Translation and Relevance

Cognition and Context

Ernst-August Gutt
First published 2000 by St. Jerome Publishing
Second edition published 2010

Published 2014 by Routledge

2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
711 Third Avenue, New York, NY 100l7, USA

Routledge is an imprint of the Taylor & Francis Group, an informa business

© Ernst-August Gutt 2000,2010

All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or
by any electronic, mechanical, or other means, now known or hereafter invented, including
photocopying and recording, or in any information storage or retrieval system, without permission
in writing from the publishers.

Notices
Knowledge and best practice in this field are constantly changing. As new research and experience
broaden our understanding, changes in research methods, professional practices, or medical
treatment may become necessary.

Practitioners and researchers must always rely on their own experience and knowledge in
evaluating and using any information, methods, compounds, or experiments described herein. In
using such information or methods they should be mindful of their own safety and the safety of
others, including parties for whom they have a professional responsibility.

To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors,
assume any liability for any injury and/or damage to persons or property as a matter of products
liability, negligence or otherwise, or from any use or operation of any methods, products,
instructions, or ideas contained in the material herein.

ISBN 13: 978-1-900650-22-9 (pbk)

Cover design by Steve Fieldhouse, Oldham, UK

Typeset by Delta Typesetters, Cairo, Egypt

British Library Cataloguing in Publication Data

A catalogue record of this book is available from the British Library

Library of Congress Cataloging in Publication Data

A catalogue record of this book is available from the Library of Congress
Contents
Preface to 1st Edition vii
Preface to 2nd edition ix
Acknowledgements xi

1. The State of the Art - Some Critical Observations 1

ANew Initiative 2
Reservations 3
The risk of (multidisciplinary) disintegration 3
The problem of determining the domain of the theory 4
The problem of evaluation and decision-making 7
Non-theoretical approaches 9
E~~~e 9
A Problem of Scientific Method (Research Programme) 17
Changes in Scientific Method 21
A shift in the domain of the theory 21
A shift from description to explanation 21
Translation as communication? 22

2. A Relevance-Theoretic Approach 24
The Inferential Nature of Communication 24
Semantic Representations 25
Context and the Principle of Relevance 26
Descriptive and Interpretive Use 35
Interpretive resemblance between propositional forms 36
Interpretive resemblance between thoughts and utterances 39
Interpretive resemblance between utterances 41

3. Covert Translation 47
The Notion of 'Covert Translation' 47
Translation - When all is Change? 54
Descriptive Use in Interlingual Communication 56

4. Translating the Meaning of the Original 69

Conveying the 'Message' of the Original 69
The Problem of Secondary Communication Situations 76
Secondary communication problems and
'dynamic equivalence' 79
Secondary communication problems and idiomatic translation 83
iv Translation and Relevance

Translating the Same 'Message' by Interpretive Use? 98

5. Translation as Interlingual Interpretive Use 105

Introduction 105
Faithfulness in Interlingual Interpretive Use 107
The Origin of Translation Principles 119
Conclusion 127

6. Translating what was Expressed 130

Style - The Importance ofthe Way Thoughts are Expressed 130
Direct Quotation, Communicative Clues,
and Direct Translation 132
Communicative Clues Arising from Semantic Representations 136
Communicative Clues Arising from Syntactic Properties 144
Communicative Clues Arising from Phonetic Properties 150
Communicative Clues Arising from Semantic Constraints
on Relevance 151
Communicative Clues Arising from Formulaic Expressions 155
Onomatopoeia and Communicative Clues 159
Communicative Clues and the Stylistic Value of Words 160
Communicative Clues Arising from Sound-Based
Poetic Properties 162
Conclusion 166

7. A Unified Account of Translation 168

Direct Translation - A Special Case of Interpretive Use 169
On the Use ofthe Original Context 173
Partial Resemblance in Linguistic Properties as Translation? 177
Translation, Relevance and Successful Communication 179
Risks of failure in translation 179
Making intentions and expectations meet 190
Conclusion 198

Postscript 202
A Decade Later 202
The Research Programme Shift (Paradigm Shift) 203
The input-output research programme 204
The competence-oriented research programme 205
Contents v

Small wonder different evaluations 206

The same in the end? 207
The Unique Mandate of Translation 208
Equivalence of message or function is not enough 208
Translation as interlingual quotation 209
Some Important Consequences ofa Quotation-Based Concept
of Translation 211
Translation is an act of communication rather
than a text genre 211
An aside on intentions 211
Translation is Communication between Translator
and Target Audience 213
Translation cannot be Covert 215
The production of 'covert translations' involves
translation but ... 216
Does it matter? 217
Manuals, tourist brochures, advertisements can be
objects oftranslation proper 221
Translations can be Interpreted 'incognito' but ... 221
Translation Proper does not Depend on the Existence of a
Concept of 'translation' in the Target Culture 222
The Perspective of Translation Practice 223
Toward a deeper understanding of 'meaning' 225
Applying the rules 226
Promises and Perils of Translation 228
Interpretation with a unique claim of authenticity 228
An authentic experience as well as an authentic message 229
Translation (alone) cannot necessarily guarantee
successful communication 229
The subtle complexity ofthe translator's task 232
A Word on Cultural Translation Studies 235
Conclusion 236
General References in Postscript 239
Reviews and Literature with Comments on Gutt 1991 240

General Bibliography 243

Index 250
This page intentionally left blank
Preface to 1st Edition
This study is the outcome of a growing personal concern over the theoreti-
cal foundations of translation. As a member of the Summer Institute of
Linguistics (SIL), I have become closely involved with matters of trans la-
tion. My linguistic work in Ethiopia under the Institute of Ethiopian Studies
ofthe Addis Ababa University from 1976-1978, 1978-1982, and 1983-1987,
and the attempt to deal with a trilingual situation in my own family (Eng-
lish, Finnish, and German) provided much practical experience in translation
and its problems.
It was in 1981 that I first tried to formulate some of my concerns about
the nature of the principles, rules, and methods advocated in translation and
especially about their validity; much of the methodology seemed to make
sense, some of it seemed questionable - but the worrying thought was that
it remained unclear what reality, if any, these reactions reflected and how
they could be dealt with objectively. Initially I tried to express my concerns
in a textlinguistic framework, but the results were not satisfactory.
During my studies for an MA degree at University College London in
1982-83 I was introduced by Deirdre Wilson to the relevance theory of com-
munication, which she was developing together with Dan Sperber (cf.
Sperber and Wilson 1986a). The cognitive basis of the theory combined
with its concern for both comprehensiveness and explicitness appealed to
me, and in the years that followed I began to apply relevance theory to a
few aspects of translation (Gutt 1985, 1986, 1987, 1988). The results were
very encouraging, but it became quite clear that the complex nature of the
issues involved required a prolonged period of concentrated research. So in
1987 I had the opportunity to return to University College London for doc-
toral studies under the guidance of Deirdre Wilson.
The results of my research surprised me; I had expected that relevance
theory would help me to formulate a general theory of translation. However,
within a year it became increasingly clear that relevance theory alone is
adequate - there seems to be no need for a distinct general translation theory.
Accordingly, the main thrust of this study, which is a revised version of my
doctoral dissertation (Gutt 1989), is to explore a range of translation
phenomena and show how they can be accounted for in the relevance-
theoretic framework.
Chapter 1 begins with a sketch of the status quo and a critical evalua-
tion. Chapter 2 introduces basic notions of relevance theory as found in
viii Translation and Relevance

Sperber and Wilson (l986a), and goes on to explore further what is gener-
ally involved when utterances are used interpretively to represent other
utterances.
Chapters 3 to 7 examine various views of translation, and at the same
time show how relevance theory can handle various translational phenom-
ena involved. Chapter 3 deals with instances of 'translation' where the
relationship to the original seems incidental rather than crucial to the com-
munication process. Chapter 4 examines the idea that translations should
preserve the meaning of the original. Chapter 5 argues that much of trans la-
tion can be viewed adequately as 'interlingual interpretive use', noting,
however, that on this view the notion of translation would cover a very
wide range of phenomena. Chapter 6 examines the possibility of a much
narrower view of translation designed to preserve also stylistic features of
the original. For this purpose, the notion of 'communicative clue' is intro-
duced. Chapter 7 rounds off the discussion by showing that both the wide
and the narrow view fall out naturally from the relevance-theoretic frame-
work; it also investigates prerequisites for successful communication by
translation.
Preface to 2nd Edition
After one reprint, this book was out of print for a few years. However, re-
quests for copies kept coming, and so I was glad when St. Jerome Publishing
approached me and asked for permission to re-publish the book.
Developments since the original writing of the book did not seem to
warrant a major revision, so the first seven chapters are essentially the same
as in the original version, except for some minor corrections and updates.
The changes in relevance theory introduced by Sperber and Wilson (1995)
do not affect the core of the argument in this book, but they have been
commented on in the notes at appropriate points.
However, a new chapter has been added in the form of a 'Postscript' to
take up and respond to points of general interest raised in reviews and peer
comments on the original publication. (For the convenience of the inter-
ested reader, the references to reviews and other peer comments of which I
have become aware have been compiled in a separate listing; see the sec-
tion 'Reviews and literature with comments on Gutt 1991', pp. 240-2.)

Ernst-August Gutt
October 1999
This page intentionally left blank
Acknowledgements
My thanks go to my colleagues at the Addis Ababa University who made it
possible for my family and me to have the privilege of working and living
in a linguistically and culturally rich environment. I especially thank the
Silt'i people of Ethiopia who allowed us to live with them and to get first
hand experience of both the enrichment and problems caused by cultural
and linguistic diversity.
I want to express my heartfelt thanks and appreciation to Deirdre Wilson,
my supervisor. It has been a great privilege and joy to work under her guid-
ance and with her encouragement. I thank her for opening my eyes to the
wonders of two of the most precious gifts we have: the human mind and the
ability to share what is in it with others. My research has brought home to
me in a new way the truth that we are 'wonderfully made' (Psalm 139:14).
I thank Neil Smith for his encouragement and for his comments on the
dissertation which underlies this book. I thank Regina Blass, Ron Olson,
Clinton Robinson, Ronald Sim, Dan Sperber, the staff and fellow-students
in the Department of Linguistics, University College London as well as other
colleagues for their feedback on my research. Of course the responsibility
for the shortcomings of this book is entirely my own.
I want to express my gratitude for the financial help I received from my
colleagues in the Eastern Africa Group (SIL) and from the Corporation
Academic Scholarship Fund (SIL) in 1987-88, and for the University Post-
graduate Studentship which I was awarded by the University of London for
1988-89.
Finally, a very special thank you goes to my wife Eeva and my children
Hannele, Hannu and Hedi for supporting a husband and father whose mind
was not always free to give them the attention they needed. It is my pleasure
to dedicate this book to them.
This page intentionally left blank
1. The State of the Art - Some Critical Observations
It is becoming commonplace for works on translation theory to acknowl-
edge that there is a vast body of literature on translation, offering a wealth
of observations and views on the sUbject. l Almost equally commonly this
acknowledgement is followed by some sort of a caveat - expressed with
varying degrees of candour - to the effect that the volume of the literature is
not necessarily indicative of the degree of understanding reached. Steiner
(1975:238) expressed this in the following words:

Despite this rich history, and despite the calibre of those who have
written about the art and theory of translation, the number of origin-
al, significant ideas in the subject remains very meagre.

Discontent seems to focus in particular on the lack of a comprehensive

approach to translation that is both systematic and theoretically sound. For
example, back in the sixties Levy observed:

Only a part of the literature on the problem of translation moves on

the theoretical plane. Until today most studies and book publications,
especially on literary translation, have not gone beyond the limits of
empirical deliberations or essayistic aphorisms. (LevY 1969: 13, trans-
lation my own)

By the end of the 1970s, the situation seemed to have changed little,
because Kelly introduces his history of translation theory and practice with
the recognition that "a comprehensive theory of translation has proved elu-
sive" (1979:1).
And so it has continued into the 1980s; in Bassnett-McGuire's view,
"the systematic study of translation is still in swaddling bands" (1980: 1);
drawing an analogy to literary studies, Wilss (1982:11) sees the literature
on translation as amounting to a "mass of uncoordinated statements" :

Slightly modifying the phrase used by Bertold Brecht to describe

literary scholarship as 'a mass of opinions', it could be said that the
many views expressed on translation in the past centuries amount to

1 For extensive reviews see e.g. Kloepfer (1967), Steiner (1975), Kelly (1979), Bassnett-

McGuire (1980), Koller (1983). For bibliographies cf. e.g. Bausch, Klegrafand Wilss
(1970/2,2 vols), and Lehmann (1982).
2 Translation and Relevance

a mass of uncoordinated statements; some very significant contribu-

tions were made, but these never coalesced into a coherent, agreed
upon, intersubj ectively valid theory of translation.

More recently still, Bell (1986) has addressed this issue in a paper with the
significant title 'Why translation theory is in a mess and what we can do
about it'.
Many different explanations have been proposed for this disappointing
situation. One is that translation theorists were preoccupied for too long
with debating unfruitful issues, such as whether translation should be literal
or free, or whether translation is possible or not. Another suggestion is that
the understanding of translation has remained inadequate because it has
never been studied in its own right, but merely as a subdomain of some
other subject, such as literature or foreign language teaching. Some scholars
have suggested the simple, if radical, explanation that translation simply is
not open to scientific investigation because it is an art or a skill. By contrast,
still others have suggested that our scientific understanding of transla-
tion is so poor because it really has not been studied in a proper scientific
manner.
This last suggestion is perhaps the most important in that it poses a posi-
tive challenge, which has already resulted in new research initiatives on
translation.

A New Initiative

Hofmann (1980) introduces his study ofthe problem of redundancy in trans-

lation as a response to this challenge:

The choice of research topic was determined by the recognition that

attempts to formulate a comprehensive theory had to fail because of
insufficient basic research into the most important invariants and
variables of literary translation .... The most urgent task is ... to de-
scribe the nature of those factors operative in the process of
translation, to identify them precisely, as far as possible, and to for-
malize them for application. (1980: 1, translation my own)

Such empirical studies pay particular attention to matters of method,

and this is indeed seen as one of the distinguishing marks of modem ap-
proaches to translation:
The State of the Art 3

What distinguishes the modem science of translation from previous

considerations of translation theory is its interest in knowledge of
methodology and its keener awareness of the problems involved.
(Wilss 1982:53)

As Wilss goes on to say, this interest in sound scientific methodology

has led to a multidisciplinary view of translation science:

Its [modem science of translation, E-AG] efforts to establish a clear

idea of its field of study ... have meant that in addition to linguistic
points of view, aspects associated with the science of communica-
tion, with psycho linguistics, sociolinguistics, textlinguistics, speech
act theory, philosophy of action (Handlungstheorie), the study of
literature, and - last but not least - with teaching, have taken on
relevance for the science of translation. (1982:53)

Schulte observes that although the interdisciplinary nature of translation

had been noted for some time, it "has received very little attention during
the last decade" (1987:2). Yet this aspect is seen as very important:

Translators do not engage in the mere transplantation of words; ...

their interpretive acts deal with the exploration of situations that are
constituted by an intense interaction oflinguistic, psychological, an-
thropological, and cultural phenomena. (1987:2)

So it seems that we have entered a new era of empirical, multidisciplinary

research on translation:

We believe strongly that the time is ripe - indeed, long overdue - for a
wholehearted commitment by linguists (broadly defined), other human
scientists, practicing translators, language teachers and translator-
trainers in a multidisciplinary approach to the description and
explanation of translation; as both process and product. (Bell 1986:7)

Reservations

The risk of (multidisciplinary) disintegration

Yet, positive as this sounds, reservations have increasingly been expressed,

not only by those who believe that translation falls outside the domain of
4 Translation and Relevance

scientific investigation, but by those advocating such investigation. Thus

Wilss concedes that "there are serious difficulties involved in designing a
paradigm for the science of translation which would withstand the test of
the theory of science and which would be capable of furnishing verifiable
results" (1982:65). Crucially, he sees the root of the problem as lying in the
multidisciplinary expansion itself:

The ensuing problems of objectification can be explained primarily

by pointing out that translation cannot be termed a purely 'linguistic
operation' ... , but rather must be thought of as a psycholinguistic,
sociolinguistic and pragmalinguistic process ... which lends itself to
an exhaustive scientific depiction only with the greatest difficulty.
(Wilss 1982:65)

Thus one problem seems to be that the multidisciplinary approach to

translation brings with it a serious threat to the very aim for which it has
been demanded: that of developing a science of translation.
De Waard and Nida (1986:185) express the problem diplomatically as
follows:

Translation is also a science in the broad sense of the term, for it

is an activity which may be systematically described and related
meaningfully to various disciplines. In the strict sense of the word,
however, translating is not a science but a technology, for it is built
upon a number of scientific disciplines, including psychology, lin-
guistics, communication theory, anthropology, and semiotics.

Thus one of the main problems with the scientific investigation of

translation seems to lie in the fact that not only linguistic factors, but many
other factors need to be taken into account. Since these factors belong to a
variety of different areas of life, there is a question whether a comprehen-
sive account of translation in the form of a coherent and homogeneous
theory can ever be achieved.

The problem of determining the domain of the theory

A second major problem concerns the question of what a general science of

translation is to be about, that is, what its domain should be. The obvious
answer is that it should be about translation - but the problem is that it is not
clear a priori what translation itself is. Krings (1986) comments that the
The State of the Art 5

notion of translation is used to refer to a variety of rather different phenom-

ena such as:

intralingual vs interlingual translation, translation of isolated words

or sentences (e.g. in foreign language teaching) vs translation of whole
authentic texts, translation vs interpretation (consecutive vs simulta-
neous interpretation), translation as process vs translation as product
of that process, translation from one language to another vs transla-
tion from a natural language into another system of signs (e.g. Morse
code), translation vs transliteration (translation into another writing
system, for example from Cyrillic to Roman script), human vs ma-
chine translation, translating from (the foreigu language) vs translating
into (the foreign language), translation vs free paraphrase or imita-
tion (1986:5, translation my own).2

There have been three major lines of approach to this issue. One has
relied on shared intuitions about the domain of the theory without any
attempt at defining it in any systematic way. Historically this has perhaps
been the line taken most often. The second approach is for the translation
theorist to delimit the domain by definition. Thus, having listed a number
of definitions of translation, Krings points out that one of their functions
is "to establish a consensus as to what translation is taken to be, or more
importantly what it is taken not to be" (1986:4, translation my own).
The third approach is a culture-oriented one: translation will be what a
culture takes it to be.
The obvious weakness of the first approach is that it does not lay a very
good foundation for an explicit science. The second approach has been criti-
cized as being potentially normative: by defining what translation is, it
implicitly sets a norm, excluding from consideration all phenomena that do
not meet the criteria of the definition. Thus van den Broeck (1980:83) states:

Much of the theorizing about translation, in our time as well as in the

past, has however largely neglected this relativistic point of view.
Most of the definitions given are prescriptive rather than descrip-
tive; they serve as norms for translation practice - or rather, for a
certain kind of practice - and fail to account for the description of
existing translations, in as far as they pay no regard to norms opera-
tive in areas and times other than those for which they were designed.

2A more comprehensive listing is given by Wilss (1982:27f). Cf. also the overview
given in Snell and Crampton (1983).
 6 Translation and Relevance

One reaction to this has been the fonnation of the 'Descriptive Transla-
tion Studies' approach to translation, which claims to have achieved, among
other things, "a considerable widening of the horizon, since any and all
phenomena relating to translation, in the broadest sense, become objects of
study" (Hennans 1985:7).3 The way this is achieved, according to the pro-
ponents of this view, is by taking the third, culture-oriented approach. This
approach starts with a corpus of target language texts suspected to be trans-
lations, and tries discover "the overall CONCEPT OF TRANSLATION
underlying the corpus" (Toury 1985:20, capitals as in original). A crucial
step in this process is that of setting up the corpus, because it will detennine
the domain of the investigation, and hence also its results. Toury himself
raises this question:

How ... are translations to be distinguished from non-translations

within the target culture, if such a distinction is to serve as a basis for
the establishment of corpora, appropriate for study within DTS [De-
scriptive Translation Studies, E-AG]? (1985:19)

He replies:

The answer is that, if one does not wish to make too many assump-
tions which may prove difficult or impossible to maintain in the face
of the empirical data, one really has no foolproof criterion for mak-
ing such a distinction a priori. The only feasible path to take seems
to be to proceed from the assumption that, for the purpose of a de-
scriptive study, a 'translation' will be taken to be any target-language
utterance which is presented or regarded as such within the target
culture, on whatever grounds ... (1985: 19f.)

What are such grounds on which a target-language utterance may be

regarded as a translation? Toury suggests a number of possibilities, "rang-
ing from its explicit presentation as one, through the identification in it of
textual-linguistic features which, in the culture in question, are habitually
associated with translations, to the prior knowledge of the existence of a
certain text in another language/culture, which is tentatively taken as a trans-
lational source for a certain target-language text" (1985:22f.).
This last criterion, that is, the existence of a text in another language

3For an introduction to 'descriptive translation studies' cf. e.g. Bassnett-McGuire (1980),

Hermans (1985) and Toury (1985).
The State of the Art 7

from which the target language text could have been translated, is empha-
sized by Toury in that it allows translation studies even in "cultures which
do not at all distinguish - on the product level, that is (since the translation
procedure should be regarded as universally acknowledged in situations
where translating is indeed peiformed) - between original compositions in
the target language and translations into it" (1985:23, italics my own).
This explanation seems rather surprising in view of the claim that 'De-
scriptive Translation Studies' is concerned with discovering "the nature of
the prevailing concept of translation" in the target culture: it would seem
that such a discovery presupposes that the target culture has a concept of
translation - yet Toury explicitly states that this is not a necessary condi-
tion: 'Descriptive Translation Studies' can be carried out even if the people
do not distinguish translations from other target-language texts.
Toury's answer is, of course, to be sought in the italicized statement.
However, this statement reveals that Toury's approach is, in fact, not culture-
determined but does make a priori assumptions about translation, or rather
about 'translating': it is assumed that people of any culture universally real-
ize that they translate when they translate. Thus Toury does after all, in the
last resort, rely on a universal concept of 'translating' as a process, if not of
'translation' as a product.
Indeed, it is difficult to see how 'Descriptive Translation Studies' could
otherwise be applied to any language other than English: only in English
are texts regarded as 'translations', and in the absence of any further
intercultural assumptions there is no a priori reason for relating an English
'translation' to a German 'Ubersetzung' or to an Amharic 'tlrgum'.4 In other
words, the study of translations as an intercultural discipline cannot be car-
ried out on purely culture-specific assumptions; it must include intercultural
assumptions as well.
Thus the culture-specific approach does not really resolve the problem
of defining the theoretical domain: either it leads to the abolition of the
intercultural study of translation or it does in fact rely on non-culture-
specific criteria for determining its domain.

The problem of evaluation and decision-making

Another major problem area that is regarded as making the scientific study
of translation difficult is that of evaluation and decision-making. Wilss

4 Amharic is an Ethio-Semitic language, and is used as the official language in Ethiopia.

 8 Translation and Relevance

comments on this as follows:

Owing to its structure, it is more difficult for the science of trans la-
tion than for the more strongly system-oriented linguistic disciplines
to acquire an epistemological foundation and arrive at a description
of translation which adequately deals with the problems involved.
This is because the translation process involves a decision-making
process in a great variety of texts that are ofpractical importance
to translation. (1982: 13, italics my own)

The need for decision-making arises from the fact that the target lan-
guage rarely allows the translator to preserve exactly what the original
conveyed. Levy illustrates this by considering how the title of Brecht's play
'Der gute Mensch von Sezuan' could be translated into English. He ob-
serves that in English "there is no single word equivalent in meaning and
stylistic value to the German 'Mensch'" (1967:1171): Englishperson would
be equivalent in meaning, but belongs to a different stylistic level. If the
translator tried to maintain the same level of style, he would be faced by the
fact that in English the range of meaning of Mensch, "denoting the class of
beings called 'homo sapiens', ... is covered by two words: 'man' and
'woman'" (1967:1171). So, since none of the options considered captures
all that the original seemed to express, the solution is not self-evident but
requires a non-trivial decision on the translator's part.
In one of his most recent books Wilss notes that despite its importance
this aspect of translation seems to have been largely neglected:

In view of the complexity of decision-making in translation one would

expect the science of translation to have launched an intensive dis-
cussion of the nature of the decision-making processes in translation,
but no such discussion of any degree worth mentioning has taken
place. In the vast technical literature on the science of translation,
the notion 'decision-making process' occurs only three times in the
title of a publication ... (Wilss 1988:93)5

5 Levy 1967 is one of these three; it sketches the possibility of combining a game-

theoretic approach with pragmatic considerations ('minimax strategy'). In addition

to its reliance on a semiotic system, this sketch seems deficient in that it does not ex-
plain how readers can recover what the translator intended to convey. The other two
titles are KuBmau1 (1986) and ReiB (1981). Wi1ss points out that neither ofthem actu-
ally addresses the problem of decision-making itself. Where decision-making is referred
or alluded to in other works on translation, one finds that the science of translation is
still in its infancy "if one wants to know, how decisions in the process of translation
come about and what causes they have" (Wi1ss 1988:94, translation my own).
The State of the Art 9

With regard to evaluation, too, he notes: "The teaching of translation

methods in most cases points out only that one has to evaluate and weigh,
but not how to do it" (Wilss 1988:97).

Non-theoretical approaches

One possible reason for this neglect may lie in the fact that while recogniz-
ing its importance in practice, some have denied that this aspect of translating
should or could be covered from a theoretical or scientific point of view.
For example, Steiner argues that the 'precisions' to be achieved in trans-
lation "are of an intense but unsystematic kind", and he concludes from this
that the study of translation as a whole is not really a science: "What we are
dealing with is not a science, but an exact art" (1975:295).
Newmark is also sceptical about an objective approach to evaluation in
translation:

Translation shares with the arts and other crafts the feature that its
standards of excellence can be detennined only through the infonned
discussion of experts or exceptionally intelligent laymen; ... After
mistakes have been 'proved' by reference to encyclopaedias and dic-
tionaries, experts have to rely on their intuition and taste in preferring
one of two or three good translations of a sentence or paragraph.
Their final choice is as subjective as the translator's choice of words,
but they must be ready to give reasons for their choice. (Newmark
1988:18)

Like Steiner, Newmark is sceptical about the scientific treatment of trans-

lation as a whole:

In fact translation theory is neither a theory nor a science, but the

body of knowledge that we have and have still to have about the
process of translating ... (Newmark 1988:19)

This is, of course, a possible position to take - but obviously not helpful
to scientific penetration of the subject.

Equivalence

However, at least with regard to translations as products, many theorists

have attempted to deal with evaluation. Traditionally, this was often done
10 Translation and Relevance

in tenns of such notions as 'faithfulness' and 'fidelity', but currently the

dominant evaluative concept in translation is that of' equivalence': the quality
of a translated text is assessed in tenns of its equivalence to the original
text. While this general maxim is widely agreed upon, the problem is that in
itself it says hardly anything:

The concept of equivalence postulates a relationship between

source-language text ... and target-language text ... The concept of
equivalence does not yet say anything about the nature of the rela-
tionship: this must be defined in addition. The mere demand that a
translation be equivalent to a certain original is void of content.
(Koller 1983: 186, italics as in original, translation my own)

In other words, the notion of 'equivalence' is meaningful only with re-

gard to a conceptual framework that spells out what aspects of the texts are
to be compared and under what conditions equivalence is thought to pertain.
This has turned out to be a major problem, because different scholars
have proposed different frames of reference: Kade (1968a) restricts his ap-
proach to the content level, Koller (1972) refers to 'textual effect', Nida
and Taber (1969) to audience response, and Koller (1983) proposes five
frames of reference (denotation, connotation, textual nonns, pragmatics, and
fonn), to name but a few. The most widely accepted frame of reference for
translation equivalence now is probably that of 'function', which amounts
to the claim that a translated text (or element of a text) is equivalent to its
source language counterpart if it fulfils the same function (Levy 1969,
House 1981, de Waard and Nida 1986). Yet even here basic questions re-
main as to the types of functions that need to be distinguished and as to
whether a text can have one or several functions at a time. Furthennore, as
we shall shortly see, (see below), a number of translation theorists have
already rejected the notion of single 'functional equivalence' as inadequate.
As Svejcer has put it: "Equivalence is one of the central issues in the theory
of translation and yet one on which linguists seem to have agreed to dis-
agree" (Svejcer 1981:321).

The problem of over-specification

One of the sources of disagreement is that texts are not only very complex
structures in themselves but are also complex with regard to the uses to
which they are put and the effects which they can have in a given situation.
This means that translation and original can be compared with regard to
The State of the Art 11

a very large number of factors, any of which can be significant for some
detail in the text, and hence needs to be taken into consideration when es-
tablishing equivalence.
Wilss (1982) gives an illustration of how very specific expectations of
the recipients can affect judgements of equivalence:

The importance of the TL [target language] recipient comes out quite

clearly in a translation example by Neubert concerning the German
translation of a passage in John Braine's novel Room at the Top
(1968:32f.):
(12) I came to Warley on a wet September morning with the sky the
grey of Guiseley sandstone. I was alone in the compartment ...
(13) Es war an einem regnerischen Septembermorgen mit einem Him-
mel wie aus grauem SandsteiniGuiseley-Sandstein, als ich in
Warley ankam. 1m Zug hatte ich das Abteil fUr mich gehabt ...

Neubert argues that the decision on the rendering of 'Guiseley sand-

stone' by 'Sandstein' or 'Guiseley-Sandstein' is determined by the
interest of the recipient. If this interest is exclusively focussed on
literary aspects of the original, the translator can confine himself to
the reproduction of' Guiseley sandstone' by 'Sandstein', without of-
fending against important TE [translation equivalence] postulates and
without shortchanging the SLT [source language text]. If, on the other
hand, the translator must reckon with additional interests of the reader
in area studies, he must react accordingly, because in a case like this
only a translation containing an explicit reference 'Guiseley sand-
stone' would meet TE expectations and thus could be regarded as
adequate. (Wilss 1982:145)

In the light of such problems, Wilss concludes:

TE [translation equivalence] cannot possibly be integrated in a gen-

eral translation theory (... ), but must be looked upon as part of specific
translation theories which are at best text-type-related or, even more
restrictedly, single-text-oriented. (1982: 135)

It is surprising that Wilss does not discuss the further implications of

such a view for the whole enterprise of constructing a theory of translation.
One of the main points of theory-construction is that it should allow us to
account for complex phenomena in terms of simpler ones: that is, one of its
main motivations is to make generalizations about phenomena. But if it
turns out that each individual phenomenon - which here is not only each
12 Translation and Relevance

text, but potentially each instance of translating it for a particular audience-

may require its own theory of equivalence, then this means that these phe-
nomena cannot be accounted for in terms of generalizations at all, and that
they actually fall outside the scope of theory. Thus recognition of the poten-
tial need for single-text based 'theories' of translation equivalence entails a
possible reductio ad absurdum of the notion of 'theory' itself.

Is 'equivalence' an evaluative concept?

Another problem is that the notion of equivalence may not be truly evalua-
tive in nature. To see this, let us take a closer look at House's (1981) treatment
of equivalence in translation, which offers one of the more detailed models
of functional equivalence.
House starts from the assumption that "in order to make qualitative
statements about a translation text (TT), TT must be compared with the
source text's (ST's) textual profile which determines a norm against which
the appropriateness ofTT is judged" (198l:5lf., italics as in original). Such
'textual profiles' are detailed analyses that allow the classification of texts
on the basis of their linguistic characteristics along eight different dimen-
sions; three ofthese are 'dimensions oflanguage users': geographical origin,
social class, time; the other five are 'dimensions oflanguage use': medium,
participation, social role relationship, social attitude, and province. 6 In terms
of this model, the evaluation seems to be straightforward enough: the pro-
file of the original text provides the norm against which the profile of the
translated text is to be compared.
However, in the course of the presentation, it becomes clear that the
matter is much less straightforward, because for an evaluation it is not suf-
ficient to note similarities and differences; one also has to place a value on
them. When the profiles of two translations differ from that of the original
along the same dimension, the problem is not so difficult:

In that two TTs [translated texts] may show mismatches on the same
parameter( s), their relative adequacy is clearly a function of the rela-
tive degree of mismatch on the particular parameter(s). (House
1981:208)

Thus their relative adequacy can be evaluated by quantification: the text

6 The notion of 'province' is explained as follows: "Province reflects occupational or

professional activity. Examples of Province would be 'the language of advertising',

'the language of public worship', 'the language of science', etc." (House 1981:40).
The State of the Art 13

whose profile shows fewer mismatches with the profile of the original is the
more adequate one.
But what if two translations differ from the original in more subtle ways?
House discusses a theoretical example where one translation has very few
"dimensional mismatches, but contains many overtly erroneous errors" and
the other "has several dimensional mismatches but no overtly erroneous
errors" (1981:208). She suggests:

It seems to us that a comparative evaluation of these two TTs can

only be arrived at from a consideration of the individual texts and
the individual translations themselves. We may, however, hypothe-
sise ... that the subgroup of overtly erroneous errors which we called
'mismatches of the denotative meaning' will be marked as a more
serious detraction from the quality of a TT than dimensional mis-
matches whenever the text has a strongly marked ideational functional
component, e.g. mismatches of the denotative meaning of items in a
science text are likely to be rated higher than a mismatch on Social
Attitude. (1981:208f.)

Then she points out again:

A detailed hierarchy of errors for any individual case can, however,

only be given for a particular comparison of two or more texts de-
pending in any particular case on the objective of the evaluation, ...
(1981:209)

This position is interesting for a number of reasons. Firstly, it seems to

show that House's model for quality assessment provides a basis only for
systematic comparison - but not for value judgements: those will follow
from an assumed "detailed hierarchy of errors" which is specific for the set
of texts to be compared and will depend "on the objective of the evalua-
tion".7 Thus for the actual evaluation no general framework is provided but
we are directed towards text-specific 'hierarchies'.
Secondly, it raises the suspicion that apart from House's particular model

7 Cf. also the main objective of Larose's (1988) 'integrated model of translation' , which

incorporates House's approach: "The goal of our work is to bring out the parameters
which form the background to the judgements made in the evaluation of the transla-
tions" (1988:XV, translation my own). About the evaluation itselflittle is said, except
that the main principle is that of functional equivalence (1988:290).
14 Translation and Relevance

the notion of 'equivalence' itself may not be truly evaluative in nature but
merely comparative, in that it allows only statements about 'sameness' and
'difference'. Such statements are, of course, useful but do not in and of
themselves constitute value judgements: they can be turned into value judge-
ments - but only on the further assumption that the more 'equivalent' a
translation is, the better it is. But this assumption is problematic in that it
shows that for evaluation equivalence is not the most basic notion of trans-
lation - it rather needs to be related to a theory of values. It is not, therefore,
surprising that equivalence-based theories have been seriously challenged;
thus Reii3 and Vermeer (1984), Honig and Kui3maul (1984), and others have
argued that the translation is not necessarily the better the more equivalent
in function it is to the original. 8 So it seems that the notion of equivalence
itself is inadequate for evaluating translations.

Hierarchical solutions
This brings us to a third noteworthy aspect of House's position as illus-
trated in the above quotation. It reveals a very general trend in current
translation theory: where problems of evaluation arise, the solution is as-
sumed to lie in some hierarchical structure that determines the priorities
between different categories and parameters.
This idea was already important in Levy's (1969) functional approach
to translation. Levy saw it as crucial for a reliable translation that in the
decision-making process "the relative importance of the values in a piece of
literature are recognised" (1969: 103, translation my own), and he suggested
the following approach for determining the importance of a given 'value'
or 'function' in the text: 9

In general, one can say that with words that have several expressive
functions, the function in the semantic complex of the higher order
is the more important one, be it the context (the sentence, the para-
graph etc.), be it the character of a person, the fable or the
philosophical objective of a work. The highest complex of expres-
sion, sometimes referred to as the idea of the work, its world view,
dominates the solution of problems in some lower unit, e.g. when
choosing the stylistic level, and this in tum determines the solution
of problems of detail. (1969: 104f., translation my own)

8 For an example, see ch. 3, pp. 54 -4.

9 Levy appears to use here the notion of 'value' ['Wert'] interchangeably with the no-
tion 'function' ['Funktion']'
The State of the Art 15

Levy himself gives only a very brief sketch of the sort of hierarchy of
functions envisaged. As we shall see in chapter 5 (cf. pp. 113 -5), the glimpse
he gives of that hierarchy raises a number of questions, and it seems fairly
clear that such a hierarchy would have to be very complex in order to be
adequate.
The following statement by Hofmann - almost twenty years later - gives
a clearer idea of the degree of complexity required, especially when one
considers that Hofmann, too, is addressing only a subset of translation prob-
lems, those concerned with redundancy of information in drama; for
Hofmann it is part of the 'highest obligation' for the translator of drama to
develop "a scheme for ranking those elements that contribute to the aes-
thetic effect perceived" (1980:23, translation my own), and this process
should be systematic:

In doing so it is important for the translator to subject his intuitive

creativity ... to a systematic method of analysis. (Hofmann 1980:23,
translation my own, italics as in original)

Hofmann spells out:

To each unit of meaning on the level of sentence and sub-scene a

grid should be applied which will determine, for a selection of the
relevant expressive means, the position of the effective elements
1 descriptive-quantitatively and
2 by qualitative evaluation with regard to the content and mean-
ing to be transferred;
in this way the grid will offer strategies for resolving the unavoid-
able conflict of the expressive means that compete with one another
and with the invariant elements/variables of content/semantics,
pragmatics (effectiveness on stage). (Hofmann 1980:23, translation
my own)

The list of aspects that need consideration in this is long, including, for
example, "rhythm, metre, verse, rhyme, nominal-verbal style, choice of
words, proverbs, puns, metaphors, illustrations, euphony and cacophony,
grammatico-rhetorical figures, syntactic means ... intonation, tempo, pauses"
and a few others (Hofinann 1980:23, translation my own).
Developing a translation theory along these lines is a truly formidable
task: it involves not only developing a framework comprehensive and de-
tailed enough to capture all these different aspects, but also relating them in
16 Translation and Relevance

terms of hierarchical structures that will rank them all according to their
relative importance in a systematic way.
However, there is another, more important concern than the feasibility
of constructing such complex hierarchies, and that is the question of what
reality lies behind them. In other words, what is it that makes translation
equivalence with regard to property X rank higher in some sense than equiva-
lence with regard to property Y?
This question is rarely addressed explicitly. Levy talks in terms of three
oppositions - 'general-individual', 'whole-part', and 'content-form' - and
suggests that the translator should emphasize the general, the whole and the
content.
He gives no further explanation as to why the first member of each pair
should normally be considered predominant, and makes clear that even these
rankings are not absolute - the second member of each pair is not to be
suppressed, "especially not in cases where it turns into its antithesis: form
must be preserved where it is the carrier of the semantic (stylistic, expres-
sive) value, uniqueness where it is a component part of a more general value,
that is, of the nationally and historically specific" (Levy 1969:108, transla-
tion my own). Thus not only does the basis of his ranking scheme remain
unclear, but it also seems to be open to the possibility of a dialectical re-
versal of rankings.lo
Hofinann appeals to pragmatics, which "embraces the purpose and goal
of the translation" (1980:27, translation my own) declaring it to be the 'in-
variant' dimension. Yet, as we shall see in chapter 5 (pp. 125-6), there are
exceptions to this ranking as well.
One of the most developed and explicit attempts to set up an evaluative
framework for translation that goes beyond statements about equivalence is
the action-theoretic approach developed by Reii3 and Vermeer (1984). Hav-
ing argued that equivalence is not the most basic concept in translation -
there is no aspect of the original that will necessarily have to be preserved
in translation - they suggest that equivalence is, in fact, only a special case
of a more general notion: that of adequacy. Adequacy in tum is always
linked to the notion of purpose ('skopos') - and it is this notion that domi-
nates translation:

10 Levy himself applies the terms 'dialectics' to these oppositions: "Closely linked to

the dialectics of the individual and the general is the dialectics of the whole and the
part" (1969: 102). On p. 108, he also speaks of the "dialectics of content and form"
(translations my own, italics as in original).
The State of the Art 17

As the highest rule of a theory of translation we propose the 'rule of

skopos': An action is detennined by its purpose (is a function of its
purpose). (1984:101, translation my own)

Thus it seems that the notion of purpose provides the evaluative dimen-
sion for translation. However, a closer look shows that this is not actually
the case, because Reii3 and Vermeer postulate that there is a set of purposes,
and furthermore that this set has a hierarchical structure: "Purposes are hi-
erarchically ordered" (1984:101). In other words, what looked like a final
answer to the problem of evaluation turns out to be only another intermedi-
ate step, raising the question of what that further dimension or principle is
that determines the hierarchical ordering of purposes.
Thus, instead of solving the problem of evaluation, Reii3 and Vermeer
only add another layer of theory to an already overwhelmingly complex
framework.

A Problem of Scientific Method (Research Programme)

In summary, it seems that the 'modem' science of translation has yet to

solve some rather fundamental problems. Furthermore, there is reason to
believe that these problems are not just particular difficulties that will be
overcome by further research. It rather seems that these problems are not
unrelated, and that they are at least partially conditioned by two basic meth-
odological factors: one is the reliance on a descriptive-classificatory approach
to science, and the other is the choice of the domain. ll In this section I want
to show how these two issues relate to the problems surveyed above.
Starting with the descriptive-classificatory approach, its basic aim is to
provide an orderly or systematic description of the phenomena in its do-
main. It starts from the observation that no phenomenon is totally different
from every other phenomenon, but that sets of phenomena seem to be simi-
lar in some respects. These similarities can be exploited for purposes of
description: they allow the theorist to describe the phenomena in sets or classes

11 The criticism made here of current approaches to translation in many ways parallels

that raised by Chomsky against the 'structuralist' or 'taxonomic' approach to linguis-

tics (cf. e.g. Chomsky 1966). However, the differences in the domains - linguistic
competence vs communicative competence - mean that different arguments are required.
I prefer the term' descriptive-classificatory' to either' structuralist' or 'taxonomic' partly
because of this difference in domain, but partly also because the latter terms seem to
have polemical overtones in certain circles.
18 Translation and Relevance

rather than individually - which lends generalising power to the theory. To

the extent that phenomena with similar properties tend to behave in similar
ways, this approach is also vested with a certain amount of predictive or
explanatory power, though this is not usually seen as its first interest.
This explains why translation theories have tended to set up classifica-
tion schemes at every level, from linguistic features via text typologies to
sets of purposes: this approach relies on classification schemes because they
are the only basic theoretical tools it has. All the generalizations are made
in terms of classes, hence in order to be covered by the theory, each phe-
nomenon in the domain must be assigned to a class. This in tum means that
the classificatory framework must be comprehensive enough to include all
phenomena in its domain, and this explains why the number of classes and
classification schemes in translation theory have kept increasing as new
sets of phenomena were found to be relevant to translating: for these phe-
nomena to fall under some generalization they must first be assigned to
classes - which requires the setting up of additional appropriate classifica-
tion schemes.
The resulting proliferation of classificatory frameworks is further ag-
gravated by the choice of domain, which is seen as consisting either of
translated texts viewed in direct relation to the original texts (the product
perspective) or of the processes and operations that lead from the original
to the translated text (process perspective).
Viewed from the product perspective, translation theory faces the prob-
lem of a virtually infinite task: since there is no upper limit to the number of
different texts a language can produce, and therefore to the number of trans-
lations that can exist in a language, corpus-based description of translation
will hardly be able to exhaust its domain. 12
From the process perspective there seem to be two alternatives, depend-
ing on whether the aim is to deal explicitly with the evaluative aspect of the
translation process or not. Krings tries to do so. For him the task of empiri-
cal, process-oriented research on translation starts with "the screening of
the psycholinguistic structure of the translation process on the basis of data
taken from concrete translation events" (1986:24, translation my own).

12 This is on the understanding that such product-oriented translation theories are essen-

tially descriptive statements about a corpus oftranslated texts (or original-translation

pairs of texts), just as structural grammars are typically conceived as "descriptive state-
ments about a given corpus" (Chomsky 1964: 11). Wilss is aware ofthis limitation, but
defends it in so far as it can lead at least to the discovery of "statistical regularities"
(1982: 16).
The State of the Art 19

However, when it comes to theory formation, he falls back on classifica-

tion, because the theory will be concerned with "how this structure of the
translation process varies depending on the characteristics of the translator
(e.g. his degree of translation competence), the characteristics of the text
(e.g. text type), and the languages involved (e.g. closely related vs typologi-
cally distant languages)" (Krings 1986:24, translation my own).
Holz-Manttari's (1984) action-theoretic model does not fall back on clas-
sification: in fact it points out some of its weaknesses (see below). However,
it does not really address the problem of evaluation either; the author keeps
emphasizing that translators must aim at 'functional adequacy' (,Funktions-
gerechtheit'), but leaves this notion undefined. In fact, this exclusion of the
evaluative aspect seems to be deliberate:

The methods, models and schemata presented are intended to make

visible these [functional] dynamics. They are not meant to over-
regulate, to choke intuition or to automatize processes .... According
to each case and need they can be used either meticulously or with
generous selectiveness. For it is the expert that decides about the
use of his instruments (1984: 127, translation and italics my own)Y

Thus while this procedural model avoids the problem of classification,

it does not explicate how the translator is to make the decisions necessary in
the translation process; this is left to expert intuition.
The multidisciplinary ambitions of translation theory can also be
seen to follow from this descriptive orientation: in order to take into ac-
count non-linguistic factors that clearly have a bearing on translation, the
descriptive-classificatory framework has to be extended to other domains
like psychology, sociology, and culture in general.
Returning to the descriptive-classificatory approach, another of its
problematic consequences is the following: since its generalizations are
made in terms of classes, the theory is geared to dealing with sets of similar
phenomena rather than with individual phenomena. However, this is not

13 In formation processing models set up for translation, and most commonly they are

proposed for oral translation (simultaneous interpretation), "try to specify in detail the
mental operations involved in a given process, and to outline the stages through which
information is coded and transformed from the input to the appropriate response"
(Flores d'Arcais 1978:381). However, the 'mental operations' surveyed do not really
address the aspect of decision-making. Cf. e.g. Massaro (1978) and Moser (1978), and
Flores d' Arcais' summary (1978).
20 Translation and Relevance

satisfactory because the translator needs to deal with individual problems,

as Holz-Manttari points out:

Typification does show up the typical but has the serious disadvan-
tage of always ignoring just that which is decisive for the case in
hand. But this is the very problem which the translator has to be able
to solve and for which he must be given higher-order rules based on
a theory. (1984: 16, translation my own)

This means that either the classificatory approach has to be satisfied

with approximate rules that deal with sets of phenomena rather than with
individuals, or its framework has to be refined to a point that allows the
unique classification of every phenomenon in the domain. This not only
further complicates the theoretical framework but leads to a loss of gen-
eralizing power: the smaller the class of phenomena to which a statement
applies, the less its generalizing power, and statements that refer to single-
member classes are no longer generalizations. Thus the problem of over-
specification which we looked at above (see pp.1O-2) can also be seen as a
natural consequence of the descriptive-classificatory approach. 14
Finally, it seems that the problem of evaluation in translation can also be
traced back, at least to a large degree, to the descriptive-classificatory ap-
proach. In the final analysis this approach is capable only of setting up and
matching categories: it can categorize the phenomena in the source text, it
can categorize those in the translated text, it can categorize functions etc.,
and make statements about matches and mismatches between phenomena
in terms of the classes to which they have been assigned. Beyond that, how-
ever, it has no other principle to offer. The postulation of hierarchical
structures is an attempt to incorporate value judgements into the classificatory
framework, but strictly speaking does not have a methodological basis in
that framework, which exhausts itself in classifYing phenomena in terms of
their intrinsic properties. Evaluation and decision-making, however, cannot
be accounted for in this way: the value, significance, importance etc. of a

14 Snell-Hornby (1988) clearly recognizes the futility of trying to capture language with

all its variations and complexities in some typological scheme. However, her proposal
of replacing typologies with prototypologies and a 'gestalt-like system' pays the price
of inexplicitness: when dealing with 'blurred edges' and 'overlappings' the proto-
typological framework gives no further help to the translator. The basic problem is that
Snell-Hornby still follows the descriptive-classificatory approach. For further discus-
sion ofthis matter see Gutt (1990).
The State of the Art 21

phenomenon do not lie in its inherent properties, but in its relation to

human beings.

Changes in Scientific Method

If this analysis is correct, then two basic changes in approach seem to be pre-
requisites for further progress in the study oftranslation: a shift in the domain
of the theory away from 'translational behaviour' (van den Broeck 1980)-
either as 'product' or 'process' - and a shift away from the descriptive-
classificatory approach. I believe that both these shifts have become possible
with the development of the relevance theory of communication by Sperber
and Wilson (l986a).

A shift in the domain of the theory

Relevance theory approaches communication from the point of view of com-

petence rather than behaviour: it tries to give an explicit account of how the
information-processing faculties of our mind enable us to communicate with
one another. Its domain is therefore mental faculties rather than texts or
processes of text production, and it is the aim of this study to explore the
possibility of accounting for translation in terms ofthe communicative com-
petence assumed to be part of our minds. IS
This does not mean that the host of different factors noted as important
in recent years are ignored: they are naturally covered in the only way in
which they can have an influence on translation anyway - and that is as part
of our mental life; no external factor has an influence on either the produc-
tion or interpretation of a translation unless it has entered the mental life of
either the translator or his audience. Its mere existence 'out there' is not
enough to influence the translation.

A shift from description to explanation

Relevance theory is not a descriptive-classificatory approach. It does not

15 This shift parallels "the shift of focus [ ... J from behavior or the products of behavior

to states of the mind/brain that enter into behaviour" initiated by Chomsky in the devel-
opment ofthe generative approach to language (Chomsky 1986:3). In other words, the
development of relevance theory itself and of this particular application to translation
can be seen as part of a wider shift in research programme (cf. Lakatos 1978).
22 Translation and Relevance

try to give an orderly description of complex phenomena by grouping them

into classes, but tries instead to understand the complexities of communica-
tion in terms of cause-effect relationships, which, applied to our mental life,
are taken to mean computational, and particularly inferential, relationships. 16
Furthermore, because it is tied in with a psychological optimization princi-
ple, relevance theory provides a natural basis for an empirical account of
evaluation and decision-making.

Translation as communication?

Before introducing relevance theory in more detail, it is worth pointing out

that the application of relevance theory entails that translation is being looked
at as part of communication.
This step may not seem unproblematic at this point in the history of
translation theory, mainly because of the strong feeling, noted above, that
translation covers issues too wide for anyone discipline.
However, this is not necessarily a valid objection, for two reasons. Firstly,
as just indicated, the motivation for a multidisciplinary approach is largely
tied in with the choice of behaviour as the domain of the theory. If we can
develop an account of translational competence that can accommodate the
influence of a wide range of factors without describing and classifying them
with respect to their various domains, then there is no a priori reason for a
multidisciplinary approach, and hence no a priori objection to a communi-
cation-based approach.
Secondly, the reason why communication-theoretic approaches have been
felt inadequate is strongly related to the particular model of communication
used. So far, virtually all attempts to treat translation within communication
theory have relied on some variety of what Sperber and Wilson (1986a) call
the 'code model' of communication, that is, a model that considers verbal
communication to be essentially a matter of the encoding, transmission, and
decoding of 'messages'. In recent years a number of translation theorists
(e.g. Reii3 and Vermeer 1984, Krings 1986, Wilss 1988) have questioned
the adequacy of this model, and in view of the basic inadequacies observed
by Sperber and Wilson (1986a), these criticisms seem well justified.
However, it is not clear why this should have been thought sufficient

16 Thus relevance theory assumes a largely Fodorian (Fodor 1975, 1981) view of cogni-

tion and mental events, which is characterized by a firm belief "in the existence of
mental states and in their causal efficacy" (Gardner 1987:82).
The State of the Art 23

reason to regard translation as beyond the domain of communication alto-

gether, a position taken, for example, by Reii3 and Vermeer (1984) and
Holz-Manttari (1984) in their action-theoretic approaches. The fact that a
particular approach to communication is inadequate does not necessarily
mean that any communicative approach is inadequate. As we shall see (cf.
especially chapter 7), there are, in fact, strong reasons to believe that trans-
lation is indeed best handled as a matter of communication.
The next chapter will be devoted to an outline of relevant parts of the
theory.
2. A Relevance-Theoretic Approach

This chapter is intended as an introduction to relevance theory as presented

in Sperber and Wilson (1986a), (1986b), (1987), and Wilson and Sperber
(1985, 1988a, 1988b).l The first three sections will introduce some basic
concepts relevant to our study. The last section will concentrate on the no-
tion of 'interpretive resemblance'; since I want to apply this notion to
translation, I shall discuss issues arising in the particular case where one
utterance or text is seen as interpretively resembling another. 2 Additional
relevance-theoretic notions will be introduced and discussed in later chap-
ters as the need arises.

The Inferential Nature of Communication

According to Sperber and Wilson (1986a), the crucial mental faculty that
enables human beings to communicate with one another is the ability to
draw inferences from people's behaviour. 3 Looked at from the communica-
tor's end, his task is to produce a stimulus - verbal or otherwise - from
which the audience can infer what he 'means', or, in the terms of relevance
theory, what his informative intention is.
Thus in the case of non-verbal communication, seeing that a colleague
is looking for a seat in a fairly full seminar room, I might wave and point to
a seat next to me; these movements would be the stimulus, and my informa-
tive intention would be that my colleague will infer from the stimulus I

1 Sperber and Wilson (1995) includes a 'Postface' which makes some revisions to the
1986 version of the theory. Since none ofthose revisions affect the essence of the argu-
ments in this book no major revision of this chapter is called for at this point. However,
references to the 1995 version will be made where this seems appropriate.
2 Following Sperber and Wilson, the term 'interpreting' and its various derivations (in-

terpretation, interpretive etc.) will be used in the wide sense of deriving the intended
meaning from a stimulus; to avoid confusion with the translation-specific notion of
'( simultaneous) interpretation' , the production of such oral representations of an origi-
nal text or utterance in another language will be referred to as 'oral translation', except
in quotations from the literature.
3 Strictly speaking, relevance theory applies not to all communication in the sense of

any kind of information transfer, but to 'ostensive communication' or, more explicitly,
to 'ostensive-inferential communication': "Ostensive-inferential communication con-
sists in making manifest to an audience one's intention to make manifest a basic layer
of information", this 'basic layer of information' being the communicator's informative
intention (Sperber and Wilson 1986a:54).
A Relevance-Theoretic Approach 25

produced that there is a free seat next to me where he could sit down.
The same is true of verbal stimuli; thus if someone asked me, 'Can you
tell me where the bus to Maidenhead leaves from?', I might reply, 'Sorry, I
am a stranger here myself', with the intention that the inquirer would infer
from my answer that I am unable to supply the information he is looking for.
Of course there is a difference between non-verbal and verbal or lin-
guistic communication, but this difference lies not in the presence or absence
of inference, but rather in the degree of explicitness which the stimulus can
achieve:

Linguistic communication is the strongest possible fonn of commu-

nication: it introduces an element of explicitness where non-verbal
communication can never be more than implicit. (Sperber and Wilson
1986a:175)

As we shall see, this extraordinary explicitness is due to a range of proper-

ties oflanguage, but foremost among these is the fact that verbal expressions
are assigned semantic representations.

Semantic Representations

Since the term 'semantic' is used in different ways by different people,

we need to clarify what we mean here when talking about 'semantic
representation' .
In the cognitive framework assumed by relevance theory, linguistic ex-
pressions are dealt with first of all by a component or module of the mind
that specializes in processing language data. 4 This component is essentially
a coding device which takes as input linguistic formulae and on the basis of
their linguistic properties assigns to them mental formulae that 'mean' or
'represent' something, that is, semantic representations. Thus, more techni-
cally, we use the term 'semantic representation' here to refer to mental
representations that are the output of the language module of the mind.
However, because the language module of the mind handles only linguis-
tic data, the semantic representations which it produces as output are not
normally complete and fully truth-conditional propositions or assumptions,

4 Thus relevance theory assumes a modular model of the mind, as proposed by Fodor
(1983). Its view of the language faculty is essentially the one assumed by Chomsky
(e.g. 1980, 1986, and others). For an overview and interesting discussion of these as-
sumptions cf. Carston (1988b.)
26 Translation and Relevance

but rather assumption schemas or "blueprints for propositions" (Blake-

more 1987:15) which need to be developed and enriched in a number of
different ways.5 Once a mental representation has been processed and com-
pleted in this way to become fully truth-conditional it is said to have a
propositional form.
Consider the following example:

(1) On a walk in the woods:

(a) Sheila: I heard something in that bush.
(b) Fiona: No, it was only the wind.

The meaning expressed linguistically in Sheila's remark is extremely vague:

the word 'something' can be used to refer to any non-human entity at all,
including the wind or the rustling it might produce in the leaves of a bush.
However, Fiona's reaction shows that she understands something much more
definite than that, and in particular that she thought Sheila intended to refer
to something other than the wind, perhaps an animal.
Fiona's answer also indicates that she took Sheila to be referring to an
event that had happened just in the last few moments or so. However, there
is nothing at all in the linguistic meaning of Sheila's utterance that ties it to
that particular time. Thus in a different context Sheila might use that same
sentence to refer to something she heard at a different time, in a different
place, involving a different bush etc.
Thus in relevance theory the fact that on the one hand linguistic expres-
sions do have some meaning, and yet that this meaning is not necessarily
identical to the meaning actually conveyed by that expression on any given
occasion, is accounted for by the claim that verbal communication involves
two distinct kinds of mental representations: semantic representations that
are the output of the language module of the mind, and thoughts with
propositional forms that are derived from semantic representations by
further processing. The way in which audiences get from semantic repre-
sentations to propositional forms crucially involves the use of context.

Context and the Principle of Relevance

The context of an utterance is "the set of premises used in interpreting [it]"

(Sperber and Wilson 1986a: 15). As such, it is a psychological concept: "A

5 Cf. also Carston (1988a). See Blakemore (1987:15ff.) for an argument as to why a

truth-conditional theory of meaning cannot be part ofthe grammar.

A Relevance-Theoretic Approach 27

context is a psychological construct, a subset of the hearer's assumptions

about the world" (Sperber and Wilson 1986a: 15). Hence in relevance theory
context does not refer to some part of the external environment of the com-
munication partners, be it the text preceding or following an utterance,
situational circumstances, cultural factors, etc.; it rather refers to part of
their "assumptions about the world" or cognitive environment, as it is called. 6
The notion of 'cognitive environment' takes into account the various exter-
nal factors but places the emphasis on the information they provide and its
mental availability for the interpretation process.
The cognitive environment of a person comprises a potentially huge
amount of very varied information. It includes information that can be per-
ceived in the physical environment, information that can be retrieved from
memory - in itself a vast store of information, including information de-
rived from preceding utterances plus any cultural or other knowledge stored
there - and furthermore information that can be inferred from these two
sources. Since any of this information could serve as the potential context
of an utterance - the crucial question for successful communication is: how
do hearers manage to select the actual, speaker-intended assumptions from
among all the assumptions they could use from their cognitive environment??
As we saw in the example just given, communication can require the use of
quite specific assumptions, and misunderstandings are likely to arise when
wrong assumptions are used.
The first consideration that needs to be taken into account here is that
not all the assumptions available from the potential context are equally ac-
cessible at any given point in time. For example, as you are reading this
text, information that you have acquired about communication and the prob-
lem of context selection will most likely be much more accessible to you

6 The technical definition of cognitive environment is as follows: "A cognitive environ-

ment of an individual is a set of facts that are manifest to him" (Sperber and Wilson
1986a:39, italics as in original). Manifestness in turn has the following definition: "A
fact is manifest to an individual at a given time if and only ifhe is capable at that time of
representing it mentally and accepting its representation as true or probably true" (Sperber
and Wilson 1986a:39).
7 More precisely, the potential context that could be used in the interpretation of a text

or utterance is that part of the cognitive environment that is mutually shared between
communicator and audience. The notion of 'mutual cognitive environment' is defined
as follows: "Any shared cognitive environment in which it is manifest which people
share it is what we call a mutual cognitive environment" (Sperber and Wilson 1986a:41).
See the discussion in Sperber and Wilson (l986a), ch. 1, section 3 for the significance
ofthis notion.
28 Translation and Relevance

than infonnation about who was present at your tenth birthday party. It might
take some hard thinking to recall that infonnation at this moment. Yet on
another occasion, perhaps when you are reminiscing with a fonner class-
mate about the' olden days', that infonnation might become easily accessible,
and it might require more effort at that point to recall what different people
have said about context selection.
This means that the different degrees of accessibility of contextual as-
sumptions make themselves felt by the amount of effort their retrieval requires
in a particular act of communication. This sensitivity to processing effort is
one of the crucial factors that make inferential communication possible: it
seems that communication, no doubt like many other human activities, is
detennined by the desire for optimization of resources, and one aspect of
optimization is to keep the effort spent to a minimum. Applied to context
selection, this means that hearers will naturally start out with those contex-
tual assumptions that are most easily accessible to them. Thus by its effect
on the accessibility of assumptions, the organization of memory provides
the basis for a very effective constraint on the selection of context, given
the general principle that people will try to spend as little processing effort
on supplying contextual infonnation as possible. Note that this constraint is
a perfectly general one - it makes no reference to the kind of contextual
assumptions involved, whether they are cultural or derived from an earlier
part of the text or whatever.
The other aspect of optimization is that of obtaining benefits. Relevance
theory assumes that - put in very general tenns - human beings have a natu-
ral interest in improving their understanding of the world around them, this
understanding consisting of the assumptions about the world which they
have stored in memory. Hence they expect that the effort spent in compre-
hension will in some way modifY the contextual assumptions they brought
to the communication act.
This requirement that the outcome of an act of communication has to
modifY some previously held assumptions in order to be found rewarding is
an important one; it captures, for example, the intuition that the newness of
infonnation alone does not guarantee its appropriateness. The registration
number of our car in Ethiopia is AA 2-10273. Most of the readers will have
been puzzled by this last utterance, perhaps suspecting a printing error. Yet
I am sure that this infonnation is new for virtually all of these readers.
Relevance theory accounts for this experience in the following way.
Though new, the utterance in question failed to modify any contextual as-
sumptions which the reader could reasonably expect to provide at that point,
A Relevance-Theoretic Approach 29

hence would fail to yield the expected benefits. This left him puzzled as to
what the utterance was meant to achieve. However, having been given this
further information, the reader is no doubt able to see now that the utterance
does indeed yield benefits: its use strengthens the contextually available
claim that context modification is important for communicative success.
Technically, context modifications are referred to as contextual effects,
and these can be of three kinds: they can consist in the derivation of contex-
tual implications, in the strengthening, or confirmation, of assumptions
already held, or in the elimination of assumptions due to a contradiction.
Let us briefly look at these three kinds of contextual effects. Contextual
implications are inferences that follow not from the propositional content
of an utterance alone, nor from the contextual assumptions alone, but only
from the inferential combination of the two sets of propositions. Consider
the following example:

(2) (a) Margaret: Could you have a quick look at my printer - it's
not working right.
(b) Mike: I have got an appointment at eleven 0' clock.

Although both the question and the answer are perfectly clear in their se-
mantic content, if this is all the information we have, we cannot know what
Mike's reply implies: will he be able to have a look at the printer or not?
What is missing is, of course, knowledge of the context, more specifically,
knowledge of the time of utterance.
Suppose first of all that the hearer has the following contextual assump-
tions available:

(3) (a) There are only five minutes until eleven o'clock.
(b) The printer problem is not an obvious one, but will require
opening it up.
(c) Opening up the printer will take more than five minutes.

Combined with these assumptions Mike's reply will imply (4):

(4) Mike is not able to have a look at the printer now.

This particular implication is a contextual implication in the sense defined:

it does not follow from the propositional content of Mike's reply (2) (b)
alone, nor does it follow from the contextual assumptions in (3) alone, but it
does follow from the inferential combination of both.
30 Translation and Relevance

The strengthening of contextual assumptions was already illustrated by

the use of the utterance "The registration number of our car in Ethiopia is
AA 2-10273". Here is another, more detailed example:

(5) (a) Dorothy: I have a hunch that Gill is looking for a new job.
(b) Francis: Yeah, she is studying job ads whenever she's got a
spare minute.

By her utterance Dorothy indicates that she is not sure of the truth of what
she said. If we now assume that Dorothy has the following contextual as-
sumption accessible:

(6) Someone reading job advertisements is probably looking for a

new job,

then Francis' answer is valuable in that it supplies information that can serve
as further evidence that Dorothy's assumption may be right. In terms of
relevance theory this is due to the fact that assumptions can be held with
varying degrees of strength, or conviction, and that the strength of an as-
sumption increases when it is implied by additional assumptions likely to
be true.
Lastly, contextual effects can consist in the erasure of previously held
assumptions:

(7) (a) Philip: We have to call another meeting. I don't think that
Christine is going to come, so we will be one person short
ofaquorum.
(b) Linda: No need for cancellation, I see Christine just coming
up the drive.

In this example, Philip has just made the assumption that the meeting will
have to be cancelled since it lacks a quorum. However, by her remark Linda
supplies information that is likely to prove Philip's assumption wrong: if
Christine is a regular member of the committee, and if she is, in fact, com-
ing to attend the meeting, her presence would give the committee a quorum
and hence make postponement of the meeting unnecessary.
The erasure of previously held assumptions can result when contradic-
tions arise: in this example the propositional content of Linda's remark
contradicts Philip's claim. In this case, relevance theory assumes that the
device in our mind that carries out this inferential processing proceeds
as follows:
A Relevance-Theoretic Approach 31

when two assumptions are found to contradict each other, ifit is pos-
sible to compare their strengths, and if one is found to be stronger
than the other, then the device automatically erases the weaker as-
sumption. (Sperber and Wilson 1986a: 114)

In our example, Philip's assumption that the meeting needs to be post-

poned is an inference involving the assumption that Christine is not coming
to the meeting. However, that assumption now stands against Linda's claim
that she sees Christine coming up the driveway. Since information available
from perception is usually assigned much greater strength than information
based on inference, it is most likely that in Philip's mind the assumption
that Christine is not coming to the meeting will be erased, and replaced by
the assumption that she is coming.
Having had a brief look at some of the effort or 'cost' and the benefits
involved in interpreting utterances, we can now introduce the notion that
shows how the two relate to each other, and that is the notion of relevance,
which Sperber and Wilson define in terms of the following conditions:

Relevance

Extent condition 1: an assumption is relevant in a context to the ex-

tent that its contextual effects in this context are large.

Extent condition 2: an assumption is relevant in a context to the ex-

tent that the effort required to process it in this context is small.
(1986a: 125, italics as in original)8

Thus we see firstly that relevance is dependent on the interplay of two

factors: contextual effects and processing effort. Secondly, since both these
factors are context-dependent, the notion of 'relevance' itself is context-
dependent, too. As we shall see later, this is an important point for under-
standing the nature of translation. Thirdly, relevance is a comparative notion
- utterances can vary according to the degree of relevance they achieve in
some context.
The central claim of relevance theory is that human communication cru-
cially creates an expectation of optimal relevance, that is, an expectation on

8 Gutt (1998) argues for the exclusion of processing effort from the concept of rel-

evance itself. Note that this revision would not significantly affect the discussion of the
nature of translation.
32 Translation and Relevance

the part of the hearer that his attempt at interpretation will yield adequate
contextual effects at minimal processing cost. 9 This fact is believed to be
part of our human psychology, and is expressed in relevance theory as the
principle of relevance:

Principle of relevance lO

Every act of ostensive communication communicates the presump-

tion of its own optimal relevance. II (Sperber and Wilson 1986a: 158)

Thus whenever someone shows that he wishes to communicate, he im-

plicitly and automatically conveys the assumption that the hearer can expect
to derive adequate contextual effects without spending unnecessary effort.
As Sperber and Wilson show, this assumption has an important consequence
for the theory of utterance interpretation: the hearer has the right to assume
that the first interpretation he arrives at that a rational speaker might have
expected (a) to yield adequate contextual effects, and (b) to put the hearer
to no unjustifiable processing effort in obtaining those effects is the inter-
pretation intended by the communicatorI2. Such an interpretation, according
to Sperber and Wilson, is consistent with the principle of relevance. And
this is the answer of relevance theory as to how hearers can infer what the
intended interpretation or meaning of an utterance is: it is the interpretation

9 Put more correctly, the expectation created in the hearer is that the communicator

believes that the intended interpretation will yield adequate effects without uunecessary
processing. The hearer may, of course, have doubts that the communicator's belief is
correct. For example, the hearer may suspect that he already knows what the communi-
cator intends to convey. For further discussion see Sperber and Wilson (1986a: 1 59ff.).
10 The postface in Sperber and Wilson (1995) has renamed it the 'Second principle of

relevance'. The 'First principle of relevance' says that "human cognition tends to be
organised so as to maximise relevance" (Sperber and Wilson 1995:262). Since only the
'Second principle of relevance' is referred to in this book, it is simply called 'the princi-
pIe of relevance'.
11 The postface in Sperber and Wilson (1995) re-defines 'the presumption of optimal

relevance' as follows:
"Presumption of optimal relevance (revised)
(a) The ostensive stimulus is relevant enough for it to be worth the addressee's
effort to process it.
(b) The ostensive stimulus is the most relevant one compatible with the communi-
cator's abilities and preferences." (Sperber and Wilson 1995:270)
12 In Sperber and Wilson (1986a) there is a certain amount of ambivalence in the formu-

lation of condition (a). In a number of places, the authors state that it is the interpretation
A Relevance-Theoretic Approach 33

that is consistent with the principle of relevance, and there is never more
than one interpretation that fulfils this condition. 13
This is why we are able to make use of implied information. In example
(2) above, the principle of relevance explains why we can understand Mike's
reply to suggest more than it expresses: the assumption that all Mike in-
tended to convey was the information that he was going to have an
appointment at eleven o'clock would not be consistent with the principle of
relevance; it would amount only to the addition of an assumption to the
stock of assumptions already held, and as we saw above, adding assump-
tions to existing ones does not count as a contextual effect, that is, is not
experienced as a 'reward' for the processing effort spent. Further process-
ing must be undertaken in the search for adequate contextual effects.
In its search for adequate contextual effects, the audience will also as-
sume that it is not being put to any gratuitous expenditure of processing
effort. And this is part of the answer to a question raised earlier: how does an
audience manage to select the right set of contextual assumptions from all it
knows, could observe or infer? In the pursuit of optimal relevance it turns
first to highly accessible information, looking for adequate contextual ef-
fects; if the use of this information does yield contextual effects adequate to
the occasion in a way the speaker could have foreseen, then it will assume
that it has used the right, that is, speaker-intended, contextual information.
The expression 'adequate to the occasion' is important because there is

{I} which is to have adequate contextual effects; for example, the definition of the 'pre-
sumption of optimal relevance' says that "The set of assumptions {I} which the
communicator intends to make manifest to the addressee is relevant enough to make it
worth the addressee's while to process the ostensive stimulus" (1986a:158). In other
places, however, they say that what the communicator conveys by claiming someone's
attention is that "the stimulus is relevant to be worth the audience's attention" (1986a: 156;
italics my own). However, as confirmed by the authors (personal communication), the
requirement should be that the process by which the interpretation is derived from the
stimulus must lead to adequate contextual effects. This seems necessary in order to
allow for the fact that implicated conclusions themselves can count as contextual ef-
fects. If the interpretation itself had to have adequate contextual effects, then the
implicated conclusions of an utterance would in their turn need to have contextual ef-
fects. This does not seem to be a necessary requirement, and it would entail that only
those contextual implications which the communicator does not intend to convey make
for relevance. Note that this is made clearer in the definition of the 'presumption of
optimal relevance' in the Postface of Sperber and Wilson (1995:270).
1 3There may, however, be no interpretation that fulfils this condition. In that event the

communication act fails, and the hearer probably has to ask for further clarification as
to what the speaker meant.
34 Translation and Relevance

no absolute level of relevance that is adequate to every occasion; thus when

engaging in a friendly chat with a stranger at the bus stop one would not
normally be expecting a high degree of relevance, hence would not be look-
ing for a wide range of contextual effects. By contrast, when listening to a
paper given by a leading scholar in one's field of study, one would expect a
much greater number of contextual effects, and would also be prepared to
invest more processing effort to recover them, perhaps even read a book
written by that scholar in order to understand better what she said.
So far we have looked at the principle of relevance mainly from the
hearer's point of view, but it is equally, if not more, important from the
communicator's end. Since it is the communicator's desire to have his in-
formative intention recognized, it is also his responsibility to express himself
in such a way that the first interpretation that will come to the hearer's mind
and that he will find optimally relevant will indeed be the intended one.
This means, in effect, that "communication is an asymmetrical process"
(Sperber and Wilson 1986a:43), where more responsibility lies with the
communicator than with the audience:

It is left to the communicator to make correct assumptions about the

codes and contextual information that the audience will have acces-
sible and be likely to use in the comprehension process. The
responsibility for avoiding misunderstanding also lies with the
speaker, so that all the hearer has to do is go ahead and use whatever
code and contextual information come most easily to hand. (Sperber
and Wilson 1986a:43)

The fulfilment of this condition is crucial, since the audience has no

other means by which to determine what the communicator wanted to
communicate.
Suppose, for example, I have met a former classmate by the name of
John Smith whom I had not seen for many years. Let us assume further that
I want to share this information with a friend of mine who had been a class-
mate of John's together with me. I could simply say to my friend:

(8) I met John Smith today.

Now it is likely my friend would find it difficult to work out which John
Smith I was referring to, given that he had not thought about this individual
for many years. In fact, given that the name John Smith is very common, my
friend would be likely to think of some John Smith other than our common
A Relevance-Theoretic Approach 35

classmate from primary school years, perhaps a business friend by the same
name, and so misunderstanding would be likely to arise.
In this situation I could increase the relevance of my utterance consider-
ably by saying something like (9):

(9) Do you remember John Smith, the fellow we used to tease way
back in our school days? I met him today.

Those introductory words would guide the hearer in searching his memory
for the intended referent and hence considerably ease his processing load.
To be consistent with the principle of relevance, an utterance must achieve
adequate contextual effects and put the hearer to no unjustifiable effort in
achieving them. In the circumstances described, (9) would be consistent
with the principle of relevance, whereas (8) would not. 14

Descriptive and Interpretive Use

It is part of our common experience that people do not always say what they
mean; for example, we say we have a thousand things to do when, in actual
fact, we can list perhaps some twenty or thirty jobs; or we talk about some-
one being 'a real gangster' when we do not mean to imply that he has
committed actual crimes but are perhaps expressing our attitude to the way
he goes about his business; normally this does not cause any problem but is
understood in the intended way.
While this behaviour may seem peculiar at first - and, for example, the
Gricean account (Grice 1975) of such uses assumes that it involves the vio-
lation of certain norms - Sperber and Wilson (1986a, chapter 4, section 8)
point out that it makes good sense within the 'cost-sensitive' framework of
relevance theory because such non-literal language allows for very economi-
cal communication. Suppose I want to communicate that I think that Bill in

14 This is on the assumption that the saving in processing effort achieved by that addi-

tional sentence outweighs the effort required to process that sentence; if not, utterance
(8) would have been the more appropriate one. This explains two important points about
communication: firstly it explains why unnecessary reminders are felt to hinder suc-
cessful communication: they involve additional processing effort without adequate gains
in contextual effects; secondly, it explains also why there are occasions when it is
appropriate to express information that is actually already known to the audience: it is
appropriate under those circumstances where the processing of the 'reminder' can be
assumed to be 'cheaper' in terms of processing effort than an unaided search of memory.
36 Translation and Relevance

his business behaviour tends to be rather threatening and to bully people,

that he does so with the help of other, equally sinister people, that he tends
to be ruthless in his practices - then rather than expressing all these thoughts
separately it is much more economical and efficient to express them with
the utterance 'Bill is a real gangster' - even though I do not mean to imply
that he is an actual criminal.
The way relevance theory accounts for such 'loose talk' (cf. Sperber and
Wilson 1986b) is by claiming that the relationship between what we say and
the thoughts we intend to communicate is one of interpretive resemblance -
in other words, we do not necessarily say what we think, but more often than
not what we say interpretively resembles what we intend to communicate.
In order to see more clearly what this involves, it might be helpful to
take a closer look at how our minds are assumed to entertain thoughts.

Interpretive resemblance between propositional forms

In the framework of cognitive psychology assumed by relevance theory, a

thought is a mental representation that has a propositional form. This
propositional form can be used to relate the thought in question to some
state of affairs in some (possible) world - that is, to that state of affairs of
which the propositional form is or would be true. This idea is, of course,
very familiar from truth-conditional semantics. Within relevance theory, this
way of entertaining a mental representation is called the descriptive use of
that representation: it is entertained as a description of the state of affairs of
which it is thought to be true.
There is, however, another way our minds can entertain a thought, and
this also follows from the fact that thoughts have propositional forms. An
essential property of propositional forms is that they have logical properties:
it is in virtue of these logical properties that they can contradict each other,
imply each other and enter into other logical relationships with each other.
Since all propositional forms have logical properties, two propositional forms
may have some logical properties in common. Accordingly, we can say that
mental representations whose propositional forms share logical properties
resemble each other in virtue of these shared logical properties. IS Such re-

15 It should be noted that relevance theory assumes a much wider notion of 'logic' than

that of 'standard logics': "Standard logics make a radical distinction between concepts
such as and, if ... then, and or, which are regarded as proper logical concepts, and con-
cepts such as when, know, run, bachelor, which are considered non-logical. Following
another tradition, we regard these other concepts as also determining logical implica-
A Relevance-Theoretic Approach 37

semblance between propositional forms is called interpretive resemblance. 16

What logical properties can give rise to interpretive resemblance? Wilson
and Sperber define interpretive resemblance as follows:

two propositional forms P and Q (and by extension, two thoughts

or utterances with P and Q as their propositional fonns) interpre-
tively resemble one another in a context C to the extent that they
share their analytic and contextual implications in the context C.
(1988a:138, italics as in original)

Turning first to matters of terminology, we need to clarify what analytic

implications are. Analytic implications are implications obtained by a pro-
cess of deduction in which only analytic rules have applied, and where an
analytic rule is formally distinct in that it "takes only a single assumption as
input" (Sperber and Wilson 1986a:104). For example, the concept 'brother'
can be assumed to have associated with it the following analytic rule (or
'meaning postulate'):

(10) (a) Input: (X - brother - Y)

(b) Output: (X - male sibling - Y)

This inferential rule expresses the fact that the occurrence of the concept
'brother' in a propositional form (symbolised as the string of concepts 'X-
brother - Y') warrants that the replacement of the concept 'brother' in this
string will lead to another true propositional form: 'X - male sibling - Y'.
Crucially, this rule requires only a single propositional form as input, hence
the analytic implications are all implications that follow from the pro-
positional form alone.
Analytic implications contrast with contextual implications, which were
already introduced above (cf. p. 29); as mentioned there, contextual impli-
cations do not follow from the propositional form alone nor from the context
alone but from the inferential combination of the two. 17
For a clearer understanding of what is involved in interpretive resem-
blance between propositional forms, let us consider some examples:

tions" (Sperber and Wilson 1986a:87). For further details see ch. 2, section 2 oftheir
book.
16 This is the term used in Wilson and Sperber (l988a). Sperber and Wilson (l986a)

talk simply about 'resemblance.'

17 Contextual implications are a subset of synthetic implications; synthetic implications

involve the application of synthetic rules, where "a synthetic rule takes two separate
assumptions as input" (Sperber and Wilson 1986a:104).
38 Translation and Relevance

(11) (a) Bill likes his new toys.

(b) William likes his new toys.

Let us assume that these two sentences represent two thoughts, and that the
two thoughts are about the same situation, that is, that they have the same
propositional form. In this case these two thoughts share all their logical
properties. They share all their analytic implications - whatever is analyti-
cally implied by one thought is also implied by the other; and from this it
follows that they also share all their contextual implications in any context -
whatever implications one thought may have in a given context, the other
thought will have in that context too.
As one would expect, resemblance does not have to be complete:

(12) (a) Jack bought a new Mercedes.

(b) Don bought a car.

Again assuming that these sentences stand for particular thoughts, they
would share, for example, the entailment (13):

(13) Someone bought a car.

They would, however, differ in other properties. For example, while (14) is
an entailment of (12) (a), it is not an entailment of (12) (b):

(14) Jack bought something.

Thus resemblance between mental representations is a matter of degree.

As may have been noticed, in both the examples considered the thoughts
in question shared at least some analytic implications, that is, they over-
lapped at least in part of their semantically determined meaning. However,
this is not a necessary condition of interpretive resemblance. In fact, two
thoughts can resemble each other interpretively in a certain context if, for
example, they share some contextual implication in that context. Thus (15)
can be said to interpretively resemble (12) (a) in the context of (16):

(15) For my colleague only the best is good enough.

(16) (a) Someone who buys a Mercedes has money to spare.
(b) Someone for whom only the best is good enough has money
to spare.
(c) Jack is my colleague.
A Relevance-Theoretic Approach 39

It is clear that while (12) (a) and (15) do not share analytic implications,
given the set of contextual assumptions in (16), both (12) (a) and (15) will
yield the same contextual implication (17):

(17) Jack has money to spare.

As we shall see, this point will be important in our discussion later on.
Now given this possible resemblance relationship between mental rep-
resentations, our mind can entertain a mental representation or thought not
in virtue of its being true of some state of affairs, but in virtue of its inter-
pretive resemblance to some other representation. This use of representations
is called interpretive use in relevance theory. For example, suppose that I
have read in the newspaper that young people in the age range from 18-25
years are higher-risk drivers. I could think about this claim in two different
ways: I could entertain it as a true thought, that is, as a description of a cer-
tain state of affairs; alternatively, I could think about this as a claim that
someone else has made: that is, I could entertain this thought in virtue of its
resemblance with someone else's thought, without commitment to its truth
as a descriptive statement. In this case, in the terminology of relevance theory,
I would be entertaining this thought as an 'interpretation' of someone else's
thought.

Interpretive resemblance between thoughts and utterances

This possibility of using a representation either descriptively or interpretively

is not limited to mental representations, but is open to any representation
that has a propositional form. Thus to the extent that utterances have
propositional forms, they, too, can be used either descriptively or interpre-
tively. In fact, this is one basic 'ambiguity' that needs to be resolved by the
audience in any instance of verbal communication: whether the speaker is
using the utterance descriptively or interpretively. While speakers have the
options of indicating by the form of their utterance how they mean what
they say, they are by no means always obliged to do so.
Consider the following exchange:

(18) (a) Sarah: I really have a rather poor appetite these days.
(b) Joe: It's the Chernobyl accident.
(c) Sarah: Do you really think so?
(d) Joe: Actually, no; but Chemobyl gets blamed for anything
these days, doesn't it?
40 Translation and Relevance

In this example, (18) (b) is obviously meant interpretively, reflecting what

people in general believe, rather than Joe's own conviction, but there is
nothing in the utterance to mark this. IS
Returning to our earlier sample utterance, 'Bill is a real gangster' (p. 35-6
above), this is an instance of interpretive use because the propositional form
of my utterance 'Bill is a real gangster' interpretively resembles the thoughts
I want to convey. In this instance the assumptions which I want to commu-
nicate - 'Bill's behaviour is threatening', 'Bill bullies people', 'Bill uses
other sinister people's help', 'Bill is ruthless' - are all contextual implica-
tions of my utterance. Thus the intended interpretation shares with my
utterance a number of assumptions contextually implied by the propositional
of that utterance in the context envisaged.
At the same time, there may well be other implications, both analytic
and contextual, that I do not want to communicate by my utterance - for
example, that Bill is part of a criminal association, that he has committed a
crime, that he ought be taken to court, that he has arms etc. - though in
other contexts these assumptions might well be communicated by the same
token sentence 'Bill is a real gangster'.
Relevance theory calls those analytic implications which the communi-
cator intends to convey explicatures; the contextual assumptions which he
intends to convey are called implicatures .19 Thus the intended interpreta-
tion of an utterance consists of its explicatures and implicatures.
This means that it can be important for the successful use of 'loose talk'
that the audience be able to find out which implications the communicator
intended to communicate. How can the audience find this out? In the same
way that it always does: on the basis of consistency with the principle of
relevance. In other words, the audience will assume that the first interpreta-
tion of the utterance consistent with the principle of relevance is the one
intended by the communicator. For example, it will assume that the assump-
tion 'Bill's behaviour is threatening' is part of the intended interpretation if
it seems that the communicator could have expected the resulting interpre-
tation to have adequate contextual effects without putting the audience to

18 This is not due to the written form of the example; Joe could, of course, give a clue to

Sarah that he was not expressing his own belief, by using some kind of mocking intona-
tion or a special facial expression, but he need not do either - and in the example in
hand, he probably did not, as Sarah's question seems to indicate.
19 As we shall see below, the distinction between communicated and non-communi-

cated assumptions is not necessarily sharp, but can be graded. It should also be noted
that implicatures can be contextual assumptions as well as contextual implications.
A Relevance-Theoretic Approach 41

unnecessary processing effort.

From the communicator's point of view this means that for communica-
tive success his utterance must be such that he can reasonably expect that
the first set of analytic and/or contextual implications it conveys in consis-
tency with the principle of relevance will share with his thoughts all those
assumptions which he intended to communicate. When the communicator
succeeds in this, his utterance can be said to be a faithful representation of
his thoughts.
Note that this also allows for the case where the communicator wants to
communicate exactly what he says - as is done in ordinary assertions; in
this case the proposition he expresses interpretively resembles the intended
interpretation in all analytic and contextual implications. When one propo-
sition resembles another completely in this way, then it is called a literal
interpretation of that other proposition. Hence an ordinary assertion is sim-
ply the limiting case of interpretive resemblance - when the utterance
produced conveys a literal interpretation of the speaker's thought, resem-
bling that thought in all analytic and contextual implications.
This means, in fact, that "every utterance comes with a guarantee of faith-
fulness" - that is, in each case "the speaker guarantees that her utterance is
a faithful enough interpretation of the thought she wants to communicate"
(Wilson and Sperber 1988a:139).
Thus we see that relevance theory already comes with a notion of faith-
fulness as a natural and central part of its theoretical framework. As we
shall see in chapter 5, it does not need to be defined for particular text or
utterance types, contexts etc. but is determined for each occasion by the
principle of relevance and the cognitive environment mutually shared by
the communicator and his audience. One of our tasks there will be to exam-
ine what this general notion entails for translational faithfulness, whether it
can stand as it is or needs to be adapted in some way. However, first we
need to take a closer look at interpretive resemblance between utterances.

Interpretive resemblance between utterances

As we saw earlier in this chapter (p. 36), interpretive resemblance is defined

primarily as a relationship between propositional forms; its application to
representations such as thoughts or utterances is only by extension - in vir-
tue of the fact that such representations have propositional forms. However,
natural language expressions seem to have a number of properties that re-
quire special consideration here, and so we shall briefly tum to these
42 Translation and Relevance

properties now.
One peculiarity of natural language expressions is that they can and of-
ten do convey information non-representationally, that is, independently of
the conceptual content usually mediated by a semantic representation. For
example, English greetings like hello or goodbye do not represent anything
truth-conditionally - there is no state of affairs of which hello or goodbye
could be said to be 'true'. Consequently, these greetings do not have propo-
sitional forms.
The way that such an utterance communicates information is through an
appropriate description of it. 20 Thus, if Alfred has greeted Bill saying, 'Hello',
Bill could construct the description 'Alfred said "hello" to me'. This de-
scription would then make manifest further assumptions. For example, an
English speaker's knowledge about his language includes the knowledge
that the word hello is used to greet people informally.21 Using this knowl-
edge, Bill could derive from the description 'Alfred said "Hello" to me' the
contextual implication 'Alfred greeted me informally'. What descriptions
the audience constructs, and which set of assumptions it assumes to be com-
municated by the greeting would, as always, be determined by consistency
with the principle of relevance.
Now it seems that such utterances that lack propositional forms can
nevertheless resemble each other in their interpretations. For example,
whether I say 'Goodbye' to someone or 'Bye bye', our intuition would be
that I am communicating essentially the same idea, that is, that I am extend-
ing a leave-taking greeting.
However, this similarity in the intended interpretation cannot be cap-
tured by the notion of interpretive resemblance between propositional forms
for the very reason that neither the greeting Goodbye nor Bye bye has a
semantic representation that can be developed into a propositional form.
Furthermore, it seems that utterances, that is, verbal ostensive stimuli,
can also resemble non-verbal ostensive stimuli. For example, at a birth-
day party a child may ask where he should sit at table. The host may reply
verbally, 'At the head of the table' or non-verbally by pointing with an ap-
propriate gesture to the chair at the head of the table.

20 This account follows a suggestion made to me by Deirdre Wilson (personal commu-

nication). It will be discussed in more detail in ch. 6, pp. 155-9 below.
21 This knowledge about language is psychologically distinct from the individual's knowl-

edge a/his language which characterizes his linguistic competence, as pointed out e.g.
by Smith and Wilson (1979), ch. 2, esp. pp. 36ff.
A Relevance-Theoretic Approach 43

Again, one would feel intuitively that the interpretations intended to be

conveyed by each of these ostensive stimuli resemble one another, but the
notion of interpretive resemblance between propositional forms could not
be invoked to account for this, since the non-verbal gesture lacks a propo-
sitional form.
Another important characteristic of natural language expressions is that
they can influence their interpretations by guiding the search for relevance,
that is, by imposing constraints on the relevance of the utterance in which
they occur (cf. Blakemore 1987, Blass 1988 and 1990, Gutt 1988a). This
can be done, for example, by the use of words or morphemes that indicate
how the proposition expressed is to achieve relevance, say, as the conclu-
sion or a premise of an argument.
Consider examples (19) and (20):

(19) So Charles has lost his car keys.

(20) After all, Charles has lost his car keys.

Assuming that these are two utterances, and assuming further that their
propositional forms are the same - referring to the same individual called
'Charles', the same car, the same set of car keys etc. - one would still feel
that these two utterances differ significantly in their overall interpretation.
The difference is, of course, due to the presence of the connectives so
and after all. As Blakemore (1987) has argued, these connectives constrain
the way that the utterance is relevant: thus the so in utterance (19) indicates
that this utterance is relevant as the conclusion to a contextually assumed
argument, but the after all in (20) indicates that this utterance is relevant as
a premise in a contextually assumed argument. In other words, two utter-
ances with identical propositional forms may differ in their interpretations
precisely because the form of the utterance imposes different constraints on
how the propositional form is to be related to the context, and hence on
what contextual effects it is to have.
Similarly, utterances (21) and (22) may well differ in their interpre-
tations, even if we assume that they refer to the same event, that is, even if
they have the same propositional form:

(21) Wife to husband: Fred has broken a window.

(22) Wife to husband: Your son has broken a window.

The interpretations of these two utterances would differ significantly in

44 Translation and Relevance

the following ways: by choosing the composite expression 'your son' rather
than simply 'Fred', the communicator makes the utterance more costly to
process. Furthermore, if it is mutually clear to the communication partners
that she could have simply said 'Fred', it would be equally clear that this
choice was intentional, and this would entail that she intended to convey
additional implicatures to compensate for the increase in processing effort.
Thus by using the composite expression 'your son' in (22), the communica-
tor indicates that she thinks it significant that the culprit is the addressee's
son and that she intends the hearer to derive implicatures along these lines.
The wording of (21) does not give rise to such additional implicatures.
As we shall see in chapter 6, there are many subtle ways in which com-
municators can exploit linguistic means to modify the interpretation of an
utterance, without changing its propositional form.
These observations, however, seem to be at variance with the notion
of interpretive resemblance between propositional forms which predicts
that sameness in propositional form entails the sameness of all contextual
implications:

If two thoughts or utterances have the same propositional form, and

hence share all their analytic implications, they also, of course, share
all their contextual implications in every context. (Wilson and Sperber
1988a:138)

While this claim holds true for propositional forms, it is not necessarily true
of utterances.
The underlying reason why the notion of interpretive resemblance be-
tween propositional forms is not directly applicable to such resemblances
between utterances seems to be that utterances are intrinsically bound up
with communication: they exist to convey interpretations; interpretations,
however, are dependent on considerations of relevance, and relevance is
context-dependent. Hence the interpretations of utterances are context-
dependent. This also means that resemblance between utterances is
context-dependent; utterances that resemble one another may not resemble
each other in a different context.
Consider the following example:

(23) Original utterance

Mary: The back door is open.
(24) Original context
(a) If the back door is open, thieves can get in.
(b) We don't want thieves to get in.
A Relevance-Theoretic Approach 45

(25) Intended implicature

We need to close the back door.

Suppose I did not understand what Mary said, and so I ask you what she
said. Two possible answers would be the following:

(26) Report 1
The back door is open.
(27) Report 2
We should close the back door.

Now if! have access to the original context (24), either report will con-
vey to me what Mary meant, hence both will resemble the original closely
in its originally intended interpretation.
But suppose that my context differs from the original context:

(28) New Context

(a) If the children are coming home from school, the back door
must be open.
(b) The children are coming home from school.

If I interpret report 1 in this new context, the analytic implications will be

the same as that of the original, but the contextual implication will be very
different:

(29) The back door should be open.

By contrast, if! process report 2 in the new context, it will yield an interpre-
tation quite similar to that of the original: it will convey both that the back
door is open and also that it should be closed.
However, the definition of interpretive resemblance between proposi-
tional forms does not allow us to capture this difference because it defines
interpretive resemblance only with regard to the same context:

two propositional forms P and Q ... interpretively resemble one an-

other in a context C to the extent that they share their analytic and
contextual implications in the context C. (Wilson and Sperber
1988a:138; p. 37 above)

But if report and original are interpreted in the same context, whether
46 Translation and Relevance

that context is the original one or the new, then both report 1 and report 2
will be found to resemble the original closely. Cases where report and original
are processed in different contexts and can therefore lead to different inter-
pretations are not addressed by the definition. Thus the definition does not
reveal that if the reports are processed in the new context, then report 2
actually resembles the original (as interpreted in the original context) more
closely than report 1 does.
To take care of such situations, which are common when utterances are
used to represent other utterances, it seems that what one has to compare
are the assumptions communicated by each utterance in its own context
rather than in the same context.
Thus we see that the notion of interpretive resemblance between propo-
sitional forms cannot be applied to resemblance between utterances without
some modification, and I want to propose the following solution.
Starting from the definition of interpretive resemblance between propo-
sitional forms, the crucial point is the sharing of analytic and/or contextual
implications. Since these implications are assumptions, we can say more
generally that interpretive resemblance is characterized by the sharing of
assumptions.
Considering further that the main purpose of utterances is to convey the
set of assumptions which the communicator intends to convey, it seems
reasonable to define interpretive resemblance between utterances in terms
of assumptions shared between the intended interpretations of these utter-
ances. Since the set of assumptions an utterance is intended to convey consists
of explicatures and/or implicatures, we can say that two utterances, or even
more generally, two ostensive stimuli, interpretively resemble one another
to the extent that they share their explicatures and/or implicatures.
This notion of interpretive resemblance is independent of whether or
not the utterances in question have a propositional form, but at the same
time it is context-dependent, since the explicatures and implicatures of ut-
terances are context-dependent.
3. Covert Translation

It is claimed at times in the literature that a good translation should read not
like a translation at all, but like a target language original. Usually this merely
expresses the requirement that in terms of style, or naturalness of expres-
sion, a translation should be indistinguishable from a receptor language
original. However, at times this claim reflects the idea that there are instances
of translation where the translated text is intended to function like a target
language original. In this chapter I want to examine this particular view.

The Notion of 'Covert Translation'

House's (1981) model for translation quality assessment explicitly provides

for such cases under the notion of 'covert translation'. She defines covert
translation as "a translation which enjoys or enjoyed the status of an origi-
nal ST [source text] in the target culture" (1981:194). She calls this type of
translation 'covert' because "it is not marked pragmatically as a TT [trans-
lated text] of an ST but may, conceivably, have been created in its own
right" (1981: 194, emphasis as in original).l Furthermore, covert translations
"have direct target language addressees, for whom this TT is as immedi-
atelyand 'originally' relevant as ST is for the source language addressees"
(1981:195).
House adopts a functional equivalence approach to translation, that is,
the translation should match the original text in function, where 'function'
is to be understood as "the application (cf. Lyons, 1968:434) or use which
the text has in the particular context of a situation" (1981 :37). Now in that
framework the notion of 'covert translation' occupies a special place be-
cause covert translations are in fact the only ones capable of actually
achieving 'functional equivalence', the main goal of the theory:

it is only in cases of covert translations that it is in fact possible to

achieve functional equivalence. (1981:204)

By way of contrast, the other major type of translation, called 'overt

translation', necessarily falls short of this goal:

1 Here and in the subsequent discussion I am concerned with House's particular theory

of functional equivalence; therefore the remarks and comments made about it do not
necessarily apply to other approaches using the term functional equivalence.
48 Translation and Relevance

In the case of overt translations, the achievement of strict functional

equivalence is, in fact, impossible; a second level function must then
be aimed at in translation. (1981:205)

Thus the notion of 'covert translation' embodies the ideal case of this trans-
lation theory: the achievement of functional equivalence.
However, even though covert translation is the only type that can actually
achieve functional equivalence, this does not necessarily mean that it can
do so easily, because of differences in the sociocultural backgrounds of the
source and target language audiences:

This functional equivalence is, however, difficult to achieve because

differences of the sociocultural norms have to be taken into account.
(1981:204f.)

Thus the translator has to be careful "to take different cultural presupposi-
tions in the two language communities into account" (1981:196).
House points out some of these difficulties when evaluating an English
translation of a German tourist booklet. It is treated as an instance of covert
translation since a tourist brochure is immediately relevant to the target
audience.
The booklet is entitled Niirnberg, and provides information about the
attractions of Nuremberg for tourists. For our discussion two passages are
relevant. The first involves a reference to the age of the mastersingers: while
the German speaks about "die Zeit des Meistergesanges, die Zeit des
Schuhmachers und Meistersingers Hans Sachs" [ the time of mastersinging,
the time of the shoe-maker and mastersinger Hans Sachs] (1981 :297, gloss
my own), the English has "the age of the mastersingers and their best-known
representative, the shoe-maker Hans Sachs" (1981:301, italics my own). In
the second passage, the German talks about "das Mannleinlaufen und der
Englische Gruss" (1981:298), while the English rendering has "the moving
figures on the ancient mechanical 'Mannleinlaufen' clock and the artistic
skill ofthe carved Annunciation, the famous 'Engelsgruss'" (1981 :302, italics
my own). As can be seen, the italicized words do not have overt counter-
parts in the German original.
When comparing translation and original, House finds that in the pas-
sages quoted there is a difference along the functional parameter of 'social
role relationship': in House's perception, by leaving the italicized information
implicit, the German brochure had a special effect on the audience: it treated
Covert Translation 49

them as people cultured enough to know who Hans Sachs was and what
'Mannleinlaufen' and the 'Englische Gruss' referred to, and hence flattered
them. By making this information explicit, the English translation is felt to
lose this effect:

TT [translation text] fails to flatter the addressees ... because the as-
sumption of the addressee's knowledge offacts about Numb erg has
not been upheld in TT. (1981:123)

Yet in her evaluation, she does not treat these differences as errors, but as
adaptations required by the different sociocultural background of the target
audience:

These two mismatches are examples of the case where a reference in

ST to the particular cultural heritage of the source language commu-
nity needs to be explained to the TT addressees, for whom this culture
is alien. Therefore, these mismatches cannot be classed among cov-
ertly erroneous errors but must be regarded as changes necessitated
by the differences in cultural background between the two language
communities ... (1981:123f.)

One may disagree with House's view that the absence of the explana-
tions in the German version flatters the average German reader of that
brochure and that this effect was deliberate - one wonders if anybody ana-
lysing the German text without comparing it to any other version would
have felt any flattery at these points at all. However, this is not our concern
here, and so, for the sake of argument, we shall assume that the author's
perception of flattery in these two places is justified. Let us also agree with
her judgement that the addition of information at these points in the English
was necessary.2
However, granting these points - could one not have expected a transla-
tor aiming at functional equivalence to have preserved the flattering effect
in some other way - perhaps by building flattery into some other part of the
translated text where the sociocultural differences would not interfere?
Thus should he not have applied the technique of compensation and
added a sentence or two - not for their informational value, but to achieve

2One could, in fact, argue that the explication of this information was unnecessary:
English readers could have felt even more flattered than Germans by being treated as
knowledgeable of details of German history and culture.
50 Translation and Relevance

functional equivalence in this flattering effect?3 After all, the purpose of

the booklet is that of "attracting tourists to Numberg" with a "good meas-
ure of exaggeration and pretension" (1981: 118), and if part of the strategy
of the original involved flattering the tourist, then surely this must be very
important for the translation, too.
In fact, it is one of House's more serious criticisms of the English ver-
sion that it fails generally in these areas of "exaggeration and pretension":
failures of the English version to "use intensifiers, figurative language and
cohesion devices ... weaken the interpersonal functional component", that
is "the attempt to induce addressees to come and see Numberg by describ-
ing its characteristics in an impressive, pleasing and attractive manner"
(1981:128).
But this raises two further questions. Firstly, how can the translator know
whether or not his translation of the original is functionally equivalent, for
example in its flattering effects? Does he do so by checking whether his
translation flatters the receptor language audience in corresponding parts of
the texts, or by making sure that the number of instances of flattery that
occur is equal between original and translation, or by some comparison of
the cumulative flattering effect of the whole text? As it stands, House's
model provides no answers to these questions.
Secondly, what if "exaggeration and pretension" are not socially accept-
able in the target culture? It seems possible that the author of the English
version might want to tone down the exaggeration and flattery of the original
deliberately - perhaps on the assumption that they might be counter-
productive in view of the sociocultural background of the target audience.
House does recognize the potential importance of "differences in values
and habits, in understating or emphasizing certain emotions etc.", even be-
tween closely related cultures (1981 :198). But she cautions that" given the
goal of achieving functional equivalence in a covert translation, assump-
tions of cultural difference should be carefully examined before any change
in ST [source text] is undertaken" (1981:198). In other words: the translator
is to follow the original unless there is a reason to depart from it, and com-
parison with the original is the ultimate measure ofthe quality ofthe receptor
language text.
While this is a reasonable guideline, it leaves unanswered a rather basic

3Levy, for example, advocates compensation: "Compensation is possible: if it is neces-

sary to weaken stylistic means in some places, one has to add the lost colour elsewhere"
(1969: 105, translation my own).
Covert Translation 51

problem - and that is, that the preservation of a function may not, in fact,
make the translation functionally equivalent: for example, maintaining the
function of flattery can make the translation non-equivalent with regard to
other functions. Thus here we have an example that shows why it seems doubt-
ful that 'functional equivalence' is the most basic concept oftranslation.4
However, even if functional equivalence did offer the right theoretical
approach, spelling out what this means for covert translation - as House
(1981) points out, the ideal type for functional equivalence - seems a for-
midable task. Consider another tourist-brochure example that illustrates these
points still more clearly. This time the texts are taken from a brochure pro-
vided for passengers on board the Finnjet car ferry operating between
Travemlinde and Helsinki. Side-by-side on one page it has two write-ups
that give information about the ferry, with all its technical advantages, and
the route. 5

1 Der schOnste Weg Finnj et - suomalaisten

nach Finnland suora tie Loma-Europpaan

[The most beautiful way [Finniet - the straight way

to Finland] for Finns to holiday -
Europe]
2 Wir laden Sie ein, mit Finnjet on yksi maailman
uns auf der Finnjet- suurimpia matkustalaivoja.
einem der grossten
Passagierschiffe der Welt
- nach Finland zu
kommen!
[We invite you to come [The Finniet is one of the
with us to Finland aboard world's largest passenger
the Finniet - one of the ships]
world's largest passenger
ships!]

4 Cf. our discussion in ch. 1, pp. 12-14 above.

5 Key to the form of presentation: I have lined up the two texts in the following way:
text parts that take up corresponding positions are placed side-by-side; where the corre-
sponding parts differ significantly in information content, the differences have been
italicized in the Finnish; information altogether absent from one version is indicated by
blank lines in the respective co1unm. For easier reference, I have numbered the blocks
of text in the German column. An English key is provided in square brackets after each
sentence.
52 Translation and Relevance

3 Trotz der machtigen Mahtavasta koostaan

Grosse ist die Finnjet ein huolimatta Finnjet on
aussergewohnlich kuitenkin hiimmastyttavan
'spritziges' Schiff! nopea, ...
[Despite its huge size, the [Despite its huge size, the
Finnjet is an extraordinar- Finnjet is surprisingly
ily 'lively'ship'] fast, .. .]
4 Denn die Finnjet wird
als einziges Passagier-
schiff der Welt von
Gasturbinen angetrieben-
wie ein Dusenflugzeug.
[For the Finnjet is the
world's only passenger
ship that is driven by gas
turbines -like ajet plane.]
5 In nur 3 Minuten '" silla jo 3 minuutissa se
kommt dieser Wellen- saavuttaa kaasuturpii-
Riese mit Hilfe seiner niensa ansiosta 30.5
Gasturbinen auf die solmun huippunopeuden,
Spitzengeschwindigkeit toisin sanoen 56
von 30.5 Knoten, das sind kilometria tunnissa!
etwa 56 Stundenkilometer! [Oo .for with the help of its
[With the help of its jets this jets it reaches the top speed
wave-giant gains the top of 30.5 knots, in other
speed of30.5 knots, that is words of56 kilometres per
about 56 kilometres per hour, injust 3 minutes!]
hour, in just 3 minutes!]
6 Eine kaum vorstellbare Taman kokoisen jattilaisen
Leistung bei einem Schiff kohdalla se tuntuu miltei
dieser Grossenordnung! uskomattomalta!
[An achievement hardly [For a giant of this size this
imaginable for a ship of seems almost incredible!]
this magnitude!]
7 1m Hochsommer erreicht Keskikesalla Finnjet
die Finnjet von Travemunde taittaakin Helsinki-
aus in nur 22.5 Stunden den Travemunde-valin perati
Hafen von Helsinki! 22.5 tunnissa!
[During the summer peak [In the summer peak
season the Finnjet reaches season the Finnjet travels
the harbour ofHelsinki from from Helsinki to
Travemunde in a mere 22.5 Travemunde in only 22.5
hours!] hours.]
Covert Translation 53

8 Genauso beeindruckend Myos muilta teknisilta

wie die Kraft ist auch die ominaisuuksiltaan Finnjet
technische Ausstattung der edustaaa huippuluokkaa: ...
Finnjet. ...
[Just as impressive as its [Also in its other technical
power is also the technical qualities the Finnjet
equipment ofthe Finnjet .. .] represents top class.]

(The remainder of this fairly long paragraph is almost identical in content

from sentence to sentence between the two versions. Quoted below are the
last two paragraphs of both versions.)

9 Etwa eine halbe Stunde Myos satamaan saapumista

vor der Ankunft in ihrem on aina mukava seurata.
Heimathafen passiert die
Finnjet das Schiirengebiet
vor Helsinki.
[About half an hour [Also the arrival in the
before arrival in its home harbour is always nice to
port, the Finnjet passes watch.]
through the skerries off
Helsinki.]
Idyllinen Travemtinde
rantabulevardeineen on
kuin satukirjasta.
[The idyllic Travemunde
with its beach boulevards
is like out of a fairy tale
book.]
10 Mindestens genauso Itse laituriin
spanned wie tiber Wasser kiinnittyminen, (on) kuin
ware ein paar Minuten taitavan kuljettajan auton
spater das Anlegemanover pysakointi jalkakaytavan
aus der 'F ischperspektive' . viereen.
[The docking manoeuvre [The docking to the quay
from the 'fISh perspective' itself is like a skilful driver
would be at least as exciting parking his car by the
as from above water.] pavement.]
11 Mit Hilfe der zwei Kahden perassa olevanja
Schiffsschrauben am keulan molemminpuolisten
Heck, der beiden potkureiden seka erinomai-
Querstrahlruder links und sen sonnenpaatuntuman
rechts und mit viel avulla Finnjet tekee
54 Translation and Relevance

FingerspitzengefUhl wird 'pehmean laskun' laituriin.

die Finnjet sanft zu ihrem
Terminal gebracht.
[With the help of the two [With the help of its two
propellers at the stern, of propellers at the stern
the two lateral-thrust and on both sides of the
rudders left and right, bow as well as with
and with much instinctive outstanding skill the
feel the Finnjet is taken Finnjet makes a 'soft
softly to its terminal.] landing' at the quay.]

What relates these two texts to each other is the fact that: a) they are
placed side by side in this bilingual brochure, on a page that has a map at
the top with the travelling route marked; b) the texts are quite parallel in
their structure; and c) they agree substantially in their information content -
about 50 per cent of the text gives virtually the same information, mostly
concerning the technical aspects.
On the other hand, there are very clear differences between them. With-
out going into great detail, the following examples may serve to illustrate
this. The titles differ almost completely in content; the information con-
tained in sentence 4 of the German version is missing from the Finnish one.
Particularly interesting is a comparison ofthe last two paragraphs (sentences
9-11); both versions talk here about arrival at the destination - but whereas
the German one describes matters related to arriving in Helsinki, the Fin-
nish gives a picturesque description of arrival in Travemlinde.
The question is: how can we account for the relationship between these
Finnish and German texts? Formulating a general translation theory offunc-
tional equivalence that is explicit, coherent and can accommodate the kinds
of differences just discussed seems a formidable task indeed.

Translation - When All is Change?

An extreme case of translation along these lines is discussed by Honig and

Kui3maul (1984). It involves an advertisement by Viyella House in the Sun-
day Times that begins as follows:

WHAT'S IN A NAME?
It sounds ordinary on paper. A white shirt with a blue check. In fact,
if you asked men if they had a white shirt with a blue check, they'd
say yes.
Covert Translation 55

But the shirt illustrated on the opposite page is an adventurous white

and blue shirt. Yet it would fit beautifully in your wardrobe. And no
one would accuse you of looking less than a gentleman ... (Cited in
Honig and KuBmauI1984:36)

Honig and KuBmaul contrast two different specifications for a transla-

tion of this advertisement: one asks for a translation to be used in a study of
the marketing strategy of Viyella House, the original company; the other
asks for a corresponding advertisement to be used in Germany. While the
first requires preservation of the content of the original, the second requires
preservation of its intention.
With regard to the second specification, the authors point out that the
aim of preserving the intention of the original could result in a German
version in which possibly "not a single word would be reminiscent of the
'original'" (1984:40); in fact, they concede, the German version might not
talk about a shirt at all but perhaps centre around an item of children's wear.
While admitting that both theoreticians and practicians might prefer to use
a term other than 'translation' for the second case, the authors maintain that
such cases do fall under translation: 6

Weare of the opinion that we are dealing here with two equally valid
basic types of translation ... These two basic types can be designated
as 'functional constancy' ['Funktionskonstanz'] and 'functional
change' [' Funktionsveranderung ']. They are completely equally valid
and equally legitimate strategies of translation between which the
translator has to choose for every text. It is therefore by no means
true that functional constancy can be taken to be the normal case of
translation, whereas functional change is the exotic exception.
(1984:40, translation my own)

Thus Honig and KuBmaul propose a theory of translation that does not rely
on text functions as factors that have to be kept constant in translation, but
considers them as variables, too.
What, then, if anything, differentiates translation from other forms of
interlingual communication? The answer the authors give is ambivalent. Its
first part seems to be at variance with the idea of function-changing transla-
tion in that it suggests that the crucial point about translations is that both
their contents and function are not subject to the translator's intentions:

6S nell and Crampton, for example, belong to those who are doubtful that advertising
should be dealt with by translation: "Translation has little to do with this fascinating
area of communication" (1983:112).
56 Translation and Relevance

Translational communication differs from other forms of written

communication essentially in that neither the contents nor the func-
tion of the text originate in the sender's and text designer's, i.e. the
translator's, individual desire to communicate. (1984: 13, translation
my own)

Where, then, do the contents and functions of translated texts come from?
The most obvious answer would seem to be that typically they are derived
from the original. Yet this seems to be denied by the very next sentence which
again affirms that the communicative functions of the original are not con-
stant factors, but variables that depend on the 'purpose of the communication':

In this way all communicative functions - from the function of the

text to the function of the individual word - become variables, which
need to be determined with regard to the respective purpose of com-
munication. (1984:13, translation my own)

It seems, then, that Honig and Kui3maul have not really succeeded in clari-
tying what the essence of translation is, especially as regards the relationship
between original and translation: if a translation can differ in virtually all
aspects from the original- what makes the translation a translation??
However, before trying to clarity how the relationship between transla-
tion and original can be characterized it seems worth asking another preliminary
question first, namely: what point is there in relating these target-language
texts to the originals at all?

Descriptive Use in Interlingual Communication

Returning to the example of the tourist brochure on Nuremberg, one has

to ask: what really is the point here of comparing original and translation
in terms of functional equivalence? The Verkehrsverein Ntirnberg, the

7 In their summary of the strategy of translation which they suggest, the authors state

that the translator "takes note ofthe source language text, and relates it to his situation
as translator. He specifies the translation task and determines the communicative func-
tion of the target language text, orienting himself toward the pragmatic expectations of
his addressees" (1984:58, translation my own). All this makes sense; the only problem
is that, as in other strategy-oriented approaches, these authors, too, content themselves
with proposing these various strategic steps and illustrating them from examples - without
spelling out explicitly the criteria involved in the various specification, determination
and derivation processes. Thus what actually takes place in these steps is largely left
implicit.
Covert Translation 57

publisher of the tourist booklet under discussion, is presumably much more

interested in whether its English brochures are as effective as possible in
attracting tourists than in how closely any of these brochures matches up in
a point-by-point comparison with a corresponding German brochure in terms
of some translation-theoretic notion like 'functional equivalence'.
The same seems true from the receptors' point of view: what they are
looking for in the brochure is not functional equivalence with a correspond-
ing German version; in fact, it would be immaterial to them whether there
were a corresponding version in German at all. All that would matter to
them is that they are given information relevant to them and their plans.
For the Finnish-German example, too, the same point can be made: the
existence of a similar piece of writing in the other language is irrelevant to
the communication act.
The example from Honig and Kui3maul is perhaps the most informative
here because, as mentioned above, the authors explicitly consider two dif-
ferent ways of translating the advertisement. A closer look will help us to
see that these two options differ in rather fundamental respects. Honig and
Kui3maul formulate the specifications for the options as follows:

The Viyella House company now want to sell their products also in
Germany. Make a translation which shows what marketing strategy
they use in England. Perhaps you will want to add a few lines as well.
Or
The Viyella House company has sent us this advertisement; we should
put together something corresponding to it. So get together with one
of our advertising experts, translate the advertisement for him and
then work with him to draft a good advertisement for our area.
(1984:39, translation my own)

As I tried to show above, in the second case the original advertisement

in English has little bearing on the German version: the German target au-
dience need not even be aware of its existence, and the success ofthe German
version would not be measured in terms of its resemblance to the original.
As the authors themselves state, there need not be any resemblance at all
between the original and the German advertisement. In fact, the English
advertisement could well tum out to be a distraction: the differences in socio-
cultural background might be such that it would be better to design a new
advertisement altogether. So the German advertisers might set the English ad-
vertisement aside and came up with a completely different advertisement,
58 Translation and Relevance

and, crucially for our discussion, there would be nothing wrong with this:
all that mattered would be whether the advertisement managed to boost the
sales of Viyell a House articles.
The first specification, however, is clearly different: here the existence
of the English original is crucial to the translation task - the request could
not be fulfilled ifthere were no English original to translate; the point of the
translation is to bring out part of the content of that English original, and to
draw further inferences from it about the marketing strategy underlying it.
In summary, then, we can make the following observation: in the first
two examples - the tourist brochure on Nuremberg and the write-up on the
Finnjet - and in the case of the second specification given for translating
the Viyella House advertisement, there is no necessary relationship between
the source language and the target-language texts. In fact, in each case the
existence of the source language original was incidental to the communica-
tion act: the brochure on Nuremberg, the article about the Finnjet ferry, and
the German advertisement for Viyella House could all have been composed
without reference or resort to the source language' originals', and the com-
munication act could still have succeeded.
With the first option of the Viyella House example, however, the trans-
lation was crucially dependent on the source language original, and this
seems to be the case in most translations: for example, an English transla-
tion of Goethe's Faust or a German translation of Shakespeare's As you
like it is crucially dependent on the existence and content of the original
works; their whole point is to represent those original works, and their suc-
cess depends on the degree to which they achieve this. One could, of course,
compose receptor language originals that had virtually the same informa-
tion content as the source language texts, or used the same style, tried to
achieve similar effects and so forth. However, they would be read and inter-
preted very differently depending on whether they were presented as works
in their own right or as representations of those famous source language
originals.
In relevance theory this difference is accounted for in terms of the dis-
tinction between descriptive and interpretive use. As pointed out in chapter
2 above (pp. 35-9), relevance theory claims that this distinction is rooted in
human psychology: human beings have two different ways of entertaining
thoughts - they can entertain them descriptively, in virtue of their being
true of some state of affairs, and they can entertain them interpretively, in
virtue of the interpretive resemblance they bear to some other thoughts, and
Covert Translation 59

this difference makes itself felt in communication as well.

In fact, the examples just considered seem to have in common that they
all constitute instances of descriptive rather than interpretive use: the receptor
language texts are intended to achieve relevance in their own right, not in
virtue of their interpretive resemblance with some source language original.
But how, then, can we account for the similarities and differences be-
tween the source and receptor language texts?
Returning to our Finnish-German example, I do not have information on
which of the two versions, if any, was the' original'. However, for the sake
of argument, let us assume that the Finnish version was produced first, then
the German one - though both might have been produced interdependently.
The intended effect of this part of the brochure was probably to impress
travellers with the special technical equipment of the ferry and to arouse
positive feelings about their destination.
Why should the accounts be so similar, for example in the technical
parts? One plausible explanation is that to draft the Finnish version required
a significant amount of effort for the communicator, such as the compila-
tion oftechnical information, then relevance-based selection and then writing
it up in a structured and coherent form. So when it came to making the
German version, since much of the information was assumed to be relevant
also to the German target audience, it was probably much more convenient
to start from the Finnish text rather than creating a totally new write-up
from scratch.
At the same time, there would have been no a priori obligation to main-
tain functional equivalence or faithfulness to the Finnish version in the
content or structure of the German one; as pointed out above, it is not even
necessary to assume that there was a complete Finnish original text prior to
the German one, though there may well have been one. Most likely the
advertising department of this shipping line would be more concerned about
getting a German version that would be effective along the general lines
spelled out above than about getting a functionally equivalent or faithful
representation of the originaI.S
Thus if the communicator responsible for the German version felt there
was information of high relevance for the German audience that was not

8 Of course some of the information, e.g. the technical information, in the two versions
should not be in conflict; however, the reason here is descriptive accuracy, not interpre-
tive faithfulness: the information content in the two versions will be the same to the
extent that they both describe the same object.
60 Translation and Relevance

included in the Finnish version, or ifhe found a more effective way of pre-
senting the same information to his German target audience, it would only
be in the interest of the shipping agency ifhe went ahead and composed the
German version accordingly. There would be no point in resisting such
changes because they violated some translation-theoretic notion like func-
tional equivalence.
The differences in title might be a case in point: they may well reflect
assumptions by the communicator about what is most relevant to the re-
spective target audience: while the first concern of the typical Finnish
traveller envisaged might be economy - hence appeal to the "most direct"
route -, it might have been thought that the typical German traveller is more
concerned about enjoying himself than about the economic aspect - hence
the appeal to "the most beautiful way to Finland". In both cases the title
chosen is the one thought most effective for the target audience, regardless
of its relationship to some other language version.
Other, more trivial factors may have affected the composition process -
such as considerations of space. Thus it seems possible that the Finnish
original did have an equivalent to the fourth sentence of the German, but
perhaps when it came to putting the brochure together, it might have turned
out that the two texts did not fit in the space provided. The editor might
have decided on cutting out this particular sentence in the Finnish version
because, Finnjet being a Finnish enterprise, Finns would quite likely know
that this ferry is jet-driven, especially since it got a fair bit of coverage in the
Finnish media when it was first taken into service. On the other hand, Ger-
mans would be much less likely to have any detailed knowledge about the
Finnjet - and so it would seem highly relevant to retain this information in
the German version - imagine you could impress your friends after your holi-
day, telling them that you went on a ship driven by jets, just like a jet-plane!
The similarity and difference between the two versions of section 9 could
also be explained in terms of estimates of relevance: travelling informa-
tion is generally more relevant to first-time travellers - hence more likely to
be read by such. Information about their destination is usually more rel-
evant to travellers than information about their point of origin; most
German-speaking first-time travellers on the Finnjet would be going from
Travemlinde to Helsinki; hence it would be more relevant to them to in-
clude information about arriving in Helsinki than about departing from
Travemlinde, though Travemlinde is what the assumed Finnish original talks
about, for exactly the same general reasons.
Covert Translation 61

Thus it seems possible to account for the differences and similarities

between these texts on purely relevance-theoretic grounds, without resort-
ing to translational concepts, such as 'equivalence' or 'faithfulness', at all.
The writer of the German text may have found the Finnish original helpful
in his task, but the text he created was intended to communicate in its own
right, and to be read and understood as such.
It may be worth pointing out that this distinction between descriptive
and interpretive use applies not only to interlingual communication - paral-
lel cases occur intralingually as well. For example, speaker A might present
a summary of how human beings communicate, and speaker B might present
a summary of Sperber and Wilson's relevance theory of communication.
The two presentations could be virtually identical in information content,
but the first would be presented descriptively, that is, as a representation of
what the speaker believed to be true of human communication, whereas
speaker B would be communicating interpretively: that is, his presentation
would be intended to achieve relevance as a representation of the views of
Sperber and Wilson of communication.
This difference could have important consequences. In the first case,
speaker A would be understood as presenting his own views, and hence as
committed to their truth: if the speaker did not, in fact, believe these views
for himself, he would be accused of untruthfulness, and he could not defend
himself by saying that Sperber and Wilson believed this information to be
true; that fact would be besides the point ifhe had not presented the infor-
mation as their beliefs.
Speaker B, by contrast, would be expected accurately to represent the
thinking of Sperber and Wilson. If the information he presented actually
differed from the views held by Sperber and Wilson, he would be liable to
the charge of misrepresentation, and he could not defend himself saying
that he had presented these views because he believed them to be true: what
mattered would be whether he had faithfully conveyed the position of Sperber
and Wilson.
Thus again we see that the distinction between descriptive and interpre-
tive use has important implications for successful communication.
Looking into the theory and practice of translation, it seems that this
failure to distinguish between descriptive and interpretive use has had un-
fortunate results for the understanding and accomplishment of interlingual
communication.
On the theoretical side, it has landed scholars with the enormous task of
62 Translation and Relevance

trying to devise "a theory of translation [that] will do justice to both Bible
and bilingual cereal packet" (Kelly 1979:226).
On the practical side it has been unhelpful, too. An interesting example
here is the massive 'translation task' that arose in the process of 'Vietnami-
zation', described by Brislin (1976): at that time the Vietnamese people
were to take over jobs from the Americans; many of these jobs involved the
handling of advanced technical equipment. Brislin writes:

Training people to handle such equipment may seem straight-for-

ward enough, but it is easy to forget that tremendous numbers of
technical documents, both for instruction in and maintenance of this
equipment, must accompany any transfer from one group of opera-
tors and maintenance specialists to another. (1976: 12)

This example is interesting for a number of reasons; firstly, it illustrates

the fact that the transfer of information from one language to another is
almost automatically seen as a task that calls for translation. There is no
indication in Brislin (1976) that any other alternative was considered.
Secondly, this example shows how influential thinking along transla-
tional lines can be, colouring even the objective of the enterprise and its
evaluation; thus when discussing what criterion should be used to evaluate
the translations produced for this task, Brislin writes:

In operational terms, the criterion meant that people using the trans-
lated materials should be able to maintain the equipment as well as
people using the material in its original language form. (1976: 14f.)

Consequently one kind of testing applied to these translations was 'per-

formance testing', conceived of as "a way of measuring the quality of
technical translations by having readers perform a task using the transla-
tions" (1976:16). Note how the result is described:

The best translation into Vietnamese resulted in performance on a

very difficult, multistep task at a level that was approximately equal
to that of United States Army technicians who used the English
manual (... ). Error-free performance was observed for 73 percent of
the tasks done by each group, United States Army and Vietnamese
Air Force. (1976:17)

What is significant here is the fact that the evaluation is done in terms of
'equivalence' in performance between source language speakers and receptor
Covert Translation 63

language speakers - rather than, for example, in terms of adequacy of per-

formance. In other words, instead of asking 'Is the performance of
Vietnamese using the translation equivalent to the performance ofthe United
States Army personnel?', one could have asked: 'Is the performance of Vi-
etnamese technicians using the translation adequate for the job?' After all,
for the overall purpose of the exercise of 'Vietnamization', the main con-
cern would be that the Vietnamese people should be able to handle the
equipment properly, rather than 'as well as group X of individuals'.
Note that this is not a purely academic point: suppose, for example, that
the 'translated' manuals had actually been evaluated in terms of adequacy;
suppose further, that the quality of an instruction manual is considered' good'
if it enables technicians using the manual to carry out the operations
described error-free to a degree of 85 percent. In that case, the manuals -
English or Vietnamese - that allowed 73 percent error-free performance
would not have been considered good, but would have been seen as need-
ing further improvement. The improvement needed may not have been
obvious in terms of translation; it might well have been that due to differ-
ences in technical and cultural background the improvements may have led to
Vietnamese versions that differed considerably from the English originals.
Snell and Crampton (1983), for example, seem to be aware of the fact
that foreign language versions of instructional material need to be evalu-
ated in terms of their own adequacy rather than by comparison with the
original text. Thus they recommend that the translator should be "extremely
thorough" in checking his translation, and when doing so, "he should work
through the instructions with the apparatus, and forgetting that he has ever
read the original, make quite sure that his finished work describes the func-
tion" (1983: 113, italics as in original). The suggestion to 'forget' the original
in the testing process is a clear indication that the ultimate measure for such
text material is its adequacy for the job rather than faithful representation of
the original as such. 9
But why then are such cases of communication commonly considered

9 Two further comments seem to be in order. Firstly, it seems that testing would be more

effective if it was done by typical members of the target audience rather than by the
translator: it would be difficult for the translator to really 'forget' all his knowledge
about the product, and if his educational standard is higher than that of the average
reader, his ability to follow the instructions may not indicate that his target audience
could do the same. Secondly, it is interesting that Snell and Crampton (1983) point out
the need that the translator may have to educate his clients about the necessity of such
practical testing:
64 Translation and Relevance

instances of translation? 10 I think there are several reasons. One is that there
is a tendency to use the word 'translation' rather loosely to refer to almost
any instance of communication that involves the transfer of information
from one language to another with the help of a bilingual person.
More importantly, as briefly mentioned with regard to the Finnjet tourist
brochure, it is in some ways very convenient to actually translate an already
existing piece of source language literature rather than to work out a new
receptor language text, even if all that matters is that the receptor language
text achieves successful communication in its own right.
This factor was clearly present in the case of the 'Vietnamization' ex-
ample just considered; it was obviously more efficient to start from the
existing English language manuals than to draft new ones in Vietnamese
from scratch. So from this perspective translation seemed to be the most
economical approach to take, especially in the hope that at least part of the
job could be taken over by computers. However, as I tried to point out above,
problems can arise when the communicator does not clearly recognize that
the role of the source language text is merely that of a convenient help for
composing a receptor language text, not of a model to be faithfully repro-
duced. If this is not clear, the objective of the communication act can be
obscured, and hence its achievement is endangered.
A third reason seems to be that in cross-language situations where cul-
tural differences are minimal, translation can go a long way toward achieving
successful communication in the receptor language. But the more relevant
the sociocultural differences are to the communication act, the less success-
ful translation will tum out to be.
Another negative effect of the failure to distinguish instances of inter-
lingual communication involving descriptive use from those involving
interpretive use is that the results which interlingual descriptive use can
achieve have been put forward as challenges to translators actually engaged
in interpretive interlingual use.
Consider the following quotes from Wonderly (1968) and Nida and
Taber (1969):

Clients may regard your insistence as rather a nuisance, but if they have
any understanding of your job they will respect you for your stand. If they
had no appreciation of the point of view of a translator beforehand, you
will have helped to educate them in the ways ofthe profession. (1983: 113)
10 Though excepting advertising from the domain of translation, Snell and Crampton

(1983) do not indicate such reservations towards the translation of either 'publicity and
sales literature' or 'instructional material'.
Covert Translation 65

a handbook on automobile maintenance written for the mechanic may

use the normal technical language of the trade; but if the same infor-
mation were to be adequately translated into language suitable for a
history professor, many technical concepts would have to be ex-
pressed by descriptive phrases and explanatory sentences, and the
resulting material would be much longer. (Wonderly 1968:40)

One specialist in translating and interpreting for the aviation indus-

try commented that in his work he did not dare to employ the
principles often followed by translators of the Bible: 'With us,' he
said, 'complete intelligibility is a matter of life and death.' Unfortu-
nately, translators of religious materials have sometimes not been
prompted by the same feeling of urgency to make sense. (Nida and
Taber 1969:1)

Both passages are presented to Bible translators in order to give them a

better idea of the nature of their task - to achieve good translations. What is
not pointed out is that their tasks are quite different in nature: while the
writer of the receptor language manual is free to make whatever changes
will enable the target audience to carry out their technical tasks better, even
if this leads to changes of content, the Bible translator is bound to the con-
tents of a unique original, the Bible. On analogy to the writer of a technical
manual, the biblical translator could, of course, see his primary task as in-
fluencing the thinking and behaviour of the receptors in the most effective
way; and he could therefore feel free to add to, delete from and generally
change the contents of the biblical texts to that end. However, it would be
doubtful whether these would or even should be considered translations of
the biblical texts by readers interested in the contents of the original.
One rather important question that remains is: what is the role ofthe trans-
lator or interpreter in such instances of interlingual descriptive use - certainly
a matter of interest to all those engaged in interlingual communication.
It is useful to distinguish two different situations here. One is the case
where the original communicator himself is trying to address the receptor
language audience. The tourist brochure examples looked at above would
be cases in point, and so would be those of the manufacturer concerned to
provide technical manuals in a foreign language for the use of his products
abroad: in both cases the source language agent can be seen as trying to
communicate relevant information to the receptor language audience.
In these kinds of situation the translator or interpreter clearly does have
a vital function. However, this function is not an independent one, but can
66 Translation and Relevance

be seen as part of a complex process of stimulus production: the obvious

handicap of the source language communicator is that he does not master
the receptor language, and so he needs the assistance of a bilingual indi-
vidual for producing a receptor language stimulus that will accomplish his
communication objectives.
While we tend to think of communication in terms of a single, individual
communicator, it is not difficult to see that there are many instances even of
intralingual communication which are much more complex than that.
An obvious example is printed publication: here the process of stimulus
production is very complex indeed - beginning with the writing of the first
pages of the manuscript, through revisions, through a process of editing,
through typesetting up to the final printed product. There are a number of
different individuals involved in this process: the writer himself, the editor
and the printer - though in many situations there will be more: the editor
might be aided by reviewers, there would be typesetters, and there may be
artists, all co-operating with the original author to produce the stimulus -
article, pamphlet, book - that will serve his desire to communicate.
Another case might be the production of advertisements: here one might
well know from the beginning what kind of effects one wants to achieve -
but it might take a group of specialists from marketing and the advertising
industry to design and produce an appropriate stimulus, and the stimulus
may not necessarily involve a single medium, like printing, but may take
the form of a multi-media show. In all these cases the process of stimulus
production is complex because the original communicator himself does not
possess all the know-how, skills and equipment necessary to produce the
stimulus he wants.
Viewed in this light, then, the case of interlingual descriptive communi-
cation considered here seems quite parallel: the original communicator lacks
a certain skill necessary for producing the desired stimulus, that is, mastery
of the receptor language code, and so he employs the help of a suitable per-
son who has this skill - the interpreter or translator.
One might say that in such situations the presumption of relevance is
shared between two individuals: the source language communicator bears
responsibility that the set of assumptions he intends to make manifest to the
addressee is "relevant enough to make it worth the addressee's while to
process the ostensive stimulus" (Sperber and Wilson 1986a:158). The trans-
lator's or interpreter's task is to ensure that the receptor language text he
comes up with" ... is the most relevant [ostensive stimulus] the communi-
cator could have used to communicate {I}" (1986a:158).
Covert Translation 67

The source language communicator may, of course, need more help than
just with 'transcoding'; he may, in fact, want to use some of the translator's
or interpreter's knowledge of the receptor language culture to ensure that
what he intends to communicate will be adequately relevant to the receptor
language audience, and in this way the bilingual agent can have an influ-
ence even on the objective of the communication act.
It is encouraging to see that - at least some parts of - the professional
world of translation today seem to sense that this kind of interlingual commu-
nication service is different from translation, as the following advertisement
suggests:

Weidner translates words into business. Consultancy-based approach.

We will consult you, both before and during the translation process.
Our close involvement with your project enables us to identify with
and understand your corporate objectives. This appreciation of the
background to and the aims of each assignment is the key to provid-
ing a translation which achieves the impact you are seeking in the
target market. 11

Crucially, the advertisement goes on to say:

More than a translation service .... Your translation requirement needs

to be managed like any other aspect of your business; it has to dove-
tail into your overall marketing and publishing plans. We understand
your environment and we offer very much more than a translation
service. (Ibid., italics my own)

In the second kind of situation the bilingual may be a communicator in

his own right, not tied to any source language agent at all. Thus carrying the
Viyella House example a bit further, the German agency asked to develop a
German advertisement on the basis of the English one may be given a free
hand by the parent company as to what exactly they should produce; it might
be left up to the German branch of the company to evaluate the advertise-
ment in terms of its likely effectiveness. In that case the sole function of the
'original' might have been to stimulate the thinking of the receptor lan-
guage communicator.
In summary, I have tried to show in this chapter that there are instances of
interlingual communication that have often been referred to as 'translation'

llAdvertisement on back cover of Language Monthly, no. 47, August 1987.

68 Translation and Relevance

and that translation theorists have tried to account for, but that differ from
other instances of translation in that the source language original is inciden-
tal rather than crucial to the communication act. I have tried to argue that in
relevance theory these instances of interlingual communication can be ac-
counted for in terms of descriptive use. If this is correct, then there will be
no need for a general theory of translation to concern itselfwith such cases.
4. Translating the Meaning of the Original

If one were to ask around what people think a translation should achieve, a
very common answer would probably be that it should communicate the
meaning of the original accurately and clearly to the readers of the transla-
tion. This has not always been so - thus certain philological traditions have
tended to stress the preservation of stylistic and other linguistic characteris-
tics of the original. l However, since the 1960s, there has been a strong trend
in translation theory and practice to pay special attention to how well the
translation communicates to the target audience. 2 This concern for the im-
pact of a translation on the receptor language audience has probably found
its fullest development in circles concerned with the translation of the Bible,
though it is not limited to this enterprise, as, for example, Larson (1984) shows. 3

Conveying the 'Message' of the Original

The first approach along these lines that developed into a comprehensive
theory is that of 'dynamic equivalence' developed by Nida (1964; Nida and
Taber 1969). It derives its name from the fact that its primary concern is "with
the dynamic relationship ... that the relationship between receptor and mes-
sage should be substantially the same as that which existed between the
original receptors and the message" (Nida 1964, p. 159). In the more fully
developed version contained in Nida and Taber (1969), the authors intro-
duce their approach in the following terms:

The older focus in translating was the fonn of the message, and trans-
lators took particular delight in being able to reproduce stylistic

1 Cf. e.g. the position of Longfellow: "The business of a translator is to report what the

author says, not to explain what he means; that is the work of the commentator. What an
author says and how he says it, that is the problem of the translator." (Quoted in De Sua
1964.) Cf. also Arnold (1861).
2 Newmark (1988) links this trend to the rise of 'modern linguistics': "Since the rise of

modern linguistics ... and anticipated by Tytler in 1790, Larbaud, Belloc, Knox and
Rieu, the general emphasis, supported by communication-theorists as well as by non-
literary translators, has been placed on the reader - on informing the reader effectively
and appropriately, notably in Nida, Firth, Koller and the Leipzig School" (1988, p. 38).
3 Wilss (1982, p. 148f.) includes Kade (1968b) among the communicative approaches;

however, Kade seems to differ from the approaches considered here in presenting equiva-
lence in semantic content as the primary concern.
70 Translation and Relevance

specialties, e.g. rhythms, rhymes, plays on words, chiasmus, paral-

lelism, and unusual grammatical structures. The new focus, however,
has shifted from the form of the message to the response of the
receptor. (Nidaand Taber 1969:1)

In line with this orientation, Nida and Taber define dynamic equiva-
lence as follows:

Dynamic equivalence is therefore to be defined in terms of the de-

gree to which the receptors of the message in the receptor language
respond to it in substantially the same manner as the receptors in the
source language. This response can never be identical, for the cul-
tural and historical settings are too different, but there should be a
high degree of equivalence of response, or the translation will have
failed to accomplish its purpose. (Nida and Taber 1969:24)

One important aspect of 'audience response' lies in correct understand-

ing of the meaning:

Correctness must be determined by the extent to which the average

reader for which a translation is intended will be likely to understand
it correctly. (Nida and Taber 1969: 1)

Thus the main objective of translation is concerned with conveying the

meaning of the original text:

Translating must aim primarily at 'reproducing the message'. To do

anything else is essentially false to one's task as a translator. (Nida
and Taber 1969:12)

However, comprehension of the original message is not enough:

It would be wrong to think, however, that the response of the receptors

in the second language is merely in terms of comprehension of the
information, for communication is not merely informative. It must
also be expressive and imperative if it is to serve the principal pur-
poses of communications such as those found in the Bible. That is to
say, a translation of the Bible must not only provide information which
people can understand but must present the message in such a way
that people can feel its relevance (the expressive element in commu-
nication) and can then respond to it in action (the imperative function).
(Nida and Taber 1969:24)
Translating the Meaning of the Original 7l

The idiomatic translation approach developed by Beekman and Callow

(1974) is also concerned with Bible translation. It resembles the dynamic
equivalence approach in its rejection of form-oriented translation, and its
emphasis that a translation should convey the meaning of the original. It
also demands that the translation should be faithful to the 'dynamics' of the
original, but it looks at these dynamics in terms of 'naturalness' oflanguage
use and ease of comprehension rather than receptor response:

A translation which transfers the meaning and the dynamics of the

original text is to be regarded as a faithful translation. The expres-
sion transfers the meaning, means that the translation conveys to the
reader or hearer the infonnation that the original conveyed to its read-
ers or hearers ... The expression, the dynamics, means that (1) the
translation makes a natural use of the linguistic structures of the RL
[receptor language] and that (2) the recipients of the translation un-
derstand the message with ease. (1974:34, italics as in original)

Larson (1984) presents an extension of the idiomatic approach to trans-

lation in general. In substance her model is very similar to that of Beekman
and Callow (1974), though she does include the aspect of audience response
in her definition of 'dynamics':

The underlying premise upon which this book is based is that the
best translation is the one which a) uses the nonnallanguage fonns
of the receptor language, b) communicates, as much as possible, to
the receptor language speakers the same meaning that was under-
stood by the speakers of the source language language, and c)
maintains the dynamics of the original source language text. Main-
taining the 'dynamics' of the original source text means that the
translation is presented in such a way that it will, hopefully, evoke
the same response as the source text attempted to evoke. (1984:6)

Taking these approaches together, we can see that they share the follow-
ing two basic obj ectives: 1) a translation must convey to the receptor language
audience the meaning or message of the original; and 2) it must do so in a
way that is faithful, viz. equivalent to the dynamics of the original - keep-
ing in mind that there are differences in what is meant by 'dynamics'.
What do these objectives correspond to in terms of relevance theory?
Before we can answer this question we need to ask what is meant by the
'meaning' or 'message' in these approaches. Nida and Taber define:
72 Translation and Relevance

Message: the total meaning or content of a discourse; the concepts

and feelings which the author intends the reader to understand and
perceive. (1969:205)

Beekman and Callow do not state explicitly what they understand by

'meaning', but their definition of what it means to 'transfer the meaning'
(see above) suggests thatthe 'meaning' of the original refers to "the informa-
tion that the original conveyed to its readers or hearers" (1974:34).4
Larson (1984) does not define her notion of 'meaning' either. At one
point she identifies it with the 'deep' or 'semantic' structure of language,
and for her this seems to correspond to information content. However, she
also talks about 'emotive meaning' - thus a communicator "may choose
one form over another in order to give a certain emotive meaning in addi-
tion to the information he wishes to convey" (Larson 1984:32, italics as in
original). She states clearly that the meaning of the original is to be viewed
as "the meaning intended by the original communicator" (1984:17).
Expressed in terms of relevance theory, what these approaches suggest
is that a translation should convey to the receptor language audience a lit-
eral interpretation of the original (cf. chapter 2, pp. 41, 46) or one closely
resembling it.
In the remainder of this chapter we shall examine whether this claim can
provide the basis of an explicit general theory of translation. Since our dis-
cussion will pay special attention to the approaches just mentioned, we shall
follow their distinction between the 'message' and the 'dynamics' of the
translation, considering each in tum.
While there are some disagreements about how these dynamics are to be
viewed, these approaches show good agreement on the need for the transla-
tion to convey the original message, or, in our terms, the set of assumptions
{I} the original communicator intended to convey. Therefore in our dis-
cussion we shall focus on this claim, though we shall also give some
consideration to the question of the' dynamics'.
To do this, we shall take the example of a biblical text, consider its -
probable - intended original meaning, and then examine the prospects of
conveying this meaning to a contemporary English audience by translation.

4 In a personal communication Callow told me that this statement should really have

talked about meaning in terms of the information the author intended to convey rather
than what was perhaps understood by the audience; however, in terms of our discus-
sion, this point is not of importance.
Translating the Meaning of the Original 73

The example I have chosen is the second chapter of the Gospel of Mat-
thew. 5 This chapter begins with a report of how some magi came to pay
homage to Jesus as the new-born king of the Jews; this visit resulted in the
flight of Joseph and his family to Egypt and the slaughter of infants in Beth-
lehem. The chapter ends with an account of how the fugitives returned from
Egypt and came to live in Nazareth.
However, Matthew does not just report these events, but combines them
with copious allusions to and quotations from the Old Testament, some-
times adapting the quotations in certain ways. This suggests that he intended
to convey in this chapter something more than just a report of certain events
- but what did he intend to communicate? Considering that this chapter is
the only one that gives information about the first thirty or so years of Je-
sus' life - why did Matthew choose this particular combination of narration
and quotations from the Old Testament?
With France (1981) we shall "want to ask simply what the author is
trying to get across by his selection of Old Testament texts in this chapter,
how he goes about communicating his meaning ... and how far we may
judge him to have been successful in communicating his thoughts to his
putative readers" (1981:233f.).
The structure of this chapter is rather straightforward: it consists of four
narrative sections each of which contains a quotation:

(l)(a) The visit of the magi (vv.I-12);quotefromMicah5:1

(b) The escape to Egypt (vv. 13-15); quote from Hoseah 11:1
(c) The slaughter of the children (vv. 16-18); quote from
Jeremiah 31:15
(d) The return to Nazareth (vv. 19-23), with the quote 'He shall
be called a Nazarene' from an unidentified source.

After careful consideration not only of this particular selection ofnarra-

tive material and quotes, but also of certain peculiarities in Matthew's use of
his quotations, France reconstructs the 'message' of this chapter as follows:

Assuming that Matthew was writing for a Jewish Christian reader-

ship ... I imagine that even the most uninstructed reader would have

5 There are a number of reasons for this choice: e.g. this chapter presents special chal-

lenges of interpretation - R. T. France refers to it as "a minefield littered with exegetical

corpses" (1981:233). It is also a chapter that seems to rely quite heavily on contextual
assumptions peculiar to the religion and culture of the Jews; but perhaps most impor-
tantly, an excellent analysis ofthis chapter has been proposed by R.T. France (1981)
who approaches the text clearly in terms of communication.
74 Translation and Relevance

had no difficulty in grasping the apologetic point which governs the

overall strncture of the chapter, that Jesus' obscure Galilean back-
ground was not a cause for embarrassment, but rather the end-result
of a series of divinely directed movements, beginning as orthodox
belief demanded in Bethlehem, but culminating in Nazareth, and that
for each stage of this process there was appropriate scriptural au-
thority. The 'surface meaning' which we have postulated for the four
formula-quotations would plainly convey this general, essentially
apologetic, message. (France 1981:249)

However, according to France, this 'surface meaning' does not exhaust

Matthew's communicative intentions:

we have seen reason at each point to believe that Matthew had more
in mind than the' surface meaning'; that he had bonus points to offer
to those whose acquaintance with the Old Testament enabled them
to spot his 'deliberate mistakes' in Mic. 5:1 and his sophisticated
creation of the Nazarene text from a minor theme of Old Testament
prophecy, or to recollect the context of Jer. 31:15 and the original
identity of the 'son' in Hos. 11:1. (1981:250)

France assumes here a "distinction between the 'surface meaning', which

any reasonably intelligent reader might be expected to grasp, and what we
may call a 'bonus' meaning accessible to those who are more 'sharp-eyed',
or better instructed in Old Testament scripture, ... "(1981:241).6
France's justification of the view that there can be layers of intended
meaning in a text is worth quoting in full:

I am sure that a distinction between surface meaning and a bonus

meaning for the initiated or alert is realistic; any adult reader of chil-
dren's classics ... will be well aware that the surface meaning may
be communicated to the great delight of the more naIve audience,

6 Here is an example of the 'bonus points' offered by Matthew's deliberate mistakes;

where the original passage in Micah 5: I speaks ofthe insignificance of Bethlehem as

'the least' of the towns in Judah, Matthew changes it to its emphatic opposite oudamoos
elachistee 'by no means the least'; in France's view, the effect ofthis alteration is "to
call attention to the dramatic alteration in Bethlehem's role from the insignificant vil-
lage of David to the birthplace of the great son of David" (1981:242). He points out that
a reader not well acquainted with the Old Testament text might miss this 'bonus point'
because this change "achieves its force precisely as an alteration: in itself it would cause
no surprise" (198:242).
Translating the Meaning of the Original 75

while at the same time a whole world of more esoteric pleasure is in

store for those who share the author's private adult viewpoint and
erudition. It is a poor author who aims to communicate only with the
lowest common denominator of his potential readership. (1981:241)

France concludes:

what any given reader will find in a chapter like Matthew 2 will vary
with his exegetical background. What I want to suggest is that Mat-
thew would not necessarily have found this regrettable, that he was
deliberately composing a chapter rich in potential exegetical bo-
nuses, so that the more fully a reader shared the religious traditions
and scriptural erudition of the author, the more he was likely to de-
rive from his reading, while at the same time there was a surface
meaning sufficiently uncomplicated for even the most naIve reader
to follow it. (France 1981:241)

This view of the intended 'message' of a text being layered, perhaps

even open-ended, is quite consistent with relevance theory, and, as France
has pointed out, agrees with our everyday use of language as well. The
question we are interested in here, however, is: what does this mean for ap-
proaches to translation that are committed to conveying the author-intended
meaning of the original to a receptor language audience?
Assuming that France's analysis is essentially right, it would only seem
reasonable to expect that any translation committed to conveying the in-
tended original meaning should convey to the receptor language audience
at least Matthew's 'surface meaning', that is, the main point of this pas-
sage, which he intended to convey to all members of his original audience. 7
However, it seems safe to say that there are few, if any, English readers
who would naturally derive this 'surface meaning' from an English transla-
tion of this chapter - not even from the 'Good News Bible' translation,
though this was produced on the principles of dynamic equivalence. Though
all regular English translations of this chapter include all the events as well
as the Old Testament quotations and allusions, they do not seem to convey
Matthew's main point here. How can this be? To understand this problem
we need to look more closely at the nature of communication.

7 At this point I am not interested in the particular interpretations that France arrives at

in his study; I am using his example more in terms of its general approach and insights,
rather than to endorse any particulars.
76 Translation and Relevance

The Problem of Secondary Communication

Situations

As pointed out in chapter 2, (see pp. 24-5), one of the central claims of
relevance theory is that human communication works by inference: the au-
dience infers from the stimulus what the communicator intends to convey.
Furthermore, in the same chapter (pp. 26-35) we saw that in verbal commu-
nication the derivation of the speaker-intended interpretation depends not
only on correct decoding, but just as much on the use of the right, that is,
speaker-intended, contextual information.
Thus the sentence 'We are about to close', said to you by a shop assist-
ant as you try to enter, would normally be taken to suggest that you should
not come in. However, if that shop assistant were your friend with whom
you had planned to go out for the evening, it would more likely be intended
to suggest to you that you should wait for him since he would be shortly
with you. The meaning available from decoding would be the same in both
instances - the difference in interpretation would be due to the difference in
the contextual information used in the interpretation process.
It follows that for communication to be successful the text or utterance
produced must be inferentially combined with the right, that is, speaker-
envisaged, contextual assumptions. Let us call communication situations
where this condition is fulfilled primary communication situations. How-
ever, it can happen - for various reasons - that in interpreting a text an
audience may fail to use the contextual assumptions intended by the com-
municator and perhaps use others instead. Such situations we shall refer to
as secondary communication situations, and in most cases they will lead to
misinterpretations.
What kind of misinterpretations can arise in secondary communication
situations? From what has just been said, the answer is: any aspect of inter-
pretation that is dependent on context:

Context detennines the disambiguation oflinguistically ambiguous expressions:

wrong contextual assumptions can lead to the choice of the wrong semantic
representation of such expressions.

Context is usually needed to determine the propositional fonn of an utterance:

again mismatches of context can lead to the derivation of a wrong proposi-
tional fonn.

Context is needed to detennine whether a propositional fonn is intended as an

Translating the Meaning of the Original 77

explicature, or whether it serves only to convey implicatures. This plays an

important role in the identification of literal versus loose or metaphorical
talk. Thus the interpretation of the sentence 'I could have killed that fellow'
as either literal or metaphorical will depend on whether its propositional
form is taken as an explicature or not. 8 Since this depends on the contextual
assumptions available, the use of inappropriate context can lead to misun-
derstandings.

Context is needed to derive the implicatures of an utterance. Use of the wrong

context can lead to the derivation of implicatures not intended - or it can
cause intended implicatures to be missed.

This list is suggestive rather than complete - but it is probably suffi-

cient to show that secondary communication situations can give rise to a wide
range of misinterpretations: ambiguities can be resolved the wrong way,
metaphorical expressions can be taken literally, implicatures can be missed
and so forth.
What is important at this point is the reason for these misinterpretations.
In all the cases considered, they arise from a mismatch in context: a given
utterance is interpreted against a context different from the one intended by
the communicator.
There is a wide range of secondary communication situations: they
begin with everyday misunderstandings that occur because the person ad-
dressed, for one reason or another, did not use the contextual assumptions
envisaged by the communicator; they extend all the way to the problem of
misinterpretations of historical documents or works of literature that origi-
nated in settings different from our own.
From all that has been said so far, it comes as no surprise that transla-
tion, too, can find itself in secondary communication situations, and where
it does, it follows naturally that misinterpretations may arise, just as they do
in other situations of secondary communication.
In fact, this is the main reason why an English translation of the second
chapter of Matthew is unlikely to communicate to the average English reader
what was most likely an obvious 'surface meaning' for the original audi-
ence: if the audience for whom Matthew wrote was a Jewish-Christian
audience, they would first of all have had a fair knowledge of the Old Tes-
tament, and secondly they would have had to grapple with the problem that

8 For other suggestions concerning the relevance-theoretic analysis of metaphors, see

e.g. Carston (1996).
78 Translation and Relevance

Jesus had come to be known as 'Jesus of Nazareth': how could a 'Jesus of

Nazareth' be the expected Messiah when it was common knowledge that
the Messiah was to be born in Bethlehem? That this question was in the air
is evidenced, for example, in John 7:4lfwhere the people questioned Je-
sus' messiahship on those very grounds, "Is the Christ to come from Galilee?
Has not the scripture said that the Christ is descended from David, and comes
from Bethlehem, the village where David was?"
To an audience vexed by this problem it must have been easy to infer
what Matthew intended to convey by this section of the text: it shows clearly
that Jesus was, in fact, born in Bethlehem and also that, as France pointed
out, the fact that he came to live in Nazareth later on was not a mistake, but
the result of a succession of divinely guided events. In other words: for the
original audience Matthew's main point here followed as a set of implicatures
from the combination of this text with some highly accessible contextual
assumptions.
By contrast, it is not surprising that contemporary English readers who
may have little knowledge of the Old Testament and who would see no
problem in the association of Jesus with Nazareth, would fail to get Mat-
thew's main point here: lacking the right contextual assumptions they would
fail to derive the right implicatures.
Perhaps it is worth pointing out that this problem is not peculiar to the
translation of written documents from ancient times, as the following ex-
ample given by Brislin (1978) illustrates:

At a meeting held recently in Japan, an American was discussing

two alternative proposals with his colleagues, all of whom were na-
tive speakers of Japanese. The American was well schooled in the
Japanese language and was, indeed, often called 'fluent' by those
around him. At this meeting, proposal A was contrasted to proposal
B, and a consensus was reached about future action, and the meeting
then dismissed. Upon leaving the room the American commented, 'I
think the group made a wise choice in accepting proposal A.' A Japa-
nese colleague, however, noted, 'But proposal B was the group's
choice.' The American continued: 'But I heard people say that pro-
posal A was better.' The Japanese colleague concluded, 'Ah, you
listened to the words but notto the pauses between the words.' (1978 :205)

In this instance communication was not only between contemporaries,

but even face to face, and yet contextual mismatches due to cross-cultural
Translating the Meaning of the Original 79

differences caused misunderstanding, leading the American to derive wrong

implicatures.
Having understood the nature of the problem, the next question for the
translation theorist is what should or could be done about such losses in
translation.

Secondary communication problems and 'dynamic equivalence'

In the dynamic equivalence approach, there is a certain amount of ambiva-

lence as to how problems arising from secondary communication should be
handled. Let us approach the matter first from the perspective of the overall
objectives of dynamic equivalence. As we have seen, this approach aims at
the comprehension of the 'message' of the original, and this 'message' is
defined as "the total meaning or content of a discourse; the concepts and
feelings which the author intends the reader to understand and perceive"
(Nida and Taber 1969:205). Given this, it certainly seems necessary for the
translator to seek to overcome obstacles to comprehension arising from the
differences in background knowledge between the old and the new audi-
ence, and one would expect the theory of dynamic equivalence to provide
and spell out the measures needed to achieve this.
In terms of our example, one would expect the dynamic equivalence
approach to provide the translator with ways and means of communicating
those main points of Matthew 2 to the modem English reader. Since the
problem was that modem readers lack the contextual assumptions neces-
sary to arrive at those main points - which are, in tum, mostly contextual
implications - in principle the problem could be solved in two different
ways: a) by supplying the contextual assumptions needed or b) perhaps by
spelling out the contextual implications themselves. Thus the translator
should be able to include in the translation either information about the
problem of Jesus' association with Nazareth in the light of Jewish expecta-
tions about the Messiah, or he should be able to spell out the main point,
perhaps in the following form at the end of the chapter: "Thus we see that
the fact that Jesus is called 'Jesus of Nazareth' is no reason to be embar-
rassed. Rather, as we have seen, God brought him there in a number of
steps, each of which he himself directed, beginning, as our belief demands,
in Bethlehem, and culminating in Nazareth".
However, though quite clearly demanded by the objective of dynamic
equivalence, these measures would not be acceptable on Nida and Taber's
conception of translation which is a linguistic one:
80 Translation and Relevance

linguistic translation: a translation in which only infonnation which

is linguistically implicit in the original is made explicit and in which
all changes offonn follow the rules of back transfonnation and trans-
fonnation and of componential analysis; opposed to CULTURAL
TRANSLATION. Only a linguistic translation can be considered
FAITHFUL. (1969:205, capital letters as in original, italics my own)

Now contextual assumptions and implications are not a matter of lin-

guistics, but of inferences that have to do with people's beliefs - cultural,
religious and so forth; consequently their explication is not warranted un-
der linguistic translation, as the authors clearly assert in the following
passage:

It must be further emphasized that one is not free to make in the text
any and all kinds of explanatory additions and/or expansions. There
is a very definite limit as to what is proper translation in this difficult
area: one may make explicit in the text only what is linguistically
implicit in the immediate context of the problematic passage. This
imposes a dual constraint: one may not simply add interesting cul-
tural infonnation which is not actually present in the meanings of the
tenns used in the passage, and one may not add infonnation derived
from other parts of the Bible, much less from extra-Biblical sources,
such as tradition. (Nida and Taber 1969: Ill, italics as in original)9

Both from the point of view of relevance theory and from common ex-
perience, it is difficult to see how such a notion of 'linguistic translation'
can as a matter of general principle aim at achieving dynamic equivalence
in terms of conveying the message of the original, especially in view of the
fact that the translation situations that Nida and Taber have in mind span
wide cultural gaps. to
The reconciliation of the goal of dynamic equivalence with that of lin-
guistic translation is even more difficult when it comes to the dynamics
themselves. Let me illustrate this with the opening chapter of Matthew's
Gospel. Matthew begins his Gospel with the genealogy of Jesus. In terms

9 The reference to "information which is not actually present in the meanigs of the terms

used in the passage" is based on a decompositional view of semantics where the mean-
ing of a term is made up of its meaning components.
10 Nida and Taber do allow for the inclusion of explanatory notes, either in a glossary or

on the page where the note is needed, but separate from the translated text (1969: Ill).
However, because of the extra-linguistic source of such information, under their defini-
tion of 'linguistic translation' these measures must be considered supplementary to
translation, not part of translation itself.
Translating the Meaning of the Original 81

of his original intentions and his original audience this was no doubt very
effective: assuming that one of the main objectives of writing this Gospel
was to assure people that Jesus was the long-expected Messiah, one of the
first requirements that the candidate would have to fulfil was to be of the
right lineage: it was common knowledge that the Messiah would be a de-
scendant of King David. The identity ofthe Messiah naturally was a burning
issue in those days of oppression by the Roman government. In other words,
as far as his contemporary audience was concerned, on the very first pages
of his Gospel Matthew began to tackle a crucial point.
But what about our typical modem English reader? The fact that he was
reading the Gospel of Matthew would indicate some interest in its contents,
presumably because of Jesus, the central character of the book. Depending
on what kind of religious education he may have had, he would perhaps
have some biblical knowledge: let us say he has heard of outstanding char-
acters like King David, or Abraham. But the majority ofthe other individuals
listed in the genealogy would very likely be unknown to him, and their
names, take, for example, 'Jehoshaphat', hard to read. Thus our reader would
have to struggle through a long list of mostly unknown and difficult names -
in terms of relevance theory, he would have to spend a lot of processing
effort on the first sixteen verses of this chapter.
Despite this high processing effort, he would find little reward, in sharp
contrast to Matthew's first century audience: our modem reader would
almost certainly not be aware that Davidic lineage was important as a pre-
requisite for Jesus to qualify as the Messiah - and even ifhe were aware of
this, the question whether or not Jesus is that Messiah-figure expected by
the Jews would probably not figure very largely in his mind. Nor would the
relevance be likely to be increased by Matthew's explicit comment after the
genealogy about the lineage consisting of three sets of fourteen ancestors.
In short, reading this text, our reader would tend to have great difficulty
in arriving at an interpretation consistent with the principle of relevance.
This contrasts sharply with the original communication act; there the back-
ground knowledge envisaged by Matthew would first of all have decreased
the processing effort: many of the names in the genealogy would have been
known to Matthew's Jewish-Christian readers; secondly, and more im-
portantly, the text would have had a rich pay-off in terms of contextual
implications: the strong messianic expectations of the day and the readers'
familiarity with (most of) the characters occurring in the genealogy would
have enabled him to see many interesting implications in it; thus the men-
tion of women in a genealogy, like Rahab, or the allusion to one of the dark
82 Translation and Relevance

spots in king David's life, addressed by the explicit statement that Solo-
mon's mother had been someone else's wife, would all have provided food
for thought.
In view of these discrepancies between the two audiences it is difficult
to see how one can seriously uphold the idea that dynamic equivalence trans-
lation can achieve if not identity, at least "a high degree of equivalence of
response" (Nida and Taber 1969:24) as a general claim: it may be achiev-
able in primary or near-primary communication situations, but it seems
unrealistic for secondary communication situations with significant differ-
ences in cognitive environment, such as are usually encountered when
translating biblical texts for present day readers.ll
At the same time, it must be recognized that dynamic equivalence trans-
lations do tend to be more easily understood than formal equivalence
translations. One reason for this is that the orientation toward the receptor's
response helps the translator to avoid awkwardness in expression that creeps
in easily due to source language interference. However, another reason seems
to be that the dynamic equivalence approach does, in fact, allow for a number
of context-conditioned adaptations, though these are presented as linguis-
tic changes.
One area in which this happens is that of figurative language. In the
framework of componential analysis adopted by Nida and Taber, "figura-
tive extensions are based upon some supplementary component in the
primary meaning which becomes essential in the extended meaning"
(1969:88). Thus in the expression 'He is a fox', the figurative meaning is "me-
diated through a supplementary - and purely conventional - component
which claims that the fox is particularly deceptive and clever" (1969:87).
Since such figurative extensions are "often arbitrary and conventional,
they are almost always specific to a particular culture and language" (1969:88),
and, accordingly, they can give rise to communication problems in translation.
To avoid such problems, the translator may need to use a different figura-
tive expression. Thus if in another culture rabbits or spiders are considered

11 Another problem in the area of Bible translation in particular is that the translator,

and also the translation theorist himself, belong to a culture and time usually very dif-
ferent from that of the original. This makes it very difficult for him to judge realistically
what the dynamics of the original were like for the original audience. (See House 1981:
8ff., for a criticism ofthe 'dynamic equivalence' approach on similar grounds.) In par-
ticular, it gives rise to the danger that - perhaps subconsciously - we judge the response
of the receptor language audience against our own response, rather than against a care-
ful reconstruction of what the original response might have been.
Translating the Meaning of the Original 83

'deceptively clever', then the translator would need to substitute accord-

ingly. The legitimacy of such a substitution would rest on the assumption
that this is a purely linguistic change: the expression in the translation is
chosen on the basis of the meaning component it shares with that found in
the figurative extension in the original.
However, as Sperber and Wilson have argued in a number of places
(e.g. 1986a, 1986b), there is good reason to believe that such 'extended
meanings' arise not from linguistic meanings or components of meaning,
but rather from information stored in the encyclopaedic entry associated,
for example, with the concept 'fox'.12 As such it is not linguistic or seman-
tic, but contextual information: it is not the case that the English word fox
has a linguistic meaning component' deceptively clever', but rather it is a
popular belief among English speakers that foxes have this particular char-
acteristic. When the expression 'He is a fox' is used in any given instance,
the audience may make use of this belief, if it leads to an interpretation
consistent with the principle of relevance.
If this view is correct, then such adaptations should not be acceptable in
a theory of linguistic translation, and this would make the goal of dynamic
equivalence still more difficult to achieve.

Secondary communication problems and idiomatic translation

Of the various problems that arise from secondary communication, in the

idiomatic approaches special attention is paid to the fact that not all the
meaning of the text is expressed:

In every text that one may want to translate, there will be informa-
tion which is implicit; that is, it is not stated in an explicit form in the
text itself. Some information, or meaning, is left implicit because of
the structure of the source language; some because it has already
been included elsewhere in the text, and some because of shared in-
formation in the communication situation. However, the implicit
information is part of the meaning which is to be communicated by
the translation, because it is part of the meaning intended to be un-
derstood by the original writer. (Beekman and Callow 1974:38, italics
as in original)

Now on the grounds that such information is already part of the original

12 For more detail on encyclopaedic entries and the organization of information associ-

ated with concepts see chapter 6, pp. 141-2.

84 Translation and Relevance

message, the idiomatic approaches not only allow, but call for the explica-
tion of such information in the translated text, if it cannot be conveyed
implicitly in a given instance:

It is clear, therefore, that the translator needs to take into careful

consideration the presence of implicit information in the original, so
that it may be used explicitly when it is needed in the RL [receptor
language] version. (Beekman and Callow 1974:47).
[Implicit information] will sometimes need to be made explicit
because the source language writer and his audience shared infor-
mation which is not shared by the receptor language audience.
(Larson 1984:42, italics as in original)

One of the examples given by Beekman and Callow (1974) concerns

Mark 1:35ff. which the King James Version, following the Greek rather
closely, translates as follows:

35 And in the morning, rising up a great while before day, he went

out and departed into a solitary place, and there prayed. 36 And Simon
and they that were with him followed after him. 37 And when they
had found him, they said unto him, All men seek for thee (italics as
in the original).

On the grounds that when dealing with a series of events, different lan-
guages tend to differ in which events they make explicit or leave implicit,
Beekman and Callow suggest that in some language(s) verse 36 of the sec-
tion cited above may need to be rendered as follows, the italics indicating
the 'explicated' information:

And when it dawned Simon and they that were with him in the house
arose and saw that Jesus was not there. They went out and followed
after him. (1974:54, italics as in the original)

Unlike Nida and Taber (1969), neither Beekman and Callow (1974) nor
Larson (1984) commit themselves to a notion of 'linguistic translation'. In
fact, Beekman and Callow state that "Occasionally ... the translator needs
to draw on information available in the remote and cultural contexts as well
as the information he can find in the immediate context" (1974:57).
On the other hand, the idiomatic translator is not free to explicate just
any information; the general rule is that implicit information "is made ex-
Translating the Meaning of the Original 85

plicit because the grammar, or the meanings, or the dynamics of the RL

[receptor language] require it in order that the information conveyed will be
the same as that conveyed to the original readers" (Beekman and Callow
1974:58).
On the face of it, it might seem that with these guidelines - those given
by Larson (1984) correspond closely to them - the idiomatic approaches
provide all that is theoretically necessary for the translator to achieve his
aim of conveying the same message as the original: he is led to explicate all
and only the information needed to that end.
However, there are significant problems here because the concepts of
meaning and communication on which the idiomatic approaches rest are
inadequate in some important respects. To see this, let me begin with a brief
outline of how these approaches view the meaning of texts.
According to the idiomatic approaches, a text has 'surface structures',
which are grammatical, lexical and phonological structures, and it also has
a 'deep structure' which consists of propositions and other elements that
give, for example, indications about speech acts and about interpropositional
relations. The 'surface structures' constitute the 'form' of the text, and the
'deep structure' its meaning, which is what the translator is to convey:

Behind the surface structure is the deep structure, the meaning. It is

this meaning that serves as the base for translation into another lan-
guage. (Larson 1984:26, italics as in original) 13

However, the relationship between 'surface' and 'deep structure' is not as

straightforward as one might have thought, and it is largely these complica-
tions that make translation so difficult. One problem is that typically language
allows for 'skewing' between 'surface' and 'deep structure': for example,
there can be 'skewing' between the grammatical form of a language and its
illocutionary force, a typical case in point being 'rhetorical questions' that
have the grammatical form of questions, but can have rather different
illocutionary force. Thus a mother who is angry with her son for not having
emptied the garbage might say, "When are you going to empty the gar-
bage?". In this case, "the semantic illocutionary force is one of command,

13 Most of the quotations are taken from Larson (1984) since she presents a more ex-

plicit account of meaning than Beekman and Callow (1974). However, the assumptions
about meaning in both works seem to be the same in all aspects relevant to our discus-
sion here. K. Callow is working on a comprehensive theory of communication (Callow
1998).
86 Translation and Relevance

but the grammatical form is that of a question which would normally be

used to ask about time" (Larson 1984:235, italics as in original). In many
instances such 'skewing' will be used to convey 'emotive meaning'.
Another problem with language is that of implicit information, which
we already mentioned above. As "that [information] for which there is no
form but the information is part of the total communication intended or
assumed by the writer" (Larson 1984:38), it also arises from the difference
between 'surface' and 'deep' structure.
Before examining these views, it is only fair to acknowledge that Larson
points out with regard to her assumptions about semantics that the main
objective of her book is not in the area of theory but of practice: "The aim
of the book is not to argue linguistic theory but to present tools which will
help translators" (1984:26). However, as we shall see, theoretical assump-
tions are important in that the value of the tools suggested depends on the
validity of the framework in which they are developed.
Perhaps the best point from which to examine the notion of meaning
used in the idiomatic approaches is that of implicit information, because
here the problems can be seen most clearly. As we just saw, implicit infor-
mation is defined as that "for which there is no form", but which is
nevertheless "part of the total communication intended or assumed by the
writer" (Larson 1984:38). Since the approach treats language as a form-
meaning correlate, this raises the first important question: if implicit meaning
is not in the 'form' of the text - how is it conveyed? It is obviously as-
sumed to be in the 'deep structure', but how does it get there? Larson
suggests:

Some infonnation, or meaning, is left implicit because of the struc-

ture of the source language; some because it has already been included
elsewhere in the text, and some because of shared infonnation in the
communication situation. (1984:38, italics as in original)

This gives some indication as to the source of implicit information, but

it still does not explain how a particular piece of shared information in the
communication situation, rather than any other piece of information equally
shared, comes to be included as part of the intended meaning of a text.
Consider the following example given by Larson:

Infonnation which is left implicit when talking to one person might

be made explicit when talking to another. A woman might say to her
Translating the Meaning of the Original 87

husband, 'Peter is sick. ' In reporting the same information to the

doctor she would say, 'My son Peter is sick, ' or 'My son is sick. '
The information my son was not needed to identify Peter when talk-
ing to her husband who knew very well who Peter was. (1984:41,
italics as in original)

Larson then goes on to explain:

There is a difference between implicit information and information

which is simply absent and never intended to be part of the commu-
nication. For instance, in the example 'My son Peter is sick, ' the
mother did not say, 'Peter has brown hair and is ten years old. 'This
is not implied. It is absent. It is not part of the communication and,
therefore, should not be added. (1984:42, italics as in original)

Thus the crucial point that distinguishes implicit information from 'infor-
mation which is simply absent' is the communicator's intention: only
unexpressed information which the communicator intended to convey quali-
fies as implicit information.
This view is certainly right from the communicator's point of view.
However, what it does not explain is how the audience can possibly tell
the two kinds of information apart, given that it has no direct access to the
communicator's intention but, in fact, has to discover that intention from
what he says.
This gap in the approach is significant from two perspectives: from the
practical point of view it leaves the translator without needed guidance when
it comes to identifying what information is or is not implied in a text; this is
one ofthe reasons why among translators following the idiomatic approaches
the matter of implicit information has been a perennial topic of debate. From
the theoretical perspective the lack of an explicit account of the nature of
implicit information has given rise to a number of misconceptions about
implicit information. As we shall see, these misconceptions have in tum
given rise to practices that seem questionable.
One of these misconceptions concerns implicit information involved in
the metaphorical uses oflanguage. According to Larson (1984), "Metaphors
and similes are grammatical forms which represent two propositions in the
semantic structure" that are related to each other by way of comparison,
where "the comparison is always that of some likeness" (1984:246, italics
as in original). Often that likeness or 'point of similarity' is left implicit:
88 Translation and Relevance

In the sentence The moon is like blood, the two propositions are:

2 The moon is (red).

2 Blood is (red). (1984:247)

The 'point of similarity' is not always as obvious as in this example.

Thus another example given by Larson is the following:

(2) John eats like a pig. (1984:248)

She comments:

The point ofsimilarity is not given. Maybe the proposition is the pig
eats too much, or the pig eats fast, or the pig eats sloppily. Until we
can fill in the comment about the pig, we do not know the point of
similarity to John. (1984:248, italics as in original)

How can the translator find out such implicit information? Larson sug-
gests: "Often the context in which a metaphor is used will give clues which
will help in the interpretation" (1984:249).
Apart from the lack of explicitness already noted, this view of similes
and metaphors is quite mistaken in the two related assumptions that these
figurative expressions always represent two propositions, and that there
is always one 'point of similarity'. Sperber and Wilson (1986a; 1986b) not
only offer an explicit account of how the implicatures of figurative lan-
guage are conveyed, but have also shown that the very point of figurative
uses of language is that they convey a wider range of propositions, even in
the case of highly standardized metaphors. Thus Sperber and Wilson dis-
cuss the following example:

(3) This room is a pigsty. (1986a:236)

Having pointed out that, perhaps due to some stereotyped assumptions, such
standardized metaphors typically convey "one or two dominant and highly
accessible assumptions", in this case perhaps "the implication that the room
is filthy and untidy" (l986a:236), the authors add:

However, the speaker must have intended to convey something more

than this if the relative indirectness of the utterance is to be justified:
an image, say, of filthiness and untidiness beyond the norm, beyond
Translating the Meaning of the Original 89

what could have been satisfactorily conveyed by saying merely 'This

room is filthy and untidy.' Thus even this highly standardized example
cannot be paraphrased without loss. (1986a:236)

Many metaphors convey much more than this:

In general, the wider the range of potential implicatures and the greater
the hearer's responsibility for constructing them, the more poetic the
effect, the more creative the metaphor. (1986a:236)

Going back to Larson's example 'John eats like a pig', the expectation
that there is one 'point of similarity' which the translator has to identify in
order to understand the simile is quite mistaken: in order for the figurative
use to be justified the audience would, in fact, invariably look for an inter-
pretation that implies more than that. Beekman and Callow (1974) propose
the same scheme of analysis for metaphorical language. We see here, then,
that one of the consequences of the failure to come to grips with implicit
information in the idiomatic approaches is that it is prone to mislead the
translator concerning the meaning of metaphorical expressions. 14
However, these problems with implicit information in metaphorical lan-
guage reveal another, more general misconception of implicit information,
and that is the assumption that it is determinate. Thus, as we saw (p. 87
above), the translator is admonished to distinguish clearly between infor-
mation that is 'implicit' and other information that is 'absent' - which would
seem to presuppose a clear distinction between the two. Yet, as Sperber and
Wilson (1986a) have shown, there is good reason to believe that this is not
the case:

14 It is interesting to note that despite their theoretical inadequacy, the idiomatic ap-

proaches do show sensitivity to the fact that there may be more to metaphorical
expressions than their analysis suggests. This may be seen in their reluctance to replace
metaphorical expressions altogether by their literal equivalents except as a last resort:
"the first approach is to retain the form of a metaphor ... If this is inadequate, then the
form of a simile is tried. If this still fails to communicate, then a nonfigurative form is
used" (Beekman and Callow 1974:145). However, they themselves view this as a con-
sequence of concern for 'form' - which is somewhat surprising in an otherwise
meaning-oriented approach: "From the general theoretical standpoint, this discussion is
a further illustration ofthe principle that while in an idiomatic translation meaning al-
ways takes precedence over form, this does not mean that the form of the original is
completely ignored. There are circumstances, as in the present discussion on translating
metaphor and simile, when the special literary form ofthe original is taken into consid-
eration when deciding on the RL form". (1974,:145, f. 3)
90 Translation and Relevance

The fiction that there is a clear-cut distinction between wholly deter-

minate, specifically intended inferences and indeterminate, wholly
unintended inferences cannot be maintained. (1986a: 199)

Rather relevance theory has shown that implicatures vary along a continuum
of relative strength, the implicature being the stronger the more necessary it
is seen to be for consistency with the principle of relevance.
Returning to example (2) above, one can think of circumstances where
one of the three propositions considered would be very strongly communi-
cated; thus if John had just left the table complaining that his stomach was
too full, the remark 'John eats like a pig' would be seen to strongly impli-
cate the contextual assumption 'the pig eats too much' rather than 'the pig
eats sloppily'.
However, suppose that John had misbehaved at table in all sorts of
respects - taking huge portions of food, eating it very hastily and noisily,
and spilling it all around his plate. Assuming that anyone or perhaps two
of those propositions could lead to an interpretation consistent with the
principle of relevance, none of the three propositions considered would
be particularly strongly implicated, but would come across as a weaker
implicature
France's analysis of Matthew 2 shows an awareness of this indetermi-
nacy and open-endedness of implicatures, and he saw their communicative
value when he suggested that "Matthew would not necessarily have found
this regrettable", but rather that "he was deliberately composing a chapter
rich in potential exegetical bonuses, so that the more fully a reader shared
the religious traditions and scriptural erudition of the author, the more he
was likely to derive from his reading ... " (France 1981:250; quoted here
again for convenience). As we shall see in chapter 5, weak communication
is an important vehicle for the achievement of poetic effects.
Relevance theory has no problem in accounting for such indeterminacy,
but recognizes it clearly as a regular part of human communication. How-
ever, it does pose considerable problems for the idiomatic approaches, with
their commitment that the translator should convey both the explicit and
implicit information of the original. One problem is how the translator can
know whether he has fulfilled his task of conveying the same 'message' in
cases involving indeterminate sets of implicatures. The assumption of the
idiomatic approaches that, for example, metaphors represent two proposi-
tions with one point of similarity has been helpful to the translator since it is
fairly easy to compare original and translation within this framework.
Translating the Meaning of the Original 91

However, it is not at all clear how such sameness of meaning is to be evalu-

ated when one is dealing with open-ended and non-enumerable sets of
implicatures of varying strength.
Another range of problems arising from these misconceptions in the idi-
omatic approaches has to do with the general solution proposed to remedy
communication problems in the area of implicit information, namely the
'explication' of this information in the translated textY
One problem is that there has been little recognition of the true extent to
which implicatures are involved in communication, and hence of the true
degree to which explication might be needed. Let us begin with as simple
an example as (2) above (p. 88). In a case where no equivalent metaphor or
simile can be found to convey such implicit information, should the transla-
tor list all three propositions explicitly, that is render this simile as 'John
eats like a pig, too much, too fast and too sloppily'? But this might involve
some arbitrariness - perhaps the similarity extends also to the noises pro-
duced by pigs when eating? Should this, therefore, be listed as well? What
if still other similarities come to mind? Where does one stop listing? Note
that there need not be any clear cut-off point, in which case no non-arbitrary
decision would be possible.
Furthermore, explication would limit the interpretation to just these three
points; having been presented with the utterance' John eats like a pig, too
much, too fast and too sloppily', the audience would have no reason to as-
sume that it was also meant to recover, for example, the idea that John ate
noisily, though the original would have allowed for this as a further weak
implicature. 16

15 In fact, two remedies are provided: 'explication' as a general measure, and 'change of

form' with regard to figurative expressions. Thus the 'metaphorical form' of a source
language expression may be changed to a 'nonfigurative form'; for example, Beekman
and Callow suggest that the metaphorical "I will make you to become fishers of men"
(Mark 1: 17) "may be cast in a nonfigurative form. It would then read, 'You have been
working catching fish, now I will give you a new work making disciples for me'"
(1974:148). However, it is clear from the point of view of relevance theory that the
change made here is not one of form, but of meaning; in fact, the new rendering would
seem to have lost virtually all the implicatures that arise from the presentation of the
disciples as 'fishers of men'.
16 Note also that the relative strength of implicatures correlates with the degree of re-

sponsibility taken for their assertion by the communicator. This makes weak
communication a very useful tool for hinting at something without assuming full re-
sponsibility for it: the weaker an implicature is, the less responsible the communicator
can be held for it. Explication will often change this sharing of responsibility. We shall
take up the significance ofthis point for translation in ch. 7, pp. 175-6.
92 Translation and Relevance

The extent of this problem of indeterminacy and open-endedness of

implicature can be seen more clearly when one considers more involved
uses of metaphorical language, as, for example, the following passage from
Matthew 2:

Then was fulfilled what was spoken by the prophet Jeremiah:

A voice was heard in Ramah,
wailing and loud lamentation,
Rachel weeping for her children;
she refused to be consoled,
because they were no more.
(Matthew 2: 17f., Revised Standard Version)

Here we are dealing with a rather elaborate metaphor that is very rich in
a comparatively large number of weak implicatures, which together create
an impression rather than convey a clearly specifiable message. Given that
this metaphor relies heavily on knowledge of the Old Testament and of
Palestinian geography, one would have to add a great deal of information to
convey the intended interpretation to readers who lacked knowledge of both,
as would be the case, say, with the average person among the Silt' e-people
of Ethiopia. 17 Also, to understand this passage fully, the reader would need
to be acquainted with the generally accepted ways of using Old Testament
texts as predictions of contemporary events.
Furthermore, in the literature on translation, attention is mostly paid to
implicit information with regard to the meaning of sentences, possibly of
paragraphs. IS Implicatures at the level of a chapter or larger unit of text, as
noted in France's analysis of Matthew 2, are not generally addressed. If
they were, for many situations involving secondary communication, the
amount of information needing explication in order to convey the same 'mes-
sage' would be of considerable size. Thus earlier on in this chapter (p. 79
above) we noted that the following explication might be needed to clearly
convey the 'surface meaning' of Matthew chapter 2 to contemporary Eng-
lish readers:

17 The Silt'e-people number around 200,000 and speak an Ethio-Semitic language. My

family and I lived among the Silt'e people from 1976-1979, studying their language.
18This tendency is reflected in the claim by Beekman and Callow that most of the im-
plicit information that needs explication is found in the immediate context: "Most of
the implicit information that is relevant to understanding the document is contained
within the document itself, and it is only rarely necessary for the translator to draw on
information from outside of it. In fact, most of the relevant implicit information within
the document is drawn from the immediate context, that is to say, from within the par-
ticular paragraph being studied or from an adjacent one". (1974:49)
Translating the Meaning of the Original 93

Thus we see that the fact that Jesus is called 'Jesus of Nazareth' is
no reason to be embarrassed. Rather, as we have seen, God brought
him there in a number of steps, each of which he himself directed,
beginning, as our belief demands, in Bethlehem, and culminating in
Nazareth.

There would be further points in this one chapter that would require the
explication of contextual assumptions and/or implications if the translator
was really fully committed to the task of communicating the originally in-
tended interpretation, leading to quite substantial additions to the text.
If such explication were done, it would almost certainly run into two
kinds of practical problems. One would be how, under these circum-
stances, the translator could comply with another requirement of the
idiomatic approaches made specifically with regard to the explication of
implicit information:

A further note of caution is needed lest this new information distort

the theme or focus intended in the original document. It is very im-
portant that the translator who is introducing information new to the
receptor audience, but not new to the original audience, does not use
forms that will make it seem like mainline or thematic material. The
discourse could become distorted, or too much emphasis given to
something which is not that important to the original author. Such
information should be presented in such a subtle or natural way that
the intended prominence of the source text is not distorted. (Larson
1984:456)

It is difficult to see how this demand could reasonably be fulfilled with

regard to the example of Matthew 2 as discussed, and there are numerous
other biblical texts where implicit information causes problems of similar
magnitude in secondary communication situations.
The other practical problem has to do with acceptability: it seems more
than likely that such a degree of explication would be found unacceptable
by many audiences, especially when applied to biblical texts. One might be
tempted to regard this matter of acceptability as not too serious: after all,
with enough re-education, perhaps people could become reconciled to this
new way of translating. However, more is at stake here than popular views
of translation, and we shall return to this matter in chapter 7 (see pp. 179ff.
where we discuss the issue of successful communication.
So far our discussion of implicit information has focussed mainly on
94 Translation and Relevance

information that might be missed due to lack of the right contextual as-
sumptions on the part ofthe receptor language audience. However, problems
of implicit information also arise when the audience uses wrong contextual
information, so that it arrives at wrong implications, as the following exam-
ple from Beekman and Callow shows:

Mark 2:4 says, "And when they could not come nigh unto him for the
press, they uncovered the roof where he was". Since no indication
was given of how four men, carrying a paralysed friend, could get
onto a roof (and the language helper tended, naturally enough, to
think in terms of his own familiar steep thatched root), the language
helper assumed a miracle, similar to Philip's sudden removal from
the presence of the Ethiopian official to Azotus. Here, the Greek
narrative left an intervening event implicit - that they climbed the
outside stairs onto the roof. It is not always possible to leave this
implicit in other languages. (1974:47)

I have proposed an analysis from the point of view of relevance theory

elsewhere (Gutt 1987). What I want to point out here is that there is little
reason to believe that wrong implicatures can generally be remedied by
explication. In his chapter 'Ruth in Central Africa: a cultural commen-
tary', Wendland (1987) provides examples of misinterpretations of the book
of Ruth that are likely to arise in a Central African context. One of these
instances refers to the timing of Ruth's return to her home town:

So Naomi returned from Moab ... arriving in Bethlehem as the bar-

ley harvest was beginning. (Ruth 1:22, New International Version)

Wendland comments:

The time reference here is important, since in a Tonga sociocultural

setting it would immediately arouse the suspicions of the people
whose village Naomi was entering. A person does not usually move
during the period extending from after the fields have been planted
until after the harvest has been completed. One's crops mean life,
and therefore it must have been some serious offence which drove
Naomi away from her former home at such a time. Perhaps it had
been that she was guilty of practising witchcraft - after all, were
not all her men now dead? (1987: 171)

Wendland does not offer a solution to this problem, and it is difficult to

Translating the Meaning of the Original 95

see how a translator could effectively prevent the receptor language audi-
ence from bringing all their particular cultural assumptions to bear on the
interpretation of this passage, especially when a number of points in the
story seem to corroborate the misinterpretation. Note that the problem here
is not one of what is said or how it is said - the problem is that the events
reported in the story readily combine with a number of highly accessible
contextual assumptions that result in a highly plausible, though mistaken,
interpretation for the receptor language audience.
Lastly, let us tum to the claim that idiomatic translations should resem-
ble the original in its 'dynamics'. Beekman and Callow (1974) write:

The naturalness of the translation and the ease with which it is un-
derstood should be comparable to the naturalness of the original and
to the ease with which the recipients of the original documents un-
derstood them. (1974:34)

They point out that some allowance needs to be made for problems caused
by differences in language and culture:

Such a comparison of the dynamics of the original with that of a

translation must bear in mind that the message may have been easier
for the original recipients to understand because Greek was the lan-
guage of both writers and readers, and they shared the same or similar
cultures. (Beekman and Callow 1974:34)

Yet they do not consider these differences as serious obstacles that might
invalidate the demand for naturalness and ease of comprehension of the
translated text because the writers "wrote to be understood":

... On the other hand, the message was not dependent upon these
local advantages since the writers were not penning abstract theses
or obscure philosophies but had a very practical aim in view; they
wrote to be understood. (Beekman and Callow 1974:34)

This statement reveals a significant lack of appreciation of the crucial

role that context plays in communication: it seems to strongly imply that
there is a way of 'writing to be understood' that is independent of differ-
ences in contextual assumptions, such as arise from historical, cultural and
other differences.
In a similar vein, Larson (1984) claims:
96 Translation and Relevance

A natural and clear translation is not dependent on familiar informa-

tion. New information, even historical facts, can be presented in a
natural and clear manner. (1984:430)

We have already looked at the problem of supplying all the needed 'new
information' in a translation. Yet there is another factor that is crucial for
communication, especially for 'ease of comprehension', that is not addressed
by these authors at all: and this is that for successful communication the
intended interpretation must not only be recoverable with ease - but also
that it must lead to adequate contextual effects. As we tried to show from
the genealogy at the beginning of the Gospel of Matthew (p. 81 above), it is
doubtful that a modem English reader with little theological or historical
interest would find that section very rewarding in terms of contextual ef-
fects. But without adequate contextual effects the criterion of consistency
with the principle of relevance would not be satisfied; when the hearer is
unable to find an interpretation consistent with the principle of relevance,
he will be uncertain about what the communicator is trying to say. More
seriously, the lack of adequate contextual effects may give him the impres-
sion that the text is irrelevant to him, and a natural reaction to irrelevance is
termination of the communication process: in other words, the receptor puts
the translation aside. There is probably no greater threat to a translation
approach committed to communication than such a complete breakdown.
We may know this reaction to perceived irrelevance from our own expe-
rience. In the case of Bible translation the research of Dye (1980) lends
further support to the seriousness of lack of perceived relevance. Dye in-
vestigated the impact of fifteen Bible translation projects in various parts of
the world, and found that among all the various parameters studied, the
single most important one was what Dye - quite independently of relevance
theory - called the 'principle of personal relevance', that is, the degree to
which the receptor language audience was able to see the relevance of the
translated texts to their lives.
It may be worthwhile pointing out that relevance theory allows us here
to draw a significant distinction with regard to relevance. Suppose a begin-
ner with computers has a problem with the screen display; he has a handbook
- but only one that is actually written for advanced users. In this situation
the handbook, with the information it contains, is highly relevant to that
beginner - it contains the solution to his problem. However, in spite of this,
it might be of little use to the person concerned because he might well find
the processing effort too high, requiring him, perhaps, to study other in-
Translating the Meaning of the Original 97

troductory literature on computers first. This difference can be accounted

for in our relevance-theoretic framework: the handbook is not irrelevant -
it does have contextual implications for the reader; however, the fact that
the derivation of such contextual implications would require great process-
ing effort makes the use of this handbook inconsistent with the principle of
relevance for this audience. 19 Similarly for the examples taken from Mat-
thew: to say that some of these passages are not interpretable in consistency
with the principle of relevance by a certain audience is not to say that they
are irrelevant to that audience.
As pointed out above, the idiomatic approaches do not address these
problems, and so it is not clear how they would want to overcome them.
In conclusion, we see here that, while the idiomatic approaches recog-
nize the importance of implicit information for translation in secondary
communication situations, and allow for its explication, the solutions of-
fered are impaired by an inadequate understanding of communication in
general and of implicature in particular. They give rise to theory-internal
inconsistencies, and there is reason to doubt that the treatment they suggest
can achieve at least in principle, if not in practice, the aim of idiomatic
translations that convey the 'message' of the original, consisting of both its
explicit and implicit information.
At this point someone might raise the following objection: granted that
secondary communication situations can cause problems for translation -
have we not perhaps exaggerated the seriousness of these problems? After
all, neither the dynamic equivalence approach nor the idiomatic approaches
insist on identity of the messages in every detail; they are content with a
high degree of approximation.
It seems to me that such an objection would betray a lack of apprecia-
tion of the nature and significance of the problems surveyed above. Firstly,
it should be clear from the examples considered that our concern here was
not with matters of detail; on the contrary, the main example from Matthew
2 was chosen because it involved problems about the main point of that
chapter. Nor is this an isolated example, as, for example, the following state-
ments by Headland (1981) about his experience in translating the New
Testament for the Casiguran Dumagat people in the Philippines show: 20

19For a different perspective on viewing the role of processing effort see Gutt (1998).
20 The Casiguran Dumagat are "a N egrito society of hunters and gatherers, living in a
tropical rain forest in Aurora Province, Philippines" (Headland 1981 :25, f. 1).
98 Translation and Relevance

I am fairly sure that no Dumagat believer has yet grasped the impor-
tant significance of the first church council in Jerusalem, as reported
in Acts 15. Luke would be more than disappointed to see how the
Dumagat misses the point. (1981: 19)

There must be a way too, I suppose, to get the message of Hebrews 7

across, that the reason that Jesus is a better intervener for us to God
than the Levitical priests is because Abraham gave tithes to
Melchizedek! That message ought to get across - it is the central
message of that section of the book. Why don't the Dumagat readers
get it? (1981:19)

Thus the real problem is not the loss of detail but of main points and
overall thrust.
Secondly, it is quite true that translation does not always encounter prob-
lems as severe as those considered here; in fact, relevance theory predicts
that the more similar the two audiences are with regard to contextual as-
sumptions needed for the understanding of the text, that is, the closer the
situation is to one of primary communication, the fewer the problems will
be. However, what we are interested in here is not a theory that will work
well only in the less problematic situations, but an account of translation in
general. Therefore our concern in this chapter has been to examine whether
as general theories of translation the dynamic equivalence approach and
the idiomatic approaches provide evidence that the goals they have set for
translation are achievable in principle in both primary and secondary
communication situations. As we have seen, due to inadequate views of
communication and meaning, this they fail to show.
Thirdly, the evidence provided by Headland (1981), Dye (1980) and
probably also by our own experience confirms that this deficiency is not
only a problem from the theoretical point of view, but that it has significant
effects on the effectiveness of translation in practice as well.
Our next question will be whether relevance theory provides a basis for
developing a general theory of translation that can perhaps guarantee com-
munication of the same message - or a close approximation of it - in both
primary and secondary communication situations.

Translating the Same 'Message' by Interpretive Use?

In the previous section we considered some basic problems with the dy-
namic equivalence and the idiomatic approaches to translation. Essentially
these problems are due to an inadequate understanding of communication.
Translating the Meaning of the Original 99

This raises the question: given that relevance theory offers a more adequate
account of communication - can it not provide a framework for what those
other translation theories tried to do? In other words, can it show how the
translator can succeed in communicating to the receptor language audience
the set of assumptions the original communicator intended to communicate
to his original audience?
Within relevance theory, such an endeavour might seem to be analys-
able as a variety of interpretive use (cf. chapter 2 pp. 39): the translator
produces a receptor language text, the translation, with the intention of com-
municating to the receptors the same assumptions that the original
communicator intended to convey to the original audience. As we pointed
out in chapter 2 (p. 40), the intended interpretation of an utterance consists
of its explicatures and/or implicatures. Thus to say that a translation should
communicate the same interpretation as that intended in the original means
that it should convey to the receptors all and only those explicatures and
implicatures that the original was intended to convey.
On closer examination this demand can be taken in two distinct ways;
on the first and stronger reading it can be interpreted as follows:

(4) The explicatures of the translation should be the same as the

explicatures of the original, and the implicatures of the transla-
tion should be the same as the implicatures of the original.

It is not difficult to see that on this reading the demand is likely to create
conflicts in secondary communication situations; thus the preservation of a
given explicature may give rise to an unintended implicature because of the
different cognitive environment of the receptors. Consider the implicature
in Wendland's Tonga example (p. 94 above): combined with the beliefs of
the Tonga people, the explicatures of the text implicated that Naomi's rea-
son for leaving Moab must have been something very serious indeed, and
that she might have been a witch. As will be recalled, this implicature had
to do with the particular timing of her departure - prior to the barley harvest -
and with the death of her two sons-in-law.
Omitting these explicatures would violate the demand of preserving the
explicatures of the original - so it is not really an option for the translator
committed to conveying the meaning of the original. Tolerating such erro-
neous implicatures would also violate this commitment. So one might
consider cancelling them by explicit statements that would contradict them.
100 Translation and Relevance

However, since the original author did not seem interested in communicat-
ing the denials ofthese implicatures, cancelling them by explicit denial would
not be an acceptable option either: it would again misrepresent the assump-
tions communicated by the original.
As we can see now, this kind of problem is not unique or exceptional,
but a perfectly general one: it follows from the inferential nature of commu-
nication that secondary communication situations can give rise to clashes
between the demands of communicating both the explicatures and the
implicatures of the original.
However, there is a weaker reading of the requirement that the meaning
of the original be translated:

(5) The sum total of the explicatures and implicatures of the transla-
tion must equal the sum total of the explicatures and implicatures
of the original.

This reading reduces the danger of a clash between explicatures and im-
plicatures in the translation because the translator is not a priori committed
to maintaining all explicatures as explicatures, and it is imaginable that in
some cases at least he can 'reshuffle' the explicit and implicit assumptions
in a way that will avoid conflict. As we saw earlier in this chapter (pp. 83-4
above) such 'reshuffling' of information is, in fact, considered a legitimate
part of 'communicative' approaches to translation.
This raises an immediate question: what reason is there to expect such a
'reshuffle' to be generally possible? As we saw in chapter 2 (especially pp.
26-35), the meaning of an utterance is not simply the proposition partly
encoded by it, but a set of interrelated assumptions; furthermore, the mean-
ing of each utterance is influenced by the meaning of its predecessors. With
such intricate interrelations it seems rather arbitrary to assume that these
assumptions can be rearranged without significant loss. Returning to our
example of Matthew 2: how likely is it that one can produce a translation
that will convey Matthew's intended 'message' to a present-day English
reader, just by 'reshuffling' its explicatures and implicatures?
Quite apart from this practical problem, such an approach would be faced
with the same issues encountered with the explication of implicit informa-
tion in the idiomatic approaches. Thus it would still have to deal with the
indeterminacy of implicature, which raises the problem of how open-ended
sets of implicatures with varying degrees of strength can be turned into
explicatures. As we saw in our discussion ofthe idiomatic approaches, there
Translating the Meaning of the Original 101

seems to be no principled way of doing this that does not involve arbitrari-
ness and distortion.
It would also still face the problem of avoiding extraneous implicatures.
This would be a particular problem in cases where the intended interpretation
of the original does not include explicatures or implicatures that would can-
cel the unintended implicatures derived by the receptor language audience.
However, in a way the issue of whether it is possible to 'redistribute' the
explicatures and implicatures of the original is only of secondary impor-
tance. Much more important is the basic assumption on which this approach
is based - that it should be possible, at least in principle, to communicate a
particular 'message' or interpretation to any audience, no matter what their
cognitive environment is like.
This assumption has, in fact, been challenged by several translation theo-
rists. Reiss and Vermeer (1984:104), for example, have challenged it along
semiotic lines, arguing that since texts are parts of larger wholes, that is, of
culture and language, their transfer into other cultures and languages will
change the texts themselves. Frawley (1984) has even claimed that transla-
tion theory has, in fact, abandoned "the ridiculous insistence on 'preservation
of meaning'" (1984: l73)Y
While I would not be sure that Frawley's evaluation is justified, rel-
evance theory does confirm that there is a problem of principle here. As we
showed in chapter 2, one of the essential conditions for successful com-
munication is that the set of assumptions to be communicated must yield
adequate contextual effects. Whether or not a given set of assumptions ful-
fils this condition is, of course, dependent on the context in which it is
processed: a set of assumptions that yielded a large number of contextual
effects in one context may yield very few such effects in another context. If
the amount of contextual effects in that other context is less than adequate,
the audience will not be able to recover the intended interpretation, and
may even lose interest in the communication. Thus the view that a 'mes-
sage' can be communicated to any audience regardless of their cognitive
environment is simply false.
This is a fundamental error of all approaches that take for granted that
the translations should convey the 'message' of the original and that see the

21In the field of biblical translation, mismatches in cognitive abilities especially related to
concepts of number and logical connectives, have lead Hnnt (1989) to the conclusion
"that there are speech commnnities for whom it will be impossible to do a full meaningful
translation of the New Testament" (1989: 18). However, the particular problems discussed
by him "only apply to a small fraction of the languages in the world" (1989:22).
102 Translation and Relevance

real problem in finding the right linguistic form for the message:

All translators are agreed that their task is to communicate the mean-
ing of the original. There is no discussion on this point. There is
discussion, however, concerning the linguistic form to be used.
(Beekman and Callow 1974:20)

Even from intralingual experience we know that we cannot necessarily

communicate the same thoughts to just anybody, regardless of their back-
ground knowledge. When addressing different audiences, we tend to change
what we want to convey to them, not only how we say it. From this perspec-
tive it seems quite arbitrary to demand that a translator should, for example,
intend to convey to a modern English audience what Matthew intended to
communicate to his first century audience in the Middle East. When Mat-
thew as the original communicator formed his informative intention, that is,
when he decided what he wanted to communicate, he was influenced by the
mutual cognitive environment he shared with his audience, that is, the
original audience. More specifically, the selection ofthe assumptions which
he intended to communicate was determined by his belief that their recov-
ery would have adequate contextual effects on his audience, under optimal
processmg.
However, as we tried to illustrate using the genealogy found in Mat-
thew's Gospel (see p. 81), for many English readers today many of the
assumptions Matthew intended to convey would seem of comparatively low
relevance. This is not to say that the text cannot be relevant to them at all -
the question is rather whether they will find it relevant consistent with the
principle of relevance.
This perspective may prove fruitful for a better understanding of the
problems encountered by Headland (1981, pp. 97-8 above). Headland re-
jected the diagnosis commonly suggested that the 'rate' of information was
too high and that therefore the problems could be overcome by 'spreading
out' the information over longer units of text:

I hypothesize that the information rate cannot be spread out for

Casiguran Dumagat ... but that secondary information would be bet-
ter eliminated altogether in those passages where it hinders the
message from getting through to the hearer. (1981:22)

Headland felt rather that the total amount of information was simply too
Translating the Meaning of the Original 103

much, no matter how much one spreads it out: the biblical texts very often
seemed to convey more information than the Dumagat people were used to
processmg.
This phenomenon can be explained, at least in part, in terms of inad-
equate contextual effects. Thus suppose the Dumagat readers were unable
to derive adequate contextual effects from much of the information offered
in the translation - which was, after all, intended for a rather different audi-
ence in the first place; then it is clear why the spreading out of information
would not really solve their problem. While conveying information in a less
condensed form can certainly reduce processing effort - it does not, in and
of itself, lead to an increase in contextual effects. What Headland calls 'sec-
ondary information' may well be information that does not yield adequate
contextual effects for the Dumagat audience. Hence from a purely commu-
nication-theoretical point of view Headland's inclination to simply omit some
of the information was right, though the nature of the biblical texts would
make one hesitant to adopt so radical a solution. 22
In short, the point is simply this: the interpretation of a stimulus is al-
ways relevance-determined, and hence context-dependent. It is therefore
not always possible to take some given 'meaning' or 'message' and pro-
duce a stimulus that will be able to communicate just this 'message' to some
particular audience. Whether or not this is possible will depend on whether
the 'message' in question is communicable to that audience in terms of
consistency with the principle of relevance. The view that the main prob-
lem in translation is that of finding the right way of expressing the content
in the receptor language has tended to obscure the problem of the commu-
nicability of the content itself.
It seems, then, that our investigation of the possibility of setting up a
general theory of translation around the requirement that the translated text
should convey the same 'message' to the receptor language audience as the
original was intended to convey to the source language audience has led us
to a negative result: while this aim may be achievable in situations of pri-
mary communication, its achievement in general becomes less likely the
more different the context of the receptor language audience is from that of

22Headland himself writes: "My own doctrine of the Scriptures keeps me from applying
this hypothesis. At least one recent consultant thought I had done too much of it al-
ready. I would like to have done more, but not until I can get a solid translation principle
to back me up" (Headland 1981:22). In ch. 7, pp. 190-8 we shall examine how such
problems can be solved.
104 Translation and Relevance

the source language audience. 23 This problem is not due to a deficiency of

any particular translation theory, nor even oflinguistic mismatches between
the languages involved, but it is a necessary outcome of the inferential na-
ture of communication and its strong dependence on context.24
So if the aim of conveying the same message does not provide a tenable
basis for a general theory of translation - can relevance theory help us to
find some valid alternative?

23Note that this idea could still serve as the goal of a translation theory that would limit
itselfto primary communication situations (cf. p. 76 above); however, such situations
seem to be the exception rather than the rule, and as we shall see in ch. 7, there is a way
of taking into account both primary and secondary communication situations within a
single, integrated approach.
24 It would be interesting to investigate why translation theorists should have pursued
this aim in the first place. Apart from its obvious attractiveness, I think the fact that
much of the thinking of translation theory in the course of history took place in a setting
that did not involve too extreme differences in cognitive environment, e.g. culture, may
have contributed to this.
5. Translation as Interlingual Interpretive Use

Introduction

If, as we argued in chapter 4, interpretive use cannot serve as the basis of a

translation theory designed to convey the same 'message', that is, both the
explicatures and implicatures of the original, could it not serve as a frame-
work for a general theory of translation of a less ambitious kind?l The
simplest possibility would be if translation were interpretive use across lan-
guage boundaries. In other words, a translation would be a receptor language
text that interpretively resembled the original.
From a theoretical point of view, such a theory would, of course, be at-
tractive in that the only stipulation needed to differentiate translation from
other instances of interpretive use would be that the original and the new
text belong to two different languages.
However, before accepting a theory of translation along these lines, we
need to examine more closely what it means to say that an utterance inter-
pretively resembles an original.
Let us look at an example. Suppose that at a linguistic conference a col-
league of mine had missed a particular session that I attended. So he might
ask me: "What did Pike say?" At this point I clearly have a wide range of
options open in answering him. I could:

(a) try to summarize in a couple of sentences what I consider to be

the main points of the lecture;
(b) try to give brief summaries of the main points of the lecture;
(c) just say, "Oh, it was all about discourse";
(d) pick out some particular topics of Pike's talk, perhaps 'cohe-
sion', and represent in some detail what he said about that,
possibly adding some explanations as well;
(e) offer to let him read the full written version of the paper that was
handed out.

What would determine which answer I chose? According to relevance

1 Sperber and Wilson mention translation in passing, suggesting that it is some kind of

interpretive use, involving resemblance in semantic structure (1986a:228), or logical

form (1988a: 136). While such resemblance plays a role (cf. esp. ch. 6, pp. 136-44 sec-
tion 3), more is involved. (See Sperber and Wilson 1995:259.) One reason is that the
logical form does not comprise all aspects of linguistically determined meaning. These
other aspects will be discussed at greater length in ch. 6.
106 Translation and Relevance

theory, my answer would, as always, be determined by considerations of

relevance, and specifically by my assumptions about what my communica-
tion partner might find optimally relevant. Suppose I know that my colleague
is not interested in discourse analysis - that might be an occasion where I
would choose to reply with something like (c). On the other hand, I might
judge that my colleague would be interested in 'cohesion', though he might
not know too much about it - in which case my reply would follow the lines
of (d). Or, if! thought my colleague was very interested in almost anything
that Pike said in his presentation, I should perhaps choose option (e).
Thus the search for optimal relevance would constrain me to express
myself so that with minimal processing effort my partner can derive infor-
mation that is adequately relevant to him. And since his question was about
what someone else said, that is, since I was engaged in interpretive use,
there would be a strong expectation that the information conveyed by my
answer would resemble what Pike was talking about rather than what some
other speaker said or what I thought.
Put more generally, in interpretive use the principle of relevance comes
across as a presumption of optimal resemblance: what the reporter in-
tends to convey is (a) presumed to interpretively resemble the original -
otherwise this would not be an instance of interpretive use - and (b) the re-
semblance it shows is to be consistent with the presumption of optimal
relevance, that is, is presumed to have adequate contextual effects without
gratuitous processing effort. This notion of optimal resemblance seems to
capture well the idea of faithfulness, and Sperber and Wilson have, in fact,
stated that in interpretive use "the speaker guarantees that her utterance is a
faithful enough representation of the original: that is, resembles it closely
enough in relevant respects" (Wilson and Sperber, 1988a: 137).2

2 Three clarifications seem to be in order here. First, strictly speaking, we are talking here
about instances of "(at least) second-degree interpretations" (Sperber and Wilson, 1986a:
238) since "every utterance is an interpretive expression of a thought of the speaker's"
(1986a:231), as we pointed out in ch. 2, pp. 39-41. However, nothing crucial for our
argument depends on this matter, and this looser use, which is also employed by Sperber
and Wilson, simplifies the exposition. Secondly, on the view I proposed above, the no-
tion of faithfulness itself already implies adequacy. Thus one could simply speak of an
utterance as being 'faithful' rather than 'faithful enough'. Thirdly, the use of the term
'guarantee' is open to misinterpretation: as Sperber and Wilson point out elsewhere, the
principle of relevance does not entail that the communication will always succeed - the
utterance may not live up to the guarantee (cf. Sperber and Wilson 1986a:158ff.). With
regard to interpretive use, this means that what is given by the speaker is better described
as a presumption of faithfulness rather than a guarantee. See the clarification on this point
in Sperber and Wilson (1995:293, note b to Chapter 1).
Translation as Interlingual Interpretive Use 107

This brings us back to the question briefly raised in chapter 2 (p. 41: is
this general notion of faithfulness useful for translation or is it not perhaps
too vague - after all, 'close enough resemblance in relevant aspects' does
not seem to determine anything very concrete?
The answer is that the principle of relevance heavily constrains the
translation with regard to both what it is intended to convey and how it is
expressed. Thus if we ask in what respects the intended interpretation of
the translation should resemble the original, the answer is: in respects that
make it adequately relevant to the audience - that is, that offer adequate
contextual effects; if we ask how the translation should be expressed, the
answer is: it should be expressed in such a manner that it yields the intend-
ed interpretation without putting the audience to unnecessary processing
effort. Hence considerations of relevance constrain both the intended in-
terpretation of the translation and the way it is expressed, and since
consistency with the principle of relevance is always context-dependent,
these constraints, too, are context-determined.
These conditions seem to provide exactly the guidance that translators
and translation theorists have been looking for: they determine in what re-
spects the translation should resemble the original - only in those respects
that can be expected to make it adequately relevant to the receptor language
audience. They determine also that the translation should be clear and natural
in expression in the sense that it should not be unnecessarily difficult to
understand. 3
Let us test this account of faithfulness by applying it to a number of
examples, and by comparing it to some of the rules and principles that have
been advocated to achieve faithfulness in translation.

Faithfulness in Interlingual Interpretive Use

Let us begin with an example from the sphere ofliterary translation. Adams
(1973) talks here about the problem of mismatches in grammatical catego-
ries between languages, the case in point being that ofthe distinction between
vous and tu in French, which is not available in contemporary English:

At a climactic moment in Stendhal's Le rouge et Ie noir (Book II,

chap. 19), Julien Sorel, after weeks of solitary suffering, has finally
climbed back into Mathilde de la Mole's good graces, and so under-
takes once more the perilous ascent, via a ladder, to her midnight

3 We shall return to the issue of naturalness later (see p. 122).

108 Translation and Relevance

bedroom. She receives him with ecstatic, unbounded delight, crying,

'C'est donc toi!' And just here C. K. Scott-Moncrieff - for whose
extraordinary gifts as a translator I have, as a general rule, only the
highest respect - slips on the insidious banana peel, and translates,
'So it is thou!' What girl of high social rank and free social manners
ever greeted a lover that way? (Adams 1973: 14)

To give an adequate account of this example, we will need to look at it

from two perspectives. Firstly, from the perspective of the translator, can
the notion of interlingual interpretive use explain why Scott-Moncrieff should
have chosen such a rendering? Secondly, from Adams's perspective - can
our account explain why he should find it unsatisfactory?
Let us look at it first from the translator's perspective. In view of Adams's
high regard for Scott-Moncrieff's abilities it seems out of the question that-
like a student in a beginner's class on translation - the translator simply
looked into an English dictionary to find a pronoun corresponding in se-
mantic meaning to French tu and then put it into the translation, without
being aware of its associations.
Neither does it seem completely adequate to me to dismiss this matter as
a kind of slip-up, as Adams seems to suggest. It seems improbable, though
not altogether impossible, that a form as uncommon in contemporary Eng-
lish as thou would slip into the text unnoticed: such an explanation would
have seemed appropriate had the translator used you. So there is reason to
assume that this was a deliberate choice on the translator's part.4
On this assumption, then, Scott-Moncrieff must have considered his
rendering faithful enough to the original, that is, he must have thought that
its interpretation resembled the original in assumptions that would make it
adequately relevant to the receptors, and, moreover, that these assumptions
would be recoverable from his rendering without unnecessary process-
ing effort.
In particular, given that you would have been the normal first choice, he
must have intended to convey special contextual effects by using the less
common, hence more costly pronominal form thou. The special effects he
had in mind become clear when one takes a closer look at the pronominal
form tu in the original: as part of their cultural background knowledge,
the French know that the 2nd person singular form of the pronoun is used

4 If it was not a deliberate choice but a slip-up due to temporary lack of attention, then

this phenomenon does not belong in an account of communication but rather in an

account of psychological errors.
Translation as Interlingual Interpretive Use 109

between people who have an intimate social relationship. Most likely this
information is stored in the encyclopaedic entry associated with the word
tu, and hence this information becomes highly accessible whenever tu or
one of its inflected forms is used. 5 Due to its high accessibility it can give
rise to quite manifest contextual implications: in this case, that there was an
intimate relationship between Mathilde and Julien, a significant implicature
at this point.
We can see, then, a possible reason why the translator should have cho-
sen thou rather than you in English: you, being indeterminate between
singular and plural, could not have yielded this implicature about the inti-
mate nature of the relationship between these two characters, and so his
choice was intended to preserve this implicature for his readers.
But why does Adams regard this solution as faulty rather than success-
ful? He explains:

'Thee' and 'thou' belong, for most people, to obsolete or ecclesiasti-

cal language; intimacy is the feeling these terms preeminently don't
express. (1973: 14, italics as in original)

Adams's evaluation touches on three distinct points: a) on the audi-

ence envisaged, b) on what thou does convey to them, and c) on what it
doesn't convey.
Let us begin with the question of audience. Adams argues in terms of
'most people', which probably must be interpreted as 'most English readers
who would be interested in literature of this kind'. If this is right, then his
further claim is that for such readers thou belongs to 'obsolete or ecclesias-
tical language '. In terms of relevance theory this means that the encyclopaedic
entry associated with thou contains the information that it is a word no longer
used, except in certain religious contexts. Note that there is no significant
difference in semantic meaning: both tu and thou semantically represent a
single addressee.
As to what it does not convey, the encyclopaedic entry associated with
thou does not contain any information about it being appropriate as a form
of address between people with a close social relationship. This informa-
tion is not present there with 'most people'. Thus Adams's observations
can easily be expressed in relevance-theoretic terms.

5 For further information about encyclopaedic entries and the organization of informa-

tion in memory see ch. 6, pp. 141-2.

110 Translation and Relevance

However, relevance theory draws our attention to another factor, and

that is processing effort. Not only does thou have information associated
that classifies it as 'obsolete or ecclesiastical' - it is also likely that rare
lexical forms like this one are stored in less accessible places in memory.
Hence such unusual forms require more processing effort, and given that
the communicator would have had available a perfectly ordinary alterna-
tive, you, the audience will rightly expect special contextual effects, special
pay-off, from the use of this more costly form.
It is, therefore, not only the mismatch in associated information, but also
the increase in processing cost that is responsible for the infelicity of this
translation. It makes the audience look for special contextual effects, hence
makes it willing to pay special attention not only to the semantics but, for
example, to information associated with the word itself. Since the informa-
tion differs from that in the original, the audience may be misled toward
unintended contextual implications.
Note, however, that the interpretation the audience will arrive at need
not necessarily be, for example, that Mathilde used obsolete language with
Julien. The reason is that it is not simply the first interpretation that comes
to mind that the audience is entitled to take as the intended interpretation,
but rather the first interpretation that comes to mind and that is consistent
with the principle of relevance. In other words, the reader usually tries to
'make sense' of what he reads, and it would probably be difficult to make
sense of an interpretation that suggested the use of either archaic or relig-
ious forms of language at this point in the story. This means that this
interpretation would run into problems: there would be no obvious way in
which this interpretation could have been intended to yield a significant
number of contextual implications, that is, in which it would 'make sense'
in this context.
At this point in the interpretation process the audience could react in
different ways: it could, for example, assume that there is something wrong
with the translation. In this case, it could, for example, leave the matter
unresolved and go on reading, or - especially if this is not the first point of
difficulty - perhaps break off the interpretation process altogether, that is,
discard that translation.
Alternatively, it could assume that there is in fact an interpretation con-
sistent with the principle of relevance, though one involving a higher
investment in processing cost. Consequently, the reader would try to expand
the context further in order to gain an adequate return in terms of contextual
implications. The results this would lead to would depend crucially on what
Translation as Interlingual Interpretive Use 111

knowledge the reader has, on his intellectual powers, experience with inter-
pretation of literature and so forth. An imaginative reader may perhaps
suspect some irony here and would consequently misinterpret this passage,
especially if he did not have access to the original.
But it is also possible that the reader is one of those 'semilanguaged'
people that Adams (1973 :xii) talks about in the preface to his book, that is, a
person less than fully bilingual with the original language, but with some
knowledge of it. Such a person might extend his context to include the rec-
ognition that he is reading a translation from French, which might lead to a
further extension that brings in knowledge of French. Thus he might realize
that English thou stands for French tu, which might make accessible the
knowledge of the social conventions relating to the use of tu, which would
enable him to derive contextual implications about the degree of intimacy
that seems to obtain now between Mathilde and Julien, and any further con-
textual implications this might have for the understanding of the novel.
Even for readers familiar with French this interpretation would be prob-
lematic, because its recovery involves considerable processing effort, and it
is not clear whether many readers would have been prepared to invest this
extra effort.
Thus we see that we can account for the problems of communication
encountered in this example in terms of interlingual interpretive use: for the
audience represented by Adams - which is probably the majority - Scott-
Moncrieff's rendering here falls short both in closeness of resemblance and
in adequate relevance.
This example involves problems of resemblance on a point of stylistic
detail. By way of contrast let us now look at a translation where there is
concern about resemblance in much more fundamental respects.
The example I want to look at concerns Levy's (1969) discussion of how a
particular poem by Morgenstern could or should be translated into English.
In Christian Morgenstern's poem 'The aesthetic weasel', in the verses

Ein Wiesel
sass auf einem Kiesel
inmitten Bachgeriesel
[A weasel
sat on a pebble
in the midst of a ripple of a brook]
(translation from Levy (1967)6

6 Levy (1967) uses the same example (1967: 1178f.).

112 Translation and Relevance

the playful rhyme is more essential than the zoological and topographic
exactness, for Morgenstern himself adds

Das raffinier-
te Tier
Tat's urn des Reimes Willen.
[The shrewd
animal
did it for the sake of the rhyme]
(translation my own)
Max Knight translates,
A weasel
perched on an easel
within a patch of teasel

and adds in the preface quite rightly that other translations would be equally
possible:
A ferret
nibbling a carrot
in a garret
or
A mink
sipping a drink
in a kitchen sink
or
A hyena
playing a concertina
manarena
or
A lizard
shaking its gizzard
in a blizzard

More important than the individual meanings in detail is here the pres-
ervation of the play on words. (1969: 103f., translation my own)7

This example is particularly interesting because it is presented by Levy

as an illustration of one of the most basic problems of translation: what the
translator should do when he cannot possibly preserve all the features of
the original:

7 All quotations from LevY (1969) are given in my own translation, unless indicated other-

WIse.
Translation as Interlingual Interpretive Use 113

In translation there are situations which do not allow one to capture

all values of the original. Then the translator has to decide which
qualities of the original are the most important and which ones one
could miss out. The problem of the reliability of translation consists
partly in that the relative importance of the values in a piece oflit-
erature are recognised. (Levy 1969: 103)

The 'values' among which the translator has to choose are described by
Levy in terms of 'semantic functions':

in Morgenstern's text some words have two semantic functions:

1. their own denotative meaning, 2. a function in a structure ofa high-
er order (and just this was retained in the translation). (1969: 104)

The words in question are Wiesel, Kiesel and Bachgeriesel. Their ap-
proximate denotations are 'weasel', 'pebble' and 'ripple ofa brook'. Their
'higher' semantic functions are that they establish the rhyming pattern of
this poem, more particularly the pattern of the 'Kalauer', a kind ofpun.8
The problem is, of course, that while English has ways of expressing
these denotations and also of rhyming, it does not happen to offer a set of
words or expressions that fulfil both conditions at once: that is, that have
these denotations and also rhyme. Therefore the translator has to make a
choice about what properties he wants to preserve.
Levy proposes that this choice follows from a 'functional hierarchy' (cf.
chapter 1, pp. 14-15) that determines the relative ranking of importance of
various aspects of word meaning:

In general, one can say that with words that have several expressive
functions, the function in the semantic complex of the higher order is
the more important one, be it the context (the sentence, the paragraph
etc.), be it the character of a person, the fable or the philosophical
objective of a work. The highest complex of expression, sometimes
referred to as the idea of the work, its world view, dominates the solu-
tion of problems in some lower unit, e.g. when choosing the stylistic
level, and this in turn determines the solution of problems of detail.
(LevY 1969: 104f., reproduced here for convenience from p. 14 above).

The particular 'hierarchy' Levy proposes for this example looks as follows:

8Kalauer is a colloquial term and refers to a 'simple, funny pun' ['einfaches, witziges
Wortspiel'; R. Klappenbach and W. Steimitz (eds) (1969), Worterbuch der deutschen
Gegenwartssprache, Akademie Verlag, Berlin, p. 2016].
 .....
.....
4

4th degree "Kalauer"-style

3rd degree pun

2th degree animal object of activity place of activity

I st degree

~
§
""
§:
Ein Wiesel sa(/, auf einem Kiesel inmiUen Bachgeriesel a"
~

(1969:104) §
i::l...
~
til
~
§
(")
til
Translation as Interlingual Interpretive Use 115

Levy claims that while the translations differ in their concrete semantic
content, they converge in preserving the agreement in rhyme between words
that correspond to each other with regard to their 'functions' at the second
level of abstraction; thus they preserve the agreement of rhyme of "1. the
name of the animal, 2. the object to which its activity is geared, 3. the place
of activity. In all five translations it is only these three abstract functions of
the three individual verses that are preserved and not the concrete meanings
of the individual words" (1969:104).
As an account of how the translator is to make his decisions, this ap-
proach raises a number of questions. Perhaps the most obvious one concerns
the nature of the 'functional hierarchy' itself: as we already remarked about
hierarchical functions in general (cf. chapter 1), it is not at all clear on what
principles Levy's hierarchy is constructed: the lowest level seems to consist
of the actual words of the text, the second level of some semantic abstracts:
'animal', 'obj ect of activity', and 'place of activity'. This already raises a
number of questions, one of which is how one determines what abstract
notions to posit. For example, it seems that the phrase 'auf einem Kiesel'
('on a pebble') is abstracted as 'object of activity' and the phrase 'inmitten
Bachgeriesel' ('in the midst of a ripple of a brook') as 'place of activity'
[' Schauplatz']. However, the fact is that both phrases refer to locations. But
if this is the case, that is, if the 2nd degree of abstraction refers to 'animal',
'pace' and 'place', then none of the four alternatives would meet Levy's
requirement of preserving the abstract functions correctly - only the first
one would.
Furthermore, levels 3 and 4 of the hierarchy seem to belong to a differ-
ent domain altogether: they do not naturally follow on from level 2,
presenting perhaps a further degree of abstraction along semantic lines,
but belong to the domain of stylistics, distinguishing as they do between
'pun' and "'Kalauer" -style'. Thus the overall organization of this hierarchy
remains unclear.
So the general question is: how can the translator know what the proper
representation of the text is at any higher level of function? Levy does not
answer these questions, yet without an answer the appeal to function only
serves to replace the translator's question 'what features should I preserve?'
by another set of questions, such as 'what abstract functions are there in the
text?' and 'what is the functional hierarchy that determines their relative
value?'. Since the answer to these questions can depend on text-external
features, such as purpose of communication, audience etc., it is doubtful
that adequate 'functional hierarchies' can be set up.
116 Translation and Relevance

However, it seems more than doubtful anyway that such 'functional hi-
erarchies' play any significant role here at all. What is actually being done
here can both be accounted for and evaluated in terms of interpretive use
within the relevance-theoretic framework.
Thus to start with, a translation of Morgenstern's poem will come with
the presumption that its interpretation resembles that of the original 'closely
enough in relevant respects'. This raises the question of what aspects of
the original the receptors would find relevant. Using his knowledge of the
audience, the translator has to make assumptions about its cognitive en-
vironment and about the potential relevance that any aspects of the
interpretation would have in that cognitive environment.
In our example, what Levy presented as 'abstract functions' are, in fact,
assumptions that he believed not only to be part of the original interpreta-
tion but also of adequate relevance to the English target audience. These
assumptions include the following:

(a) The text constitutes a play on words;

(b) The text refers to an animal, an object and a location at the end
of each line, and the words used to refer to them rhyme;
(c) The text is of the 'Kalauer' kind.

Assumptions (a) and (c) would be implicated conclusions of appropriate

descriptions of the original poem, with (c) assuming knowledge of the liter-
ary category 'Kalauer'. The assumptions in (b) are of mixed origin: the first
half of (b) would result from some analytic implications, such as 'a weasel
is an animal', 'a pebble is an obj ect', but the second half would again in-
volve a description.
As to differences, our observation was that the abstractions 'object' and
'place' rather than 'place' and 'place' seem somewhat arbitrary. In terms of
relevance theory this means that while both the original and Knight's cho-
sen translation have 'an animal was situated in place X in place Y' as one of
its analytic implications, the four alternative translations do not share this
implication but have the following implication instead: 'an animal did some-
thing to object X in place Y'. Levy does not comment on this difference.
What about Levy's suggestion that somehow the preservation of the as-
sumptions above is more important than the preservation of the fact that the
animal referred to in the original was a weasel rather than a ferret or lizard,
that it was perching on a pebble rather than playing a concertina or that it
was in the middle of a stream rather than in a kitchen sink or arena?
Translation as Interlingual Interpretive Use 117

I think there is a sense in which Levy's intuition is right - how can rel-
evance theory account for this?
We can account for this intuition if we assume that the relevance of the
original lay not in the assumptions it conveyed about what a certain weasel
did, that is, sat on a pebble in a stream, but rather in the assumption that
the animal acted in this way with a literary motive in mind: to give rise to a
rhyming poem. It is this amusing assumption that seems to be primarily
responsible for the relevance of the original, and hence Levy's intuition
that this assumption is particularly important can again be accounted for in
terms of relevance.
However, it should be noted that this condition - the comparative de-
gree of relevance of a certain assumption in the original interpretation - is
not a sufficient condition for its inclusion in the translation. This can be
illustrated from the auxiliary translation given above in addition to Knight's
renderings:

'Auxiliary translation':

A weasel
sat on a pebble
in the midst of a ripple of a brook
(p. 111 above)

This translation obviously does not attempt to preserve the rhyme, hence
would not serve well to convey the main assumption just mentioned, and
yet would seem to be appropriate to our discussion. Again, this follows
from our definition of faithfulness which calls for resemblance in relevant
respects: on the assumption that some readers may not know enough Ger-
man to understand the semantic content of the poem, this translation helps
them by giving them easy access to the semantically determined meaning of
that poem, and knowledge of that meaning of the original is relevant to the
overall thrust of our discussion.
Furthermore, it does not follow that preservation of those more impor-
tant, 'abstract' features necessarily frees the translator from the obligation
to preserve any of the more 'concrete' semantic properties, as Levy's func-
tional treatment seems to suggest by treating the four other renderings as
equally possible translations. As we saw above, those four alternatives dif-
fer from the original in certain assumptions that they could reasonably be
expected to share, as Knight's actual translation shows. Thus there is a sense
118 Translation and Relevance

in which the four alternatives differ from the original in unnecessary and
rather arbitrary respects.
This intuition can be explained in terms of our relevance-based account
offaithfulness: the translation is presented by virtue of its resemblance with
the original in relevant aspects. All the four alternatives considered miss,
for example, resemblance in that Morgenstern's poem was about a weasel,
rather than a ferret, lizard, or some other animal. This fact may well be rel-
evant, for example, for ease of identification and reference. If so, then these
four versions are less faithful than they could reasonably have been, since
this resemblance could have been retained without increase in processing
effort, as Knight's translation shows.
This last point is an important consideration. Sometimes it is possible to
achieve a higher degree of resemblance but only at the cost of a decrease in
overall relevance because it involves an increase in processing effort that is
not outweighed by gains in contextual effects. Under those conditions the
rendering showing less resemblance will usually be the one required for
successful communication.
We have already looked at cases illustrating this point. When Scott-
Moncrieff chose the rendering thou, he probably did so on the assumption
that resemblance in the second person singular form of the pronoun was a
feature of the original worth preserving, that is, one that would have ad-
equate contextual effects. However, what he apparently failed to consider
was not only that English thou conveyed quite strongly features not part of
the original interpretation, but also that the relevance of this increase in
resemblance was actually jeopardized for many readers by the increase in
processing cost that it required.
This brings out another important point: whatever decision the transla-
tor reaches is based on his intuitions or beliefs about what is relevant to his
audience. The translator does not have direct access to the cognitive envi-
ronment of his audience, he does not actually know what it is like - all he
can have is some assumptions or beliefs about it. And, of course, as we
have just seen, these assumptions may be wrong. Thus our account of trans-
lation does not predict that the principle of relevance makes all translation
efforts successful any more than it predicts that all ostensive communica-
tion is successful. In fact, it predicts that failure of communication in general
is likely to arise where the translator's assumptions about the cognitive en-
vironment of the receptor language audience are inaccurate.
Thus we see that the relevance-based account of faithfulness is not, in
Translation as Interlingual Interpretive Use 119

fact, vague at all. Since it is subj ect to constraints of relevance, it constrains

translational faithfulness with full sensitivity to context, and yet without
any need for rules and principles of translation that appeal to functional or
other classificatory schemes. In fact, it seems that the bulk of rules and
principles that have been advanced in writings on translation are concerned
not so much with matters of general translation theory but rather deal with
matters of relevance. Let us examine this claim in the next section.

The Origin of Translation Principles

Starting with Levy's treatment of Morgenstern's poem 'The aesthetic wea-

sel', one general guideline given by Levy was that the translator "has to
decide which qualities of the original are the most important and which
ones he can miss out" (1969:103). We saw that the notion of 'importance'
alluded to here could be explained in terms of relevance.
In fact, once one pays attention to this matter, one is struck by how
frequently guidelines in translation involve concepts like 'importance', 'sig-
nificance', and even 'relevance' itself. Vernay (1974), for example, defines
translation as - "an act which transfers information given in language A
into a language B in such a way that the amount of relevant information
received in language B will be identical with that in language A" (1974:237,
translation and italics my own). However, these notions are not treated as
theoretically interesting, and so one ofthe key factors in translation is missed
out on. Relevance theory helps us to spell out the crucial role played by this
factor, not only with regard to general translation principles, but also to
rather specific ones.
Take, for example, the guidelines given in Beekman and Callow (1974)
concerning "Lexical equivalence across languages - when things or events
are unknown in the RL" (1974:191-211). There Beekman and Callow list
three main options: "equivalence by modifying a generic word", "equiva-
lence using a loan word" and "equivalence by cultural substitution". Looking
at the first option first, this involves the addition of a 'descriptive modifica-
tion' to a 'generic term' to supply specific meaning absent from the generic
term itself. One of the examples given is the following:

the word passover has quite a few significant components of mean-

ing, including feast, religious, Jewish, the passing over of the angel
without hurting them, deliverance from Egypt, and eating sheep.
However, a descriptive equivalent including all of these would be
120 Translation and Relevance

cumbersome, and in a case like this, a good equivalent can usually

be arrived at which focuses on those components which are most
significant to the context, leaving the others to be implied or taught.
Some renditions of passover have been 'the feast at which they ate
sheep', 'the Jewish feast about God delivering them' , and 'the feast
remembering when God's angel passed by'. (Beekman and Callow
1974:192)

This guideline is a straightforward application of the principle of rel-

evance; it draws the translator's attention to the fact that, due to differences
in cognitive environment, the receptor language audience lacks informa-
tion associated with a concept in the original, here the concept 'passover'.
The observation that expressing all the missing information in translation
would be too cumbersome refers to the fact that this would involve too
much processing effort, especially since not all of that information would
yield contextual effects in every context. The guideline given as a solution,
that is, to focus "on those components which are most significant to the
context", follows straightforwardly from the principle of relevance; it means
that the translator should bring out such information that will make for op-
timal relevance in that particular context.
Often the guidelines provided have to do with particular sets of assump-
tions that are part ofthe receptor language audience's cognitive environment.
Thus regarding 'equivalence by cultural substitution' (cf. p. 119 above),
Beekman and Callow give the following ruling:

For historical references, it is inappropriate to make use of cultural

substitutes, as this would violate the fundamental principle of his-
torical fidelity. (1974:203)

Thus, while unknown concepts in 'didactic passages' used to illustrate

a teaching point can be translated by cultural substitutes, - for example in
Mark 4:21 "on a candlestick" has been rendered in Korku, an Indian lan-
guage, as "on a grain bin" - this cannot be done in historical passages:

In Matthew 21:19-21 and Mark 11:l3,14, Jesus curses a fig tree.

This is again a historical incident, so the translation should refer to a
fig tree, not an avocado or some other better known, local tree.
(Beekman and Callow 1974:203)

However, the 'principle of historical fidelity' to which Beekman and

Callow allude here does not follow from principles of translation theory as
Translation as Interlingual Interpretive Use 121

such, but from the high importance attached to matters of history in the
Christian faith. For the Christian audience the accurate preservation of his-
tori cal detail is seen as so highly relevant that it outweighs the additional
processing effort required. This is why in 'historical passages' the transla-
tor is advised to be content with less' dynamic' but historically more accurate
renderings.
This relevance-based account of translational fidelity or faithfulness can
also account for principles of translation that recommend 'explications',
such as we considered in chapter 4 (pp. 83-5, section 2.2). On our account,
they are motivated by the assumption that certain implicatures of the origi-
nal are highly relevant to the audience, but cannot be derived by them from
the semantic contents alone, due to contextual differences. Therefore the
translator attempts to communicate these assumptions to the receptors as
explicatures.
Newmark (1988) lists the following collection of specific guidelines,
covering a wide spectrum of different kinds of translation:

A technical translator has no right to create neologisms ... whilst an

advertiser or propaganda writer can use any linguistic resources he
requires. Conventional metaphors and sayings ... should always be
conventionally translated ... but unusual metaphors and comparisons
should be reduced to their sense if the text has a mainly informative
function ... The appropriate equivalents for keywords ... should be
scrupulously repeated throughout a text in a philosophical text ... In
a non-literary text, there is a case for transcribing as well as translating
anykey-wordoflinguistic significance ... (Newmark 1988:15)

It is not difficult to see that each of these rules is an application of the prin-
ciple of relevance to an audience with particular kinds of interests.
Other translation principles arise from the influence of processing effort
in communication. It seems that the translation rule that common expres-
sions in the original should be translated by equally common receptor
language expressions is rooted in this cost factor. For example, Newmark
(1988) observes that "to translate 'Ich habe keine Ahnung' as 'I have no
premonition' would give Ahnung too much particularity", and such a trans-
lation would violate, among other things, the principle of "equivalent
frequency of usage" (1988:145). Looked at in terms of relevance theory,
for many speakers of English the word premonition is rarely used, much
more rarely than, for example, the word idea. Since the organization of our
memory reflects frequency of usage, with less often used entries stored in
122 Translation and Relevance

less accessible locations, accessing premonition in memory would require

more effort than accessing idea. 9 Consistency with the principle of relevance
would make the audience look for an amount of contextual effects that would
justify the effort spent: however, if the original German expression was
intended to communicate no more than 'I have no idea', the audience would
not find adequate contextual effects, and hence would find the utterance
less than optimally relevant. This would explain the feeling of unnatural-
ness, without reference to either a 'principle of equivalent frequency of usage'
or to the notion of 'naturalness' itself.
'Unnaturalness' in translated texts often seems to involve gratuitous
processing effort on the receptor audience's part: perhaps due to interfer-
ence from the original language or insufficient mastery of the receptor
language, the expression used by the translator may tum out to require more
than optimal processing cost on the audience's part.
However, when evaluating such instances of 'unnaturalness', one has to
keep in mind that the feeling of inconsistency with the principle of relevance
may also be due to contextual differences. Thus it is possible that the com-
plexity of the receptor language expression was indeed justified because it
would lead to adequate contextual effects if processed with the right con-
textual assumptions. The reason why the audience failed to recover those
contextual effects could have been that it interpreted the utterance without
the intended contextual assumptions. In chapter 6 (pp. 144-6) we shall dis-
cuss an example where the translators themselves felt that the syntactic
structure of the original was unnecessarily complex and hence 'simplified'
it in their translation. What they did not realize was that the more complex
structure was intentional and led to special contextual effects, if processed
in the right context. I believe that many instances of 'unnaturalness' in trans-
lation can be accounted for in terms of inconsistency with the principle of
relevance, if both processing effort and contextual effects are considered.
The principle of relevance can also be seen behind guidelines given for
oral translation (simultaneous interpretation). Thus Namy views "good si-
multaneous interpreting" as "the art of re-expressing in one language a
message delivered in another language at the same time as it is being de-
livered", and he stipulates that "the re-expression should be clear,

9 Because of its frequency of use, the expression 'to have no idea' may actually be
stored in memory as a unit, with some kind of ready-made meaning, rather like an idiom;
the expression 'to have no premonition' would probably not be stored in this way, but
would need to be interpreted step by step on the basis of its linguistic structure, and
hence again be more costly.
Translation as Interlingual Interpretive Use 123

unambiguous and immediately comprehensible, that is to say, perfectly idio-

matic, so that the listener does not have to mentally re-interpret what reaches
him through the earphones" (1987:26). To achieve such "good simultane-
ous translation", the oral translator "can and, I contend, must take as much
liberty with the original as is necessary in order to convey to his audience
the intended meaning ... of the speaker" (Namy 1978:27). Thus Namy asks
rhetorically:

When a French Polytechnicien, addressing his American counter-

part, says: 'Quelle est la proportion de main d'oeuvre indirecte que
vous appliquez a l'entretien du capital installtS?' should the inter-
preter say 'What is the proportion of indirect labour you apply to the
maintenance ofthe fixed capital?' or should he say, 'How many people
do you employ to keep the place clean and maintain the equipment?'
(Namy 1978:27)

Namy's general answer is that "The interpreter should never hesitate to

depart - even considerably - from the original if in doing so he makes the
message more clear" (1978:27). From the relevance-theoretic point of view
this guideline is motivated by the fact that the translation will be taken up
aurally. Since the stream of speech flows on, the audience cannot be ex-
pected to sit and ponder difficult renderings - otherwise it will lose the
subsequent utterances; hence it needs to be able to recover the intended
meaning instantly. Accordingly, the translator will often settle for render-
ings that resemble the original less closely but get across easily what he
considers to be adequately relevant aspects of the origina1. 1o
While our interest here is not so much with the history of translation and
translation theories, it is tempting to suggest that diachronically, too, the
different ways in which people have translated at different times in his-
tory can be attributed to differences in what the translator believed to be
relevant to his contemporary audience. Thus Bassnett-McGuire (1980) sug-
gests that the adjustments made, for example, by Wyatt should not be simply
discarded as 'adaptations' but be seen as attempts at relating the meaning of
a poem to the readers of the time. Taking from Wyatt's translation of
Petrarch's sonnet a few lines that deal with the death of Cardinal Giovanni
Colonna and of Laura in AD 1348, she points out that the changes made by
the translator indicate that "the translator has opted for a voice that will

10 The importance of the time factor with regard to relevance has been pointed out by

Sperber and Wilson (l986a:l60).

124 Translation and Relevance

have immediate impact on contemporary readers as being of their own time"

(1980:57), allowing them perhaps even to relate it to the downfall of
Cromwell in AD l540.u Put in terms of relevance theory this would mean
that Wyatt focused on bringing out those assumptions from the original in-
terpretation that would readily yield contextual effects in the cognitive
environment he shared with his target audience.
In each case, the actual 'translation principle' is the same: do what is
consistent with the search for optimal relevance. What differs are the spe-
cific applications of this principle that take into account differences in
cognitive environment that may change the accessibility of particular pieces
of contextual information and hence may affect the relevance of the text or
utterance in that context. 12
Once this is recognized one can see why so much of the literature on
translation is useful, and yet only in a limited way: it is extremely useful in
making the translator sensitive to the importance of the assumptions present
in the cognitive environment in which he produces his translation, not only
with regard to the content of the text to be translated, but also with regard to
the nature of the whole act of communication in which he is involved. On
the other hand, the usefulness of such guidelines is limited because each
guideline is an application of the principle of relevance to some set of cir-
cumstances; it is, therefore, valid only under those circumstances. When
the circumstances change, that guideline no longer applies.
This is one reason why translation principles and rules need to be modi-
fied with regard to exceptions or else contradict one another. For example,

11 Wyatt's translation reads as follows:

The pillar pearished is whearto I lent;
The strongest staye ofmyne unquyet mynde:
(CCXXXVI)
The original reads:
Rotta e l'alta colonna e'l verde lauro
Che facean ombra al mio stanco pensero;
(CCLXIX)
This is translated by Bassnett-McGuire as:
Broken is the tall column (Colouna) and the green laurel tree (Laura)
That used to shade my tired thought. (1980:57)
12 In the original edition there was here a metaphorical reference to 'rankings' of rel-

evance; despite its enclosure in single quotes, this expression was misunderstood in a
literal sense by some, who then suspected that relevance theory required some scheme
of ranking after all (cf. e.g. Hatim 1998: 182). This is not so: differences in relative
relevance simply arise from differences in processing cost due to the accessibility of
information. There are no 'hidden' hierarchies or schemes for the grading of relevance.
Translation as Interlingual Interpretive Use 125

according to de Waard and Nida one of the conditions under which "changes
offonn can and should be made" (1986:37) in translation is "when a fonnal
correspondence involves a serious obscurity in meaning" (1986:38). Yet at
the same time they give the following exception to this rule:

On the other hand, there are certain important religious symbols

which, though often obscure in their meaning, are necessarily im-
portant for the preservation of the integrity and unity of the biblical
message, e.g. expressions like 'Lamb of God' , 'cross' or 'sacrifice'.
(1986:38)

A similar case is found in Hofmann's treatment of drama translation.

Hofmann proposes a trichotomous model, involving the "expressive level",
the "content level" and the "pragmatic level" (1980:28). In view of the spe-
cial requirements in drama, he holds the view that pragmatics embraces
"purpose and objective of the translation", and "opens up for the receptor
the ... components of cognitive understanding and aesthetic pleasure, the
challenging nature and effectiveness on stage" (1980:37). Hofinann there-
fore declares the pragmatic aspects to be an invariant, not a variable in drama
translation:

From the well-known fact that e.g. philological reliability, i.e. literal
fonnal and/or semantic invariance, does not always equal theatrical
reliability it must be concluded that the variable pragmatics, here
its component 'effectiveness on stage', is raised to an invariant.
(1980:37, translation and italics my own)

Hofmann further supports his position by reference to other translation

theorists; he names Kloepfer, Lev)!, Vinay, Reiss and quotes Mounin "who
presents the demand for this invariance in a particularly rigid way":

Prior to faithfulness to the wording, to grammar, to syntax and even

to the style of each individual sentence in the text must be faithful-
ness to that which made this piece a success in its original country.
One has to translate its effectiveness on stage first before giving con-
sideration to the reproduction of its literary or poetic qualities, and if
in this conflicts should arise, then priority must be given to the effec-
tiveness on stage. (Mounin 1967: 137; Hofmann 1980:27, translation
my own)

However, despite this strong commitment to pragmatics and effectiveness

126 Translation and Relevance

on stage, when it comes, for example, to the treatment of symbolisms in

drama, Hofmann sees the need to make exceptions. Because of the poten-
tial relevance ofthese symbolisms for possible film-productions ofthe drama
in question "all verbalised emblems should be retained in the translation,
too", even though this will mean that "as a rule, the recipient will in spite of
possible iconic functional actions remain on the level of understanding of a
Claudius ('I have nothing with this answer', Hamlet, III, ii, 93) or Rosecrantz
('I understand you not, my lord', Hamlet, IV, ii, 21) ... " (Hofinann 1980: 66).
A last example may be taken from Levy (1969). The author sees an im-
portant difference between the translation of measures and weights on the
one hand, and that of currencies on the other. He claims that unfamiliar
measures can be converted into 'metres and kilograms' because the reader
may have no idea of the content of less familiar foreign measuring units.
Regarding currencies, however, Levy makes the following claim:

Foreign currencies cannot be converted because a currency is always

specific to a certain country and the use of Mark would localise the
translation in Gennany. (1969:97)

However, this distinction between weights and measures versus cur-

rencies does not seem to be so general after all, when one reads a little later
that "one will, for example, keep foreign measures, weights and currencies
in a report of travels [Reisebeschreibung], but in the English verses 'When
first my way to fair I took, Few pence in purse had l' one will be able to
translate as 'ein paar Heller (Groschen) hatte ich'" (1969:107).
In all three examples the guideline proposed built on a generalization
about the comparative relevance of certain aspects of the original, and this
generalization held true for most cases. Yet for each guideline there were
instances where the relevance relations were different, and so the guideline
in question had to be supplemented by further rules to take care of the ex-
ceptions. The principle of relevance, however, accounts for rules and
exceptions alike.
As the last example in particular showed, the different guidelines are
not always reconciled with each other in terms of rules and exceptions but
sometimes contradict one another. Such contradictions can be seen even
more clearly when one compares the principles of translation held by dif-
ferent translators and translation theorists, a situation which led Savory
(1957) to the following, rather negative evaluation of translation principles:
Translation as Interlingual Interpretive Use 127

It would almost be true to say that there are no universally accepted

principles of translation, because the only people qualified to formu-
late them have never agreed among themselves, but have so often
and for so long contradicted each other that they have bequeathed to
us a welter of confused thought which must be hard to parallel in
other fields ofliterature. (1957:49)

Savory follows this claim up with his widely-quoted list of pair-wise

contradictory translation principles:

2 A translation must give the words of the original.

2 A translation must give the ideas of the original.
3 A translation should read like an original work.
4 A translation should read like a translation.
5 A translation should reflect the style of the original.
6 A translation should possess the style of the translator.
7 A translation should read as a contemporary of the original.
8 A translation should read as a contemporary of the translator.
9 A translation may add to or omit from the original.
10 A translation may never add to or omit from the original.
11 A translation of verse should be in prose.
12 A translation of verse should be in verse. (1957:49)

Savory's intuition that these paradoxes can be resolved through 'reader-

analysis' goes in the right direction, since the different 'translation principles'
do reflect differences in what different readers consider to be relevant. What
Savory does not bring out is that the link between different readerships and
different translations lies in the principle of relevance. Thus the contradic-
tions can be resolved when each prinicple is not stated in absolute terms,
but qualified by the condition: "when required for consistency with the prin-
ciple of relevance."

Conclusion

Thus it seems that an account of translation as interlingual interpretive use

has much to commend it. In fact, it could be said to achieve what translation
theory has been attempting to do for a long time - that is, to develop a
concept of faithfulness that is generally applicable and yet both text- and
context-specific. It is generally applicable in that it involves only notions
believed to be part of general human psychology - the principle of relevance
128 Translation and Relevance

and the ability to engage in interpretive use. It is text-specific in that inter-

pretive use will link the communicative intention of the translator to the
intended interpretation of the original text. It is context-specific in that the
search for consistency with the principle of relevance always brings in the
particular cognitive environment of the audience addressed. All this is
achieved without recourse to typologies of texts, communication acts and
the like.
Note that the resulting notion of translation can also be distinguished
clearly from non-translation: as we saw in chapter 3, instances of descrip-
tive use across language boundaries would be excluded, as would instances
of interpretive use not involving two languages. 13 It would even offer the
basis for a principled distinction between translation and paraphrase, if one
defined paraphrases as those instances of interlingual interpretive use that
failed to confirm the presumption of optimal relevance, as discussed in some
of the examples above.
Furthermore, as I tried to point out briefly, it seems that an account
of translation as interlingual interpretive use is also able to do justice to
the historical dimension of translation addressed by Kelly in the follow-
ing words:

If a comprehensive theory be possible, it must seek the essential har-

mony between the practice of all ages and genres, and give a
satisfactory analysis of differences. (1979:227)

However, the very flexibility of this notion will no doubt be felt objec-
tionable by some who would not feel comfortable in allowing summaries as
well as elaborated versions to quality as translation. An advocate of this
opinion would be Newmark (1988), who sees such practices as instances of
'restricted translation' which fall outside the scope of translation theory
proper:

13 Of course, the notion of 'two languages' introduces a certain amount of fuzziness into

the concept, given the notorious sociolinguistic problems surrounding the notion of
'language' itself. However, note that this particular problem would not have any bear-
ing on the interpretation process or its result, since those are only determined by the
notion of interpretive use. In other words, while one may disagree as to whether a ren-
dering of a text from a Bavarian dialect in High German is a translation, on the grounds
that the two language varieties are perhaps viewed as 'dialects' rather than 'languages',
this does not make any difference to the comprehension ofthe contents ofthe 'transla-
tion' in High German, as long as the reader is a competent speaker of High German.
Translation as Interlingual Interpretive Use 129

There are also other restricted methods of translation: information

translation, ranging from brief abstracts through summaries to com-
plete reproduction of content without form. (1988: 12)

Newmark also lists here a wide variety of other kinds of "restricted trans-
lation", such as "plain prose translation (as in Penguins)", "interlinear
translation", "formal translation, for nonsense poetry (Morgenstern) and
nursery rhymes", and so forth (1988:12). Having presented this list, Newmark
concludes: "Translation theory, however, is not concerned with restricted
translation" (1988:12).
Intuitively there seems to be something right about the desire to distin-
guish between translations where the translator is free to elaborate or
summarize and those where he has to somehow stick to the explicit con-
tents of the original. Let us therefore consider whether relevance theory can
help us to explicate this intuition and perhaps provide a notion of transla-
tion that will do justice to it.
6. Translating What was Expressed

Style - The Importance of the Way Thoughts are

Expressed

As pointed out at the end of chapter 5, it may be felt by some that transla-
tion as interlingual interpretive use allows for too much variation. This may
be so especially for those who are interested not only in what the original
writer intended to convey, but also in how he conveyed it, that is, in the
style of the original.
This wider, stylistic dimension of communication is, of course, of special
interest to literary studies, and so it is not surprising that theorists concerned
with literary translation have paid considerable attention to the preservation
of the stylistic properties of texts:

Style is the essential characteristic of every piece of writing, the out-

come of the writer's personality and his emotions at the moment,
and no single paragraph can be put together without revealing in some
degree the nature of its author. (Savory 1957:54)

Without wanting to embark here on the adventure of defining 'style',

there is probably sufficient agreement on the fact that style is, in some sense,
the way the writer or speaker expresses himself - resulting, for example,
from the words he chooses or the way he constructs his sentences. 1
The following example from Savory, set up by him in terms of a com-
parison between "the Moderniser" and "the Helleniser", illustrates this
concern for the preservation of stylistic features in translation:

The Greek of st. Luke viii, 8, reads:

kai heteron epesen epi teen geen teen agatheen
[and other it-fell on the ground the good]

for which the Authorized Version has 'And other fell on good ground' .
This is a Moderniser's translation, giving good, plain English. A
Helleniser would feel that the characteristic way in which the adjec-
tive agatheen followed the noun geen was lost. st. Luke had not

1 I do not attempt to define the notion of 'style' because it seems to refer to a variety of

different properties of utterances that do not form a homogeneous domain. As we shall

see in this chapter, relevance theory deals with these various aspects without needing to
invoke a distinct notion of 'style'.
Translating What was Expressed 131

written epi teen agatheen geen, he had written words which had a
slightly different emphasis on the adjective - 'on ground which was
good', or something like this. A Helleniser would try to express the
distinction; whether he would succeed is another matter. (1957 :63f.,
gloss my own)

The preservation of such stylistic features is felt to be desirable because

it gives a more accurate representation of the original:

One of the reasons for a preference for a literal translation is that it is

likely to come nearer to the style of the original. It ought to be more
accurate; and any copy, whether of a picture or poem, is likely to be
judged by its accuracy. (Savory 1957:54, italics my own)

But this is where the problems begin, as Savory himself points out in the
very next sentence:

Yet it is a fact that in making the attempt to reproduce the effect of

the original, too literal a rendering is a mistake ... (1957:54f.)

Chukovskii's (1984) chapter 'Imprecise precision' is devoted to this very

dilemma: that the translator should be precise - but that the wrong kind of
precision is disastrous for the translation:

A precise, literal copy of a poetic work is the most imprecise and

false of all translations. The same can be said of translations of artis-
tic prose. (1984:49)

As an example, he cites the Russian translation of Dickens' Pickwick

Papers by Evgeny Lann:

although each line of the original text is reproduced with mathemati-

cal precision, not a trace has survived of Dickens , youthful, sparkling,
stormy hilarity. (Chukovskii 1984:51)

Unfortunately, despite its disastrous effects, it seems very difficult to

deal with this kind of wrong precision in translation:

Lexical discrepancies are very easily caught. If the original says 'lion'
and the translation has 'dog', it is obvious the translator made a mis-
take. But if the translator misconstrues not individual words and
132 Translation and Relevance

phrases, but the basic colouration of an entire piece, ifhe offers safe
and hackneyed verses instead of explosive, innovatively bold verses,
sugary phrases in place of ardent ones, halting instead of flowing
syntax - we are almost powerless to convince the ordinary reader a
fraud has been foisted on him. (Chukovskii 1984:48)

A little further on the author calls this kind of translation "slander", adding:

But there is nothing more difficult than exposing this kind of slan-
der, because it is stated not in words or phrases but in elusive tonalities
of speech for which no methods of definition have yet been worked
out. (Chukovskii 1984:48)

Quotations from the literature that address these very problems could be
multiplied, because this is one of the central questions ofliterary translation:
when faithfulness in matters not only of content but also of style is demanded
of a translator - in what terms should such faithfulness be measured or evalu-
ated? Can relevance theory as an explicit theory of communication provide
new insights into these problems?

Direct Quotation, Communicative Clues, and

Direct Translation

When we are concerned with preserving not only what someone meant, but
also the way it was expressed we seem to be touching on the difference
between direct and indirect speech quotations in intralingual communica-
tion; after all, direct speech quotations preserve exactly what was said,
whereas indirect speech quotations give an indication of what was meant.
Indirect speech quotation seems to fall naturally under interpretive use, and
so in a way we might say that the account of translation as interlingual in-
terpretive use considered in the last chapter is the interlingual parallel to
indirect quotation. By analogy, then, what we are looking for now is an
interlingual parallel to direct quotation. 2
2 Mossop (1987) uses the difference between direct and indirect speech quotations, in
his terms "direct-discourse reports" versus "indirect-discourse reports" to differentiate
"standard-form translation" from other, freer kinds of translation; for Mossop the es-
sential similarity between "direct-discourse reports" and "standard-form translation" is
that "standard-form written translations are like directly reported discourse as regards
viewpoint' (1987:9, italics as in original). He explains: "The meaning ofthe word T in
a translation is important, and in standard-form translation, T is a stand-in for the source
Translating What was Expressed 133

One of the first issues to discuss here is, of course, what direct speech
quotations are.
It seems that the notion of direct speech quotation can be derived from
the nature of stimuli used in communication; in relevance theory such stimuli
are defined as follows:

A stimulus is a phenomenon designed to achieve cognitive effects.

(Sperber and Wilson 1986a:153)

N ow phenomena here are understood to be "perceptible objects or events"

(Sperber and Wilson 1986a:40), and they achieve the desired cognitive ef-
fects in virtue of the properties they have.
This means, then, that stimuli can be looked at from two different points
of view. They can be looked at from the point of view of the cognitive
effects they have - for example, what explicatures and/or implicatures they
convey - but they can also be looked at from the point of view of the intrin-
sic properties they have as phenomena, and this is the perspective that seems
to be relevant to direct quotation, for, as Wilson and Sperber (1988a) state,
"Direct quotations are chosen not for their propositional form but for their
superficial linguistic properties" (1988a:137).
Thus, whereas indirect quotations depend on resemblance in cognitive
effects, direct quotations depend on resemblance in linguistic properties.
When a speaker produces a direct quotation, he reproduces the original stimu-
lus with its various linguistic properties:

(1) (a) Bob: 1 will be there at five o'clock exactly.

(b) Margaret to Jane: When did Bob say he would come?
(c) Jane: He said he'd come at five.
(d) Jane: He said, "I will be there at five o'clock exactly".

Here (1) (c) is obviously an indirect quotation, resembling the original (1)
(a) in its propositional content and implicatures but differing almost com-
pletely in actual linguistic properties; in (1) (d), however, Jane reproduces
the original with all its linguistic detail: the same syntactic construction, the
same semantic representation, the same lexical items, and so forth. 3

text narrator; 'I' is not the translator" (1987:18, italics as in original). Our relevance-
theoretic account of direct translation naturally keeps the 'viewpoint' of the original
and preserves other features in addition.
3 Note that some phonetic variation, at least, is generally permissible; for example, one

does not usually reproduce idiosyncratic or dialectal peculiarities of pronunciation,

134 Translation and Relevance

Returning to translation, the obvious problem is that in translation we

need to talk about resemblances between texts and utterances that belong to
different languages. While there is a certain consensus that it is often possi-
ble to achieve a fairly good degree of resemblance in semantic representation
across languages, the same cannot be said of stylistic properties, which of-
ten consist in linguistic features that are far from universal. Thus it would
be impossible for languages that lack the phonological property of focal
stress or the syntactic category of passive to resemble utterances of English
in these properties. As, for example, Mossop (1987) has observed, verbatim
reporting "is of course impossible by definition in the case of translation"
(1987:20, n. 4).
However, one might well argue that the point of preserving stylistic prop-
erties lies not in their intrinsic value, but rather in the fact that they provide
clues that guide the audience to the interpretation intended by the commu-
nicator. We shall refer to such clues as communicative clues.
Taking up the example of focal stress, its communicative value is that it
draws the audience's attention to that part of the utterance that is most rel-
evant, that is, intended to make the greatest contribution to contextual effects.

(2) (a) The DEALER stole the money.

(b) The dealer STOLE the money.

In English it does so in virtue of the fact that "different stress assignments

induce different focal scales" (Sperber and Wilson 1986a:208) - that is,
they determine the order in which the analytic 'background' implications
are considered, where a 'background' implication is an analytic implication
obtained by replacing the focal constituent with a semantic variable. 4 Thus
the first 'background' implications of (2) (a) and (b) respectively are the
following:

(3) (a) Someone stole the money.

(b) The dealer did something to the money.

In Sperber and Wilson's (1986a) account, this function of focal stress in

English is not linguistically encoded but can be explained in terms of reduc-
ing processing effort. However, as Sperber and Wilson state, in other

though on occasion this may be desirable. This seems to indicate that even direct quota-
tions are subject to considerations of relevance.
4 For a detailed account ofthis see Sperber and Wilson (l986a:202-2l7).
Translating What was Expressed 135

languages the same function may well be "taken over by some purely lin-
guistic device, syntactic, morphological or intonational ... " (1986a:262, n. 16).
So for example, while in some of the Ethio-Semitic languages these fo-
cal effects cannot be achieved by stress, syntactic means like clefting can
be used for this purpose, as in these sentences from the Silt'e language 5 :

(4) (a) farankaay yasarak'ay naggaadeenii

money-OBJ-DEF REL-he-stole-it dealer-is-DEF
'It is the dealer who stole the money.'
(b) naggaadeey yaasheey farankaay sirak'ootin
dealer-DEF REL-he-did-it money-OBJ-DEF to-steal-is
'What the dealer did is stealing the money.'

In temporal sequence, the front-shifted relative clause would be pro-

cessed first, providing the following assumption schema for (4) (a):

(5) Someone stole the money.

Hence the clefting in (4) would have the effect of making the same analytic
implication most accessible as focal stress in the English example (2) (a),
that is, it would provide the same clue for interpreting the Silt' e sentence as
focal stress does in English. The same would apply to (4) (b).
If this were true more generally, then this notion of 'clue giving' would
allow us to define translation along lines parallel to direct quotation: as
direct quotation calls for the preservation of all linguistic properties, so this
kind of translation calls for the preservation of all communicative clues.
Furthermore, again paralleling direct quotation, by preserving all the com-
municative clues of the original, such translation would make it possible for
the receptors to arrive at the intended interpretation of the original, pro-
vided they used the contextual assumptions envisaged by the original author.
Thus in the context of our current discussion, this notion of translation
seems to be attractive from two points of view. First of all, it seems to pro-
vide a fixed rather than flexible concept of translation. Secondly, it seems
very suitable for the preservation of stylistic features in that communicative
clues reflect not only the information content of what was said, but also the
way in which it was expressed and the special effects that such stylistic
features would achieve.

5 Doubly written letters represent phonemically long segments; Ik'i stands for the velar

ejective. For further details on the phonology of the language see Gutt (1983). DEF
represents the definite marker, OBJ the object suffix, and REL the relative clause marker.
136 Translation and Relevance

In view of the close correspondence between this notion of translation

and direct quotation, we shall refer to it as direct translation. By analogy,
we shall call the concept of translation defined in the previous chapter as
interlingual interpretive use indirect translation.
So let us now investigate in more detail what kind of phenomena could
be covered by the notion of 'communicative clue' and what light it can
throw on the preservation of style in direct translation.

Communicative Clues Arising from Semantic

Representations

According to Sperber and Wilson (1986a), "Verbal communication proper

begins when an utterance ... is manifestly chosen by the speaker for its se-
mantic properties" (1986a:178). From what we have said so far, then, one
of the first concerns of the translator aiming at resemblance between stimuli
will be to consider the semantic representation of the original stimulus. This
might seem a rather common-place statement - do not all approaches to
translation stress the importance of looking at the semantics of the text?
Two points need to be made here. Firstly, it is not necessarily true that
all translation gives first priority to achieving resemblance in semantic in-
terpretation. The 'phonemic translation' of poetry discussed by Lefevere
(1975) is a case in point. The example Lefevere discusses is a translation of
Catullus by C. and L. Zukofsky (1969):

Celia and Louis Zukofsky's translation of Catullus opens with the

following preface: "This translation of Catullus follows the sound,
rhythm and syntax of his Latin - tries, as is said, to breathe the 'lit-
eral' meaning with him." Fidelity to the source text means, purely
and simply, fidelity to its sound, to the near exclusion of all other
elements. (Lefevere 1975:19)

This approach leads to renderings like the following:

"T'my Thetis this Peleus incandesced fair thru his armor" corresponds
to "tum Thetidis Peleus incensus fertur amore". (Lefevere 1975:20).

While this is a rather extreme and exceptional approach, there are other,
more normal instances where the translator may feel compelled to disregard
faithfulness in semantic representation in favour of other factors. Rhymed
poetry is one example:
Translating What was Expressed 137

Rhyme imposes a constraint upon the writer, a constraint which bears

most heavily on the essential feature of the translator's art, his choice
of words. It is scarcely possible to find a rhymed translation of a
lyric which does not contain evidence of this as shown either by the
omission of something that the original author wrote, or the inclu-
sion of something that he did not. (Savory 1957: 85)

Savory gives the following example:

Translating a Welsh poem, 'Hen Benillion,' [Mr. Oliver Edwards]

wrote the line

And still the crow feeds by the shore

and in discussing his work, admits "some liberty" with the crow, which
in the original was tending its nest "and won't rhyme in English".
(Savory 1957:85)

Translation of works where the playful use oflanguage is important also

tends to subordinate semantic resemblance to other kinds of resemblances.
One example was Morgenstern's nonsense poem about the 'aesthetic wea-
sel' (cf. pp. 111-2), but numerous examples can be found in translations of
comics, for instance in Goscinny and Uderzo's Asterix series. In his study
of this series, Grasegger (1985) states:

The invariant element in such a transfer is evidently not the specific

form nor the content, but the idea of a play on words, in favour of
which one is often satisfied with a translation only partially equiva-
lent in content. (1985: 100, translation my own)

Here is an example:

(6) (a) Original: "Je me demande si nous sommes du bois dont on

fait les heros?"
[Literal translation (my own): I wonder whether we are of
the wood of which one makes heroes?]
(b) English translation: "I'd as lief not be here either, old bean."6

6 R. Goscinny and A. Uderzo, Le Combat des Chefs; English translation: Asterix and
the Big Fight; reported by Grasegger (1985:70). The utterances in question belong to a
picture where some Roman soldiers have camouflaged themselves as bushes.
138 Translation and Relevance

The English translation here does not even show partial overlap in content.
The second point to be made regarding faithfulness to the semantics is
that semantic representations cannot be equated with 'the meaning' of an
utterance:

Linguistically encoded semantic representations are abstract mental

structures which must be inferentially enriched before they can be taken
to represent anything of interest. (Sperber and Wilson 1986a: 174)

As pointed out in chapter 2 (pp. 25-6), the semantic representation of an

utterance forms an assumption schema that needs to be developed inferen-
tially until it yields the propositional form of the utterance that can be
evaluated as true or false of some state of affairs. Thus the usefulness of
these semantic representations is that they serve as 'a source of hypotheses'
about the communicator's intention - that is, they provide communicative
clues.
One problem for the translator is that "semantic representations of sen-
tences are mental objects that never surface to consciousness" (Sperber and
Wilson 1986a:193). What we can be aware of as human communicators is
the content of the fully developed propositional form, but not of the seman-
tic representation from which it was derived - and usually we are aware not
only of the propositional form expressed by an utterance, but also its
implicatures. In short: we are not usually aware of the communicative clues
provided by the semantic representation of an utterance, but only of its in-
terpretation as a whole.
If this is true - does it not mean the end of this approach to transla-
tion? How can the translator attempt to identify entities of which he cannot
become aware? The answer is that while it is true that the semantic repre-
sentations themselves never surface to consciousness, hence are not open to
direct introspection, this does not necessarily mean that the translator can-
not know anything about them. There are many phenomena, for example,
in the physical world that we cannot have a direct awareness of. Yet we can
acquire knowledge about them by drawing inferences from related phenom-
ena of which we can be aware. In the same way, it is possible in principle to
use our awareness of propositional forms to draw inferences about proper-
ties of the underlying semantic representations. And if relevance theory is
right, this is, in fact, the only way for semanticists to develop their theories. 7
However, because we cannot necessarily equate our intuitions about the

7 Gutt (1987b) explores the possibilities for developing theories of linguistic semantics.
Translating What was Expressed 139

meaning of an utterance with the communicative clues it provides via its

semantic representation great care is required in this area. Thus while we
may feel that a translation somehow differs from the original, it may not be
obvious whether this difference is due to a misrepresentation of its commu-
nicative clues or perhaps a mismatch in contextual assumptions used to
interpret it. The following example may serve to illustrate this point.
Yuasa discusses the problems of translating into English a haiku by
Basho. 8 Concerning the opening word of the poem he says this:

The first word in the text isfuruike, which is a compound noun con-
sisting of the adjective furushi meaning 'old' and the noun ike
meaning 'pond'. If a translator could satisfy himself with this simple
explanation and end the whole matter by saying' an old pond', his
job would be easy. Somehow, however, I found this English equiva-
lent to furuike unsatisfactory, and I wondered why. First, I sensed
that the English equivalent was too weak, far too abstract and gen-
eral, to convey the landscape suggested by the original word. It is
true that the original word itself is not so precise in its representa-
tional quality; it is in fact far from 'the direct treatment of the thing' .
Nevertheless, I thought it was the responsibility of a translator to say
more than 'an old pond' to give a little more sense of the presence of
the poet by the pond. (1987:233f., italics my own)

Yuasa's evaluation illustrates the ability to make comparative judgements

about the interpretations of stimuli; note that he seems to realize that the
meaning missed out was not actually part of the meaning of the original
word. It also illustrates the problem of distinguishing between these differ-
ent kinds of 'meaning' in a generally applicable, yet explicit way. The general
terms he uses - "too weak, far too abstract and general" - are too vague to
identify the problem, and the specific expressions he uses address the prob-
lem in terms particular to this poem and to the situation it refers to.
From the point of view of relevance theory, it is fairly clear that the
problem Yuasa is grappling with here is the fact that the meaning commu-
nicated by a text is not attributable to the stimulus alone, but results from
the interaction between stimulus and cognitive environment. The reason
why the average English reader will miss the italicized information - the
impressions of the landscape of a Japanese garden and of the presence of

8The text ofthis haiku can be found in the anthology Haru no Hi, 'Spring Days'. See
(Yuasa 1987:231).
140 Translation and Relevance

the poet by the side of an ancient pond - is not necessarily that the expres-
sions are semantically different; it seems more likely that it will be lost
because that original scenery is not part of an English person's cognitive
environment - he may never have seen it nor heard of it, and so cannot use
it to enjoy the rich interpretation it made available to the original audience.
It is interesting to note that this failure to see the interpretation as result-
ing from the appropriate combination of stimulus and cognitive environment
leads Yuasa to a view of translation rather similar to that of the communica-
tive approaches. Thus when he thinks that "it was the responsibility of a
translator to say more than 'an old pond' to give a little more sense of the
presence of the poet by the pond" (Yuasa 1987:234), he is, in effect, sub-
scribing to the view that a translation should communicate the
author-intended meaning regardless of differences in background knowl-
edge between original and receptor language audience.
His solution, too, resembles the 'explication' strategy of the communi-
cative approaches:

To give a greater sense of presence by the pond, I decided to add the

word 'silence' ... it suggests by implication the presence ofa listen-
ing ear. (Yuasa 1987:234)

Yuasa's discussion illustrates well the problem of sorting out 'meaning' -

a problem only too familiar to translators. What is particularly interesting
for our discussion is his awareness that there are two different 'kinds' or
'degrees' of meaning. There is the meaning represented by the English ex-
pression an old pond - but then there seems to be another kind of meaning
that, as Yuasa himself feels, "is not so precise in its representational qual-
ity" but "is in fact far from the direct treatment of the thing", and that is
what he describes as "the landscape suggested by the original word" and "the
presence of the poet by the pond" (Yuasa 1987:234).
Here is one point where relevance theory can offer the translator signifi-
cant help. It can not only make him aware of the pervasive influence of the
cognitive environment on the interpretation of the text, but also help him to
distinguish between the different aspects of meaning and thus to get some
orientation in the 'jungle of meaning' that seems so confusing and over-
whelming at times.
In terms of the example considered, the translator could first of all try to
compare the original expression furuike and, for example, its translation
'an old pond' in terms of their respective semantic properties. As pointed
Translating What was Expressed 141

out above, this is no trivial matter, but one that has to proceed by inference
from the full interpretation to the characteristics of the semantic representa-
tions that gave rise to them. And though it is probably true to say that we are
still a long way off an adequate theory of semantic representations, the more
explicit and closer to our mental reality the semantic framework the transla-
tor uses, the better his prospects for an adequate understanding ofthe problem
and its possible solutions.
The assumptions made by relevance theory about the nature of concepts
provide a promising starting point for such an explicit and empirical frame-
work. Thus relevance theory assumes that the concepts in our mind are
associated with three sets of information, or 'entries': a logical entry, an
encyclopaedic entry and a lexical entry:

The logical entry for a concept consists of a set of deductive rules

which apply to logical forms of which that concept is a constituent.
The encyclopaedic entry contains information about the extension
and/or denotation of the concept: that is, about the objects, events
and/or properties which instantiate it. The lexical entry contains in-
formation about the natural-language counterpart of the concept: the
word or phrase of natural language which expresses it. (Sperber and
Wilson 1986a:86)

According to relevance theory, at least part of the semantic meaning of a

concept is stored in its logical entry, and consists, in fact, of the deductive
rules - or meaning postulates - it contains. For example, Sperber and Wilson
(l986a) suggest that the concept associated with the English word mother
has a logical entry which contains the following deductive rule (or meaning
postulate) :

Mother-elimination rule
Input: (X-mother-Y)
Output: (X-female parent-Y)
(Sperber and Wilson 1986a:90)

The crucial point for our discussion is that these deductive rules that
give the semantic meaning of a word apply automatically: the mere pres-
ence of a concept in semantic representation is sufficient for the hearer to
apply all of its meaning postulates, regardless of context. This accounts for
the fact that the word mother in English intrinsically denotes a female par-
ent. Furthermore, logical entries are assumed to be "small, finite and
142 Translation and Relevance

relatively constant across speakers and times", and it is the state of the logi-
cal entry which reflects whether an individual has grasped a certain concept:
"There is a point at which the logical entry for a concept is complete, and
before which one would not say that the concept had been mastered at all"
(Sperber and Wilson 1986a:88). Thus the logical entry contains informa-
tion essential to that concept.
By contrast, the encyclopaedic entry contains all sorts of information
that is incidental to the concept: thus the assumption that mothers do a lot of
housework might be part of the encyclopaedic entry associated with the
concept 'mother'. However, this information would not be accessed by au-
tomatic deductive rules, hence would not be a necessary part of this concept.
Encyclopaedic entries are assumed to be open-ended, allowing for the con-
stant addition of new information; none of the information they contain is
essential for mastery of the concept, nor is there a point at which an ency-
clopaedic entry could be said to be complete.
Furthermore, what is important for our discussion here is that logical
entries and encyclopaedic entries play distinctive roles in the process of
utterance interpretation; thus "the content of an assumption is determined
by the logical entries of the concepts it contains, while the context in which
it is processed is, at least in part, determined by their encyclopaedic entries"
(Sperber and Wilson 1986a:89). Thus the distinction between logical and
encyclopaedic entries essentially corresponds to the distinction between the
content and the context of an utterance.
Returning to our example, the translator could examine his intuitions
about the meaning of furuike in terms of a concept with logical, encyclo-
paedic and linguistic entries; he could ask himself what meaning elements
might be part of what entry. Thus he could try to determine whether the
information about "the presence of the poet by the pond" (Yuasa 1987:
234) is part of the logical entry of the concept associated withfuruike or of
its encyclopaedic entry. If part of the logical entry, then, given the assump-
tions about logical entries just considered, it should be true, for example,
that any speaker of Japanese who did not recognize that this word has 'the
presence of a poet by a pond' as part of its 'meaning' had not, in fact, mas-
tered that concept at all.
Not knowing Japanese, I do not know what the right answer is. How-
ever, considering Yuasa's own judgement that "the original word itself is
not so precise in its representational quality", it seems more likely that here
we are dealing with the kind of information that is typically part of the en-
cyclopaedic entry of a concept: that is, information in some way associated
Translating What was Expressed 143

with that concept, but not an integral part of it.

Similarly, one would suspect that the impression of a certain type of
landscape that Yuasa gets from the expressionfuruike stems from the ency-
clopaedic entry, perhaps from a stereotype stored there, rather than from its
logical entry.
Whatever the concrete outcome for any given word, it will be of impor-
tance for the translator. The properties of a semantic representation are, at
least in part, determined by the logical entries of the concepts it contains.
Therefore, if resemblance in communicative clues, hence resemblance in
semantic representation, is attempted, then clearly the translator must pay
attention to these matters of semantics. Alternatively, if the information in
question stems from the encyclopaedic entry of a concept that appears in
the semantic representation, then it is part of the context and so, under the
approach we are currently discussing, would fall outside the scope of a
translation concerned with the preservation of communicative clues. It would
be the responsibility of the audience to familiarize themselves with such
information.
It seems worth pointing out that such a relevance-theoretic account has
the potential of meeting one of Steiner's most serious criticisms of transla-
tion theory:

the debate over the extent and quality of reproductive fidelity to be

achieved by the translator has been philosophically naive or fictive.
It has postulated a semantic polarity of 'word' and 'sense' and then
argued over the optimal use of the 'space between' ... The theory of
translation ... ought not to be held to account for having failed to
solve problems of meaning, of the relations between words and the
composition of the world ... The fault, so far as the theory goes, con-
sists of having manoeuvred as if these problems of relations were
solved or as if solutions to them were inferentially obvious in the act
of translation itself. Praxis goes ahead, must go ahead as if; theory
has no licence to do so. (Steiner 1975:277f., italics as in original)

While it is true that the proposals made by Sperber and Wilson are "specu-
lative" in some respects and "the boundaries between logical and
encyclopaedic entries are not always easy to draw" (l986a:93), the impor-
tant point is that these issues have now been taken out of the realm of the
philosophical and been assigned a place in cognitive science. The question
of whether or not a certain piece of information is stored in the logical or
encyclopaedic entry is an empirical one which, at least in principle, has an
empirical answer.
144 Translation and Relevance

Furthermore, Sperber and Wilson make clear that their account "though
speculative, is as far as we know compatible with the available empirical
evidence" (1986a:85).9 Thus the distinction between logical and encyclo-
paedic entries is bound up closely with the distinction between representation
and computation - a distinction on which "the whole framework of current
cognitive psychology rests ... " (1986a:89). So, despite its tentative nature,
the claim that there are distinct logical and encyclopaedic entries is not 'philo-
sophically naive or fictive', but bears the mark of a serious scientific claim.
It is compatible with scientific insights we currently have about the mind,
and it is open to empirical falsification.

Communicative Clues Arising from Syntactic

Properties

Interlinear translations are probably the clearest examples of the concern to

preserve certain syntactic properties of the original, such as word order and
syntactic categories, with as little change as possible. tO
However, concern for the preservation of syntactic properties is not lim-
ited to this rather special form of translation. According to Lev)!, "the literary
translator is concerned with equivalents that have as many common denomi-
nators with the original as possible" (1969: 17, translation my own).
Accordingly, the table he compiles to what properties should be kept invari-
ant in different types of texts indicates that, among other properties, sentence
structure is to be considered invariant in the translation ofliterary prose and
drama, blank and rhymed verse, musical texts and dubbing (1969:19).
The following example, involving the opening passage of Dickens' A
tale of two cities, may serve to show the potential importance of sentence
structure in literary translation.

It was the best of times, it was the worst of times, it was the age of
wisdom, it was the age of foolishness, it was the epoch of belief, it
was the epoch of incredulity, it was the season of light, it was the
season of darkness, it was the spring of hope, it was the winter of
despair, we had everything before us, we had nothing before us, we
were all going to heaven, we were all going direct the other way ...

9 For further comments on the speculative nature of the current stage of development in

relevance theory see Sperber and Wilson (1987), esp. pp. 709ff.
10 I shall say more about interlinear translations in chapter 7, pp. 177-8.
Translating What was Expressed 145

On this passage Chukovskii comments:

There is an almost poetic cadence in this excerpt. The sound sym-

metry conveys its ironic tone extremely well. (1984: 144)

By contrast, he feels that the following translation misses these effects:

It was the best and worst of times, it was the age of wisdom and
foolishness, the epoch of unbelief and incredulity, the time of en-
lightenment and ignorance, the spring of hope and the winter of
despair. 11

Chukovskii feels that the problem is that "[the translators] did not catch the
author's intonations and thus robbed his words of the dynamism stemming
from the rhythm" (1984:144). Chukovskii apparently attributes the special
effect achieved by the original to such phonological properties as 'sound
symmetry' and 'rhythm'. While it seems unlikely that the 'ironic tone' and
the 'dynamism' are due to actual phonological characteristics, we can give
an explicit account of these effects if we pay attention to the syntactic pat-
terns involved.
Let us begin with Chukovskii's observation that there is an ironic qual-
ity to this passage. As Sperber and Wilson (1986a) have shown, irony is an
instance of the echoic use of utterances, that is, of instances of interpretive
use where utterances that attribute thoughts to others "achieve relevance by
informing the hearer of the fact that the speaker has in mind what so-and-so
said, and has a certain attitude to it ... " (1986a:238, italics my own). This
seems to be the case in the passage cited. Dickens informs his readers that
he has in mind a number of evaluations that people have had of the late 18th
century - and crucially that he has a certain attitude to them, or at least to
some of them, an attitude perceived by Chukovskii as 'ironic.'
Why does the translation cited not achieve this ironic effect? This has
largely to do with a change in the sentence structure: while Dickens uses a
string of sentences in juxtaposition, the translators combined these pair-
wise into single sentences with conjoined complements. Innocuous as this
syntactic change may seem, it makes a great difference to the interpretation
of the text. Note that in both versions the reader is faced with a series of
contradictory statements. In terms of the information processing that goes
on in our minds, such contradictions pose a problem, and given the fact that

11 S. P. Bobrov, and M. P.Bogoslovskaja (1957), Pavest' a dvukh garodakh, Sobranie

sochinenii, Moscow, (1957-63; vol XXII; p. 6).

146 Translation and Relevance

the contradictions are obviously deliberate, the audience will look for some
way of resolving them. At this point the fact that Dickens presents these
contradictions in the form of pairs of independent sentences provides im-
portant communicative clues. It allows the reader to view each sentence as
an independent statement - which allows for the possibility that each might
be 'echoing' the opinion of a different group of people, that is, might be
intended to represent what different people thought about those times; this
interpretation shows that Dickens is not contradicting himself, and makes it
possible to recover the ironic overtones.
By contrast, the form of syntactic coordination used by the translators
fails to provide the clues necessary for such an echoic interpretation: the
conjoined assertion "It was the best and worst of times" cannot readily be
interpreted as a reflection of the conflicting opinions "It was the best of
times" and "It was the worst of times" held by two different individuals or
groups of individuals. As a result, the translation loses the ironic flavour of
the original. Thus here we have an example of how the change of syntactic
characteristics in the translation can lead to the loss of subtle, but neverthe-
less important clues to the intended interpretation.
In the example just looked at, these communicative clues could prob-
ably be preserved quite simply in most cases by preserving the relevant
syntactic properties themselves, that is, by using a series of non-conjoined
sentences. However, due to the linguistic differences between languages,
matters are not always as straightforward. Savory's example used at the
beginning of this chapter (p. 130) is a case in point. It is of particular inter-
est because scholars have different judgments about what effects are involved
here. For convenience, I reproduce the example here.

kai heteron epesen epi teen geen teen agatheen

[and other it-fell on the ground the good]
(Luke 8:8)

This sentence is interesting because of the syntactic peculiarity that

the adjective agatheen 'good' is placed following the head of the noun
phrase rather than preceding it. This structure is grammatical in Greek,
but not in English, and hence cannot be imitated as such without violating
English grammar. More importantly, apart from being ungrammatical, such
a literal imitation of the Greek structure in English would not have the
effect of emphasizing the adj ective agatheen, an effect which Savory senses
in the original:
Translating What was Expressed 147

St. Luke had not written epi teen agatheen geen, he had written words
which had a slightly different emphasis on the adjective - 'on ground
which was good', or something like this. (Savory 1957:63f; cf. pp.
130-1 above, reproduced here in part for convenience)

This view that the postposition of the adjective is likely to convey em-
phasis is shared by a number of other scholars (e.g. Reiling and Swellengrebel
1971; Winer 1882).
However, before pursuing this special effect further, we should briefly
take note of other scholars who differ from Savory in this intuition. Blass
and Debrunner (1961), for example, claim that in constructions where the
adjective is postposed the emphasis is "more on the substantive (eis teen
geen teen agatheen Lk. 8:8, in contrast to petran etc.)"(196l:l4l).
Plummer (1922) does not refer to emphatic effects at all, but only notes
that "The double article in all three accounts presents the soil and its good-
ness as two separate ideas: 'the ground (that was intended for it), the good
(ground)'" (1922:219). Similarly Turner (1976) seems to see no special sig-
nificance in the postposition of the adjective in Biblical Greek:

The position of participial and adjectival phrases, qualifying an ar-

ticular noun, is regularly between article and noun in non-Biblical
Greek, unless there is a special reason. However, in Jewish Greek
the tendency is to place the adjectival phrase after the noun, as in
Semitic languages, with the article repeated. (1976:110)

Let us start with Turner's point, trying to see why the position of the
adjective would be important for the interpretation of this (and other) utter-
ances. It seems intuitively clear that if the postposition of the adjective is
nothing special in biblical Greek, then we should not give its occurrence
any special attention. Relevance theory can explain this intuition.
As we saw in chapter 2, (p. 32), the principle of relevance establishes a
cost-benefit correlation between the effort needed to process a stimulus and
the contextual effects to be expected as reward. A communicator can ex-
ploit this correlation for special contextual effects, as Sperber and Wilson
(1986a) have shown for utterances that involve repetition:

the task of the hearer faced with these utterances is to reconcile the
fact that a certain expression has been repeated with the assumption
that optimal relevance has been aimed at. Clearly, the extra linguistic
processing effort incurred by the repetition must be outweighed by
148 Translation and Relevance

some increase in contextual effects triggered by the repetition itself.

(1986a:220)

Put in general terms: if a communicator uses a stimulus that manifestly

requires more processing effort than some other stimulus equally available
to him, the hearer can expect that the benefits of this stimulus will outweigh
the increase in processing cost - otherwise the communicator would have
failed to achieve optimal relevance.
One of the factors that affect processing cost is the structural complexity
of the stimulus. Another is frequency of use: the more commonly a certain
type of structure occurs in stimuli, the less processing effort it seems to
reqUIre.
It is in relation to this latter point that Turner's observation becomes
important. Ifin the Greek of Luke and his audience the postposed adjective
was quite a frequent or even the normal construction, then its use would not
involve a perceptible increase in processing cost, and hence would not give
rise to expectations of increased contextual effects. Turner does give some
statistics, but unfortunately for Luke they cover only the first two chapters
(which are believed by many scholars to show particularly strong Semitic
influence in the Greek), and so it is difficult to establish whether his point is
valid for the passage in question or not.
Since our main interest is not in the particular interpretation of this text,
but in an explicit account of how syntactic structure can be used to provide
communicative clues to the interpretation of utterances, let us assume that
the scholars who do perceive here a clue to some special effect are right. In
this case, can relevance theory help us to understand why there should be
such seemingly divergent intuitions about what this effect is? I believe that
relevance theory enables us, in fact, to see that these intuitions are not nec-
essarily contradictory, but more likely express different aspects of a
somewhat complex set of effects.
Let us take our clue from Blass and Debrunner's remark (1961) that
there is more of an emphasis on the noun, that is, geen 'soil', "in contrast to
petran etc." As will be recalled, the 'Parable of the Sower', of which this
utterance is a part, describes how in the process of sowing parts of the seed
fall into different locations:

(7) (a) ho men epesen para teen hodon ...

'some fell by the wayside'
Translating What was Expressed 149

(b) kai heteron katepesen epi teen petran ...

'and other fell upon a rock ... '
(c) kai heteron epesen en mesoo toon akanthoon
'and other fell among thorns ... '
(d) kai heteron epesen eis teen geen teen agatheen
'and other fell on good ground ... '

As is obvious, these expressions are remarkably parallel in their syntac-

tic, semantic and phonological structure. Sperber and Wilson (l986a) have
pointed out that such structural parallelisms "reinforce the hearer's natural
tendency to reduce processing effort by looking for matching parallelisms
in propositional form and implicatures" (l986a:222). So having processed
the first three expressions (a)-(c), the hearer may have strong expectations
that (d) will have the same structure, and particularly he may be likely to
expect the most relevant part of the fourth expression to be found again at
the same point as in the preceding expressions, that is, in the head of the
prepositional phrase, which in each case very concisely indicates a new
location: hodon 'wayside', petran 'rock', and akanthoon 'thorns'.
With this strong expectation of parallelism, an adjective placed between
the article and the head noun may easily be given less attention than it would
otherwise have; it might possibly be overlooked. But if the communicator
wants to draw attention not only to the fact that the fourth part of the seed
fell on soil, but also that it was good soil, then he would need to make the
adjective more noticeable - and using the adjective in postposition would
seem a very suitable clue for achieving that effect. It gives the adjective
more prominence by placing it in a phrase of its own, without, however,
breaking the parallelism in the overall structure of the sentence.
Ifthis analysis is right, then this sentence is indeed an example of a very
effective use of language. By maintaining parallel structures in all four
sentences, the communicator keeps the focus on the different locations
throughout, and by postponing the adjective in the fourth sentence, he gives
it the desired degree of prominence at the same time.
Now we can also account for the apparently conflicting intuitions of
Savory on the one hand and Blass and Debrunner on the other: Savory
reacts to the relative prominence of the adj ective - Blass and Debrunner
react to the fact that the postposition of the adjectives allows the noun
geen to retain the expected focal position in the expression, which puts it,
as Blass and Debrunner perceive it, 'in contrast' to the other three loca-
tions mentioned.
150 Translation and Relevance

Communicative Clues Arising from Phonetic

Properties

Let us now tum to phonetic properties, and their written or graphological

counterparts. They typically serve to convey how a word is pronounced or
spelled, and this can be an important clue in translation when it comes to
the rendering of proper names. Thus it can be relevant to know what an
individual is called: it makes reference to that individual easier and at times
it may provide additional clues about him or her, for example as regards
nationality. Lev)!, for example, recognizes transcription as one of three
"working procedures of translation" the translator can utilize (1969:88, trans-
lation my own) under certain conditions with names and also with certain
kinds of onomatopoeic expressions.
At the same time, the relationship between translation and transcription
remains somewhat unclear. On the one hand, transcription is seen as a le-
gitimate "working procedure of translation", but on the other, it is treated as
clearly distinct from translation, as can be seen from the following passage
from Levy on the treatment of onomatopoeic expressions:

Translation is possible when the onomatopoeic sound sequence ob-

tains conceptual values or word character, as is the case with the
'language' of domestic animals and the most common sounds of
nature. However, it is not possible to translate or substitute sound-
imitating sequences that constitute a unique imitation of a sound of
nature created ad hoc. Here phonetic transcription alone is
possible.(1969:90f., translation my own)

Sometimes Levy uses the expression "translation in the true sense of the
word" (1969:91, translation my own) to contrast it with transcription. This
conceptual problem is not peculiar to Levy - consider, for example, the
following statement by Newmark:

In theory, names of single persons or objects are 'outside' languages,

belong, if at all, to the encyclopaedia not the dictionary, have ... no
meaning or connotations, are, therefore, both untranslatable and not
to be translated. (Newmark 1988:70)

Rather the problem is general in the sense that virtually all translation
work requires transcription at one point or another, and yet transcription
seems to be different from translation.
Translating What was Expressed 151

In the approach suggested here, transcription causes no special concep-

tual problems because it involves genuine properties of the original that can
contribute to relevance.
At the same time, transcription of names remains clearly distinct from
'translation proper' because each measure arises from a set of properties
clearly distinct from the other - that is, phonological (or graphological) ver-
sus semantic ones.
Before leaving the topic of the treatment of proper names, it seems worth
pointing out that our approach can also account straightforwardly for cases
where proper names are not transcribed but translated on the basis of such
semantic meaning as they might have; this is to be expected where that
semantic meaning may be felt more relevant than the phonological form of
the name - as is the case e.g. with names intended to give evidence of the
person's character. For example, in a German translation of Sheridan's
School for scandal, Careless has become Ohnsorg (Stuttgarter Ubersetzung,
cited in Levy 1969:87f.). However, this brings in the potential need for se-
lecting which of a range of possible clues a translator should intend to
preserve - a problem to which we shall return in chapter 7.

Communicative Clues Arising from Semantic

Constraints on Relevance

We have already looked at the requirement that resemblance in communi-

cative clues should involve resemblance in semantic representation between
the translation and the original. However, there is other information that is
linguistically encoded and yet not included in the semantic representation -
not because this information is contextual, but rather because it is not truth-
conditional at all. One group of words that fall under this category are
'pragmatic connectives' (cf. e.g. Blakemore 1987, Blass 1988 and 1990,
Gutt 1988a), perhaps better-known by some as 'discourse markers'.
So let us look at 'pragmatic connectives', and see how resemblance in
the clues they provide can be important for translation, in this instance for
an English translation of a German original.
The German original is a moral anecdote by Peter Hebel; the part that
interests us is the following:

2 Man findet gar oft, wenn man ein wenig aufmerksam ist, dass
Menschen im Alter von ihren Kindem wieder ebenso behandelt
152 Translation and Relevance

werden, wie sie einst ihre alten und kraftlosen Eltem behandelt
haben.
2 Es geht auch begreiflich zu.
3 Die Kindem lemen's von den Eltem; sie sehen und hOren's nicht
anders und folgen dem Beispiel.
4 So wird es auf die natUrlichsten und sichersten Wege wahr, was
gesagt wird und geschrieben ist, dass der Eltem Segen und Fluch
auf den Kindem mhe und sie nicht verfehle. 12

As will be noted, sentence 4 is introduced by the particle so; if we were

to paraphrase this sentence without the use of this particle, we might arrive
at a rendering like the following:

(8) In this way [i.e. by observing and following the example of their
parents] what is said and written comes tme ... that the blessing
and curse of the parents descend on their children and do not
pass them by.

Now let us look at one English translation of this text.

(9) 1 We find very often, if we are just a bit observant, that peo-
ple are treated by their children in old age just as they once
treated their old and helpless parents.
2 And this procedure is understandable.
3 The children learn it from their parents; they see and hear noth-
ing else, and follow the example set them.
4 So, what is said and written comes true in the most natural and
surest way; that the blessing and curse of the parents descend
on their children and do not pass them by.13

Ifwe tried again to paraphrase sentence 4, avoiding the use of so, we would
probably arrive at something like (10):

(10) As a consequence [of the fact that children learn from their
parents and follow their example], what is said and written
comes tme ... that the blessing and curse of the parents descend
on their children and do not pass them by.

12 Johann Peter Hebel, 'Kindesdank und Undank'. In H. Steinhauer (ed.), First German
reader. Bantam Books, New York, 1972, pp. 6-10; p. 6, italics my own.
13 H. Steinhauer (ed.), First German reader. Bantam Books, New York, 1972, p. 7,
italics my own.
Translating What was Expressed 153

When we compare the two paraphrases (8) and (10), we feel that the
translation differs from the original: 'in this way' and 'as a consequence of
clearly differ in their meanings. Thus in the case of the German original, the
particle so seems to contribute to the truth-conditions of the utterance in
which it occurs, asserting essentially that 'children inherit the blessing and
curse of their parents by following their example'. In the English transla-
tion, however, the actual assertion is limited to 'children inherit the blessing
and curse of their parents', and this assertion is represented as following
from another assumption: that children follow their parents' example.
While the meaning of the German so here is essentially anaphoric, in an
adverbial function, it may seem less clear how the English so can give rise
to the' consequential' interpretation of the utterance. 14
As briefly mentioned in chapter 2, (p. 43), Blakemore's (1987) relevance-
theoretic study has shown that English so has as meaning "an instruction to
interpret the proposition it introduces as a logical consequence" (1987:87,
cf. also Blakemore 1988a). Thus when processing so in our English transla-
tion, the addressee uses the contextual information available to him to
construct an argument to which sentence 4 is the conclusion, as schematized
in (11) below.

(11) Premise 1

Premise n

Conclusion: What is said and written comes true in the most natu-
ral and surest way; that the blessing and curse of the parents descend
on their children and do not pass them by.

Under the principle of relevance the addressee will expect the premises to
be highly accessible contextual assumptions. What assumptions could these
be? Since part 3 of this section of text has just been processed by the ad-
dressee, it is very highly accessible at this point, and would naturally be
considered as a possible premise. (12) represents the state of interpretation
so far:

14 English so can also be used in an anaphoric sense. In the English translation consid-

ered here, however, so does not lend itself naturally to an anaphoric interpretation. Note
that, in contrast to German so, the most common use of so in English seems to be 'con-
sequential' (often corresponding to German also); its anaphoric use seems to be
comparatively rare.
154 Translation and Relevance

(12) Premise 1: The children learn it from their parents; they see
and hear nothing else, and follow the example set them.

Premise n

Conclusion: What is said and written comes true in the most natu-
ral and surest way; that the blessing and curse of the parents descend
on their children and do not pass them by.

Now the conclusion certainly does not follow from premise 1 alone - other
premises must be involved to make this a valid inference. In this case, the
premise (or set of premises) required would be something like the following:

(13) Premise 2: If the children follow the example of their parents,

then their blessing and curse descend on them.

Thus the use of so in sentence 4 of the example leads the audience to con-
struct an argument along the following lines:

(14) Premise 1: The children learn it from their parents; they see
and hear nothing else, and follow the example set them.

Premise 2: If the children follow the example of their parents,

then their blessing and curse descend on them.

Conclusion: What is said and written comes true in the most natu-
ral and surest way; that the blessing and curse of the parents descend
on their children and do not pass them by.

It is interesting to note that despite her detailed quality assessment of this

English translation, House (1981) does not point out this mismatch, though
it would presumably count as a lexical mistake, and hence come under her
category of 'overtly erroneous errors'. This could simply be an oversight;
however, it could also be indicative of the difficulty non-relevance-based
approaches to translation have with handling linguistic items whose pri-
mary function is to specify relevance-related relationships.
As just shown, relevance theory allows an explicit account of the seem-
ingly elusive 'meaning' of such connectives, and so makes a valuable
contribution to explicit comparisons of translated texts and their originals;
in this case it shows that the original and the translation differ significantly in
the clues they provide. While the original so has a vague, but truth-conditional
Translating What was Expressed 155

meaning that needs to be developed in order to obtain the propositional

form of the utterance, the translation - mistakenly - uses a pragmatic con-
nective that provides a clue as to how the propositional form is to achieve
relevance.

Communicative Clues Arising from Formulaic

Expressions

As mentioned in chapter 2, (p. 42) other verbal expressions that do not en-
ter the semantic representation are formulaic expressions like greetings,
standard openings and closings of formal letters and the like. We saw there
that, the greeting hello, for example, does not have a propositional form
that could be evaluated in truth-conditional terms; rather the way such ex-
pressions convey meaning is by entering into an appropriate description of
the utterance, such as 'Alfred said "hello'" (cf. p. 42 above); through this
description they activate the hearer's knowledge about the language, and
thus provide clues for the intended interpretation. Let us examine more
closely what this involves. 15
In the third section of this chapter (pp. 141-2) we said that concepts can
be associated with three distinct entries: logical entries, lexical entries and
encyclopaedic entries. The kind of concepts we dealt with there, like 'mother'
or 'pond', happened to refer to concrete objects. However, we can, of course,
also have concepts of abstract objects, - including the words of a language.
For example, I can have a concept not only of a cow, but also of the (Eng-
lish) word cow, and the two concepts would be distinct. In the same way I
can have a concept of the English word hello, and, like other concepts, it
could be associated with different kinds of information. Thus, since hello
has a phonological form that we can recognize, it must obviously have a
lexical entry providing information, for example, about its pronunciation.
It does not appear to have a logical entry since it does not have truth-
conditional properties. Its 'meaning' is rather contained in its encyclopaedic
entry - it consists in what we know about the word hello: that it is a greet-
ing, that it is used on informal occasions, and any other pieces of information
one may have about the use and appropriateness of this word.
It is in virtue of this encyclopaedic information that formulaic expres-
sions contribute to the interpretation of utterances - that is, in this way they

15 As indicated in chapter 2, I am indebted to Deirdre Wilson (personal communication)

for this account.

156 Translation and Relevance

provide communicative clues for the intended interpretation. So this ency-

clopaedic information can serve as a basis for handling such expressions in
direct translation. When translating into another language, one can look for
an expression that has similar information associated with it, an expression
that is, for example, also recognized as an informal greeting.
It is interesting to note that there are formulaic expressions that could be
assigned a semantic representation, but normally seem to function in terms
of their encyclopaedic properties rather than semantic properties. A case in
point would be the stereotyped expression Yours sincerely used at the end
of formal letters. Clearly this expression can be assigned a semantic repre-
sentation, involving the concepts 'you' and 'sincere' - but it seems equally
clear that these concepts do not normally enter into our interpretation of
such utterances at all. We should not normally interpret this expression as
some kind of a claim relating to the sincerity of the writer, but rather rec-
ognize it simply as a set expression commonly used to close a letter -
information typically stored in encyclopaedic entries of words or other lin-
guistic expressions.
It is this fact that gives translators the option of translating, for example,
the Amharic greeting t'ena yist'min, literally 'May he give (you) health on
my behalf!' by an English expression like hello. 16 Though this expression
does have a semantic representation, involving concepts like giving and
health, no thought of these seems to cross people's minds when they ex-
change this greeting. Rather, it seems that speakers of Amharic have a
concept of this expression as a whole, and the encyclopaedic entry would in
this case presumably contain little more information than that this is a gen-
eral greeting. There seem to be no special restrictions on its use - it can be
used on virtually any occasion, at any time of day or night. Given that this is
the information made accessible by the original stimulus, the translator would
then look for a stimulus in the receptor language that provides this same
kind of encyclopaedic information, and in English, hello might be consid-
ered a possible rendering on the grounds of its generality, though it may be
judged less formal than the Amharic greeting.
One might wonder why expressions that have semantic properties should
not be used in virtue of those properties. However, once we take into ac-
count that utterance interpretation is sensitive to processing effort, this
phenomenon does not seem unexpected. Ifwe take the example of Amharic

16 In this transcription t' stands for the alveolar ejective, rior the unrounded high central

vowel, and doubly written letters represent phonemically long segments.

Translating What was Expressed 157

t' ena yist 'min, rather than derive the assumption schema provided in virtue
of its syntactic and semantic properties and then enrich it into a propositional
form every time the expression is used, it seems much more economical to
recognize it as a standard greeting and infer that the person uttering the
expression was greeting the addressee. Historically, such expressions would
probably start out as ordinary utterances, but their stereotyped recurrence
could be expected to lead to the development of cost saving processing
strategies.
It may be worthwhile at this point to digress briefly because this possi-
bility of storing (part of) the interpretation of an utterance in memory and
retrieving it from there has some interesting ramifications. In particular, it
can shed new light on the often deplored phenomenon that expressions which
are felt to be very meaningful and rewarding when they are first coined tend
to lose their impact the more often they are used. Steiner (1975), for exam-
ple, expresses this experience as follows:

Words seem to go dead under the weight of sanctified usage; the

frequence and sclerotic force of cliches, of unexamined similes, of
worn tropes increases. Instead of acting as a living membrane, gram-
mar and vocabulary become a barrier to new feeling. (1975:21)

What is described here in somewhat poetic terms can be accounted for,

at least in part, in the relevance-theoretic framework. Steiner refers prima-
rily to non-literal uses oflanguage, that is, tropes. As we saw earlier on (cf.
chapter 4, pp. 88-9), tropes are rewarding in that they tend to convey a wide
range of comparatively weak implicatures, thus creating an impression. On
the first hearing or reading, many of the implicatures are derived as contex-
tual implications, and hence they themselves contribute to relevance as
contextual effects.
However, the more often a certain trope is used, the more of its interpre-
tation is likely to be remembered. So on later occasions the audience will be
able to retrieve more and more of the interpretation directly from memory.
But from the point of view of relevance, there is an important difference
between the retrieval of information from memory and its recovery by in-
ference. Assumptions retrieved from memory do not in themselves constitute
contextual effects, because contextual effects are defined as resulting nei-
ther from the utterance alone nor from the context alone but only from the
inferential combination of both (cf. chapter 2, p. 29). Therefore, while on
later occasions the audience may still derive the same set of assumptions, the
158 Translation and Relevance

more of these assumptions are recalled from memory, the fewer the contex-
tual effects will be compared to the original occasion of utterance. The fewer
the contextual effects, the less relevant the utterance is judged to be, and so
the feeling of staleness and dissatisfaction with well-worn tropes, which
may eventually become 'dead metaphors', follows naturally from consid-
erations of relevance. 17
To return to our original topic of discussion, standard notices, like No
smoking or Wet paint, seem to be treated in a way similar to that of formu-
laic expression used as greetings. Again, it is unlikely that the reader of
these notices takes the trouble of developing these elliptical expressions
into full propositional forms - all he needs to do is recognize the communi-
cative clues they provide: a prohibition of smoking and a warning against
getting clothes soiled by wet paint respectively. If this is correct, then the
often-recommended strategy of translating such standard expressions by
corresponding standard expressions in the receptor language - even if their
semantic contents are different - can be accommodated within this range of
direct translation. Thus the German Frisch gestrichen could be considered
a direct translation of the English Wet paint on the grounds that the two
expressions provide the same communicative clue: they resemble each other
in the characterization that they are the standard expressions used to warn
people of wet paintY
An interesting suggestion by D. Wilson (personal communication) is
that proverbial sayings could be accounted for along the same lines. One
noteworthy property of such sayings is that although their semantics-based

17 Familiarity with such tropes will also make their processing 'cheaper', and at a cer-

tain stage this may balance the loss in contextual effects. However, low processing effort
in itself does not make for consistency with the principle of relevance - the audience
will be looking for contextual effects adequate to the occasion, and in literary writings
the level of adequacy is likely to be high.
18 The cautious form of expression here is deliberate because the fact that these commu-

nicative clues arise from encyclopaedic entries may suggest that this information is
contextual rather than intrinsic, hence does not really constitute a communicative clue.
(Cf. the claim that the distinction between logical and encyclopaedic entries corresponds
to the distinction between content and context, p. 142 above). To this two comments
seem in order. Firstly, one could argue that the encyclopaedic information associated
with concepts of words that lack any semantic content differs from encyclopaedic infor-
mation associated with the extension of other concepts in that information associated
with words is somehow intrinsic to the language. Secondly, as we shall see in the next
chapter, the concept of' communicative clue' is introduced only as a transitory notion -
our final treatment of direct translation will do without it. Therefore it seems unneces-
sary to discuss its further theoretical ramifications.
Translating What was Expressed 159

meaning often presents no problem, it is still difficult to say what that

proverb 'means'. For example, there is no problem in understanding the
semantics-based meaning of the German proverb Man muss mit den Wolfen
heulen, literally, 'One must howl with the wolves.' However, even with this
knowledge, a foreigner learning the language would probably still wonder
what this expression 'really means'. At the same time, it does not seem to
be the case that proverbs get assigned a different semantic representation,
as idiom chunks do - they seem to be much vaguer in their 'proverbial
meaning', and could be paraphrased in different ways on different occa-
sions. Another significant property of proverbs seems to be that they need
to be recognized as proverbs to have their full effect; thus children, and also
foreigners, often seem to be puzzled the first time they hear a proverb used,
and it seems that proverbs derive their 'authority' from the fact that they
express popular insights, rather than someone's personal opinion.
Under our current approach, it seems plausible to assume that these ex-
pressions are treated in our mental lexicon as units - in which case it would
be possible for us to have a concept of that lexical unit, that is, of the prov-
erb itself, and also an encyclopaedic entry associated with it. On first
encounter, we should try to understand it by ordinary processes of utterance
interpretation, presumably along the lines of interpretive use as 'something
people say'. However, as, in the course of time, this set expression recurs, it
seems likely that some of these interpretations get stored in the encyclopae-
dic entry associated with the concept of this proverb. Thus we should come
to store a number of possible paraphrases and information about occasions
of use in memory.
This proposal would account not only for the observations mentioned
above, but it would also explain why in translation there is the expectation
that proverbs should be translated holistically rather than in terms of their
semantic meaning: "One translates as lexical units standing expressions and
most of the popular proverbs and sayings" (Levy, 1969: 102). Thus When in
Rome, do as the Romans do could be justified as a direct translation of the
German Man muss mit den Wolfen heulen along the lines of stimulus repro-
duction. It would be based on information shared by the encyclopaedic entries
of both proverbs.

Onomatopoeia and Communicative Clues

The same framework can help us to deal explicitly with the translation of certain
onomatopoeia. In the quotation given on p. 150 above, Levy distinguished
160 Translation and Relevance

between two kinds of onomatopoeia; one kind were "sound-imitating se-

quences ... created ad hoc" (1969:91), and these can be reproduced in a
translation by transcription only. The second kind was characterized by Levy
as sound sequences that have acquired "conceptual values and the character
of a word, as is the case with the 'language' of pets and with the most com-
mon sounds of nature" (1969:91, translation my own). For such
onomatopoeia Levy claimed that translation proper is possible.
The second kind of onomatopoeia are not words with 'normal' semantic
meaning whose sound structure may in addition be felt to reflect something
of the event or thing for which it stands, as the English word slither might
exemplity. Rather, as indicated in the quote, Levy refers to expressions used
to imitate noises produced, for example, by animals or otherwise found in
nature. Thus he gives the following example:

In Waste Land Pierre Legris replaces the onomatopoeic allusion' Jug,

jug, jug, jug, jug, jug' to a nightingale by the French equivalent 'Tio,
Tio, Tio, Tio, Tio, Tio'. (Levy 1969:90, translation my own)

The problem is that neither the Englishjug nor the French Tio are 'con-
ceptual' in the semantic sense - they do not, for example, contribute to the
propositional form of utterances - and yet there seems to be something right
about Levy's way of referring to them as having 'conceptual value'.
Here our current approach offers a solution: it provides for the possibility
of having a concept ofthe onomatopoeic expression itself. In this way it can
account for Levy's intuition without upsetting the notion of truth-conditional
semantics. It can also explain why Levy should attribute 'the character of a
word' to such expressions - the reason being precisely that we should have
a concept, a mental representation of such an onomatopoeic expression as
we have of any other word of the language.
As far as his observation about its translatability is concerned, since the
concept representing the onomatopoeic expression can have information
about its use stored in its encyclopaedic entry, and since this information
can constitute a communicative clue, it can be translated into another lan-
guage ifthat language has an expression associated with similar information.

Communicative Clues and the Stylistic Value of Words

The assumption that there are concepts of words that can have encyclopae-
dic information associated with them allows us also to deal with properties
Translating What was Expressed 161

of verbal expressions that are sometimes called 'stylistic' or 'connotative',

and described in terms of sociolinguistic characteristics like 'register' ,dialect',
'accent' and the like. The literature on translation presents many examples
that exhibit violations in this area. Here is one given by Chukovskii:

In Shakespeare's Julius Caesar, Brutus' wife reproaches her

husband:

You've ungently, Brutus,

Stole from my bed: and yesteruight, at supper,
You suddenly rose, and walk'd about. (II,i)

In his translation of the tragedy, A.A. Fet used the word nevezhlivo,
'impolitely,' for 'ungently,' the colloquial word vechor instead of
vecher for 'yestemight,' and the ultra-formal word trapeza for 'sup-
per.' Critics of the time noted the disparity of style: "What a strange
conjunction in three short lines of three words with such totally differ-
ent nuances - polite, colloquial, and formal!" (Chukovskii 1984:98)

In fact, the translation ofJuruike in Yuasa's example (cf. p. 139) also

involves a stylistic problem of this very kind. We noted there how relevance
theory helped us to see that some of the information that Yuasa felt was
missing from certain translations - that is, information about the presence
of the poet by the side of a pond in a Japanese garden - was contextual
rather than semantic. Yuasa notes another deficiency with the translation
'old pond':

Second, I noticed thatfuruike was not the same thing asfuruki ike.
The latter is perfectly conversational, even prosaic, but the former is
unlikely to be used in ordinary conversation: being a compound, it
has a bookish Chinese flavour, is even somewhat archaic. (Yuasa
1987:234).

In this passage Yuasa is obviously not saying that the pond had 'a book-
ish Chinese flavour', but that this is a property of the word Juruike. In terms
of the account given above, this can again be explained by saying that there
is a concept of the word Juruike, and that part of its encyclopaedic entry has
this information about its Chinese association and archaic nature.
Once one realizes that this information relates to a particular word of the
Japanese language, one is not surprised that Yuasa could not find an Eng-
lish word that would convey this same information - there would most likely
162 Translation and Relevance

not be an English word that would be felt to have a 'bookish Chinese fla-
vour'. However, there are words in English that are perceived as 'archaic',
and this is the aspect of style Yuasa tried to approximate in his translation:
"to convey the archaic flavour of the original word, I decided to use the
word 'ancient''' (Yuasa 1987:234).
Thus we find that the assumption that our minds store concepts of words
in addition to concepts of the objects and events which those words de-
scribe is rather consequential. It allows us to give an explicit, empirical
account not only of the 'meaning' of such 'meaningless' expressions as
hello, but of a whole range of stylistic phenomena. Note also that this model
provides a principled account of how there can be differences in connota-
tive meaning between synonyms, which by definition are supposed to mean
the same. The identity in meaning, that is, the fact that two words make the
same contribution to the truth-conditions of an utterance, as, for example,
copper and policeman would, can be accounted for by the fact that both
words are associated with a concept of the same kind of person. The differ-
ence in connotation, on the other hand, would naturally be provided for by
the fact that each word is mentally represented by a distinct concept, and
that each of these concepts could have its own encyclopaedic entry.

Communicative Clues Arising from Sound-Based

Poetic Properties

For all utterances that involve an expression without semantic properties,

we said that the communicative clues it provides are mediated by a descrip-
tion ofthat utterance. Such descriptions are not exceptional, required perhaps
only for such special expressions, but playa part also with verbal expres-
sions that do have a semantic representation. Thus, whenever a person says
something, this fact itself will be manifest to the audience; for example, if
Daisy says to Douglas 'There is a draft from the window', this will make
manifest to Douglas the following assumption, which is a description of the
stimulus, just as 'Alfred said "hello'" (cf. p. 155 above) would be: 19

(15) Daisy has uttered the sentence 'There is a draft from the window'.

19 As Sperber and Wilson show, such an utterance makes manifest other assumptions as

well- for example, that someone has made a sound, that someone is present, that Daisy
is present and so forth. For further discussion see section 2 of chapter 4 in Sperber and
Wilson (1986a).
Translating What was Expressed 163

Hence the involvement of descriptions of utterances with expressions that

lack semantic properties is nothing extraordinary.
This realisation that verbal stimuli regularly affect communication via
some appropriate description of their properties can help us also to account
for the possibility of exploiting the sound of words for communicative ef-
fects. Let us look at the following example discussed by Savory (1957):

When Virgil wrote

Quadrupedante putrem sonitu quatit ungula campum

[With galloping sound the hoof strikes the crumbling ground, gloss
my own]

he wrote a line which has ever since been famous because its rhythm
and its accents suggest the thudding of the hooves of a galloping horse,
but no translator can preserve and reproduce this. (Savory 1957:79)20

When read with the proper dactylic metre, the rhythm could bring to the
hearer's mind the realization that this utterance sounds like "the thudding
of hooves of a galloping horse". The recovery of such a description would
no doubt be aided by the semantic content of the utterance. This description
could then influence the overall interpretation of the utterance as a further
communicative clue.
Since the rhythm and accents of the utterance are due to the particular
phonetic properties of the Latin words of which it consists and to Latin
poetic conventions of stress placement, it would indeed be unlikely that a
sequence of words in some other language would resemble the original in
its semantic representation and warrant a description of similarity in sound
to the 'thudding hooves' at the same time. So Savory's claim that "no trans-
lator can preserve and reproduce this" seems quite understandable.
Of course poetry exploits the phonetic properties of words and utter-
ances not only on the grounds that they may be perceived as resembling the
sound of phenomena in our world. Thus Levy observes that rhyme and
rhythm impose structural constraints on the construction of the utterance
that are independent of syntactic structure, and that can, therefore, weaken
the influence of the syntactic relations on the interpretation:

The line quoted is from the Aeneid, book 8, line 596; cf. O. Ribbeck (1862), P. Vergili
20

Maronis, Aeneidos libri VII-XII, Teubner, Leipzig, p. 104.

164 Translation and Relevance

A weakening of syntactic relations in verse follows simply from the

fact that these are not the only organising factors. The continuity of
the sentence in verse is broken by verse boundaries (oo.), and con-
versely individual parts that do not cohere syntactically are linked
by rhyme and other formal parallelisms. All this contributes to the
independence of smaller segments and a weakening of connectives
and syntactic functions. (1969: 174, translation my own)

These observations are very valuable, but it may not be obvious why such a
weakening of syntactic relations should be desirable in poetry.
According to relevance theory, poetic effects arise essentially when the
audience is induced and given freedom to open up and consider a wide
range of implicatures, none ofwhich are very strongly implicated, but which
taken together create an 'impression' rather than communicate a 'message'.
(Cf. our discussion of the indeterminacy of meaning in chapter 4, pp. 89-
90). The reason why rhyme and rhythm can have such a poetic effect is
that they can provide this kind of freedom for interpreting the text in ques-
tion. As Levy points out, in prose the interpretation of the utterance follows
the syntactic organization of the utterance; concepts are grouped together
and interpreted in terms of their syntactic relations, and of course one im-
portant function of syntactic relations is to specifj; the semantic relations
that hold between the various constituents of the sentence. In this way syn-
tactic structure is one of the essential properties of natural language that
allow it to work with a degree of precision not normally afforded by non-
verbal communication.
However, on the assumption made in relevance theory that poetic ef-
fects require the freedom to explore a wide range of comparatively weak
interpretations, it is clear that the 'precision' of syntactic structure will of-
ten be found to inhibit poetic effects: it reduces rather than extends the
range of possible interpretations.
Rhyme and rhythm, however, impose phonological patterns that are in-
dependent of syntactic structure and may indeed cross-cut it. These patterns
tend to enrich the interpretation, not only because they give rise to additional
groupings, but also because, in contrast to syntactic relations, the relations
they suggest are unspecified and so allow greater freedom in interpretation.
The following lines from Shakespeare's Sonnet V may serve to illus-
trate this:

Then, were not summer's distillation left

A liquid prisoner pent in walls of glass,
Translating What was Expressed 165

Beauty's effect with beauty were bereft,

Nor it, nor no remembrance what it was;
But flowers distill' d, though they with winter meet,
Leese but their show: their substance still lives sweet.

Here, the boundary at the end of the first line induces the reader to consider
this verse on its own, though syntactically the end of the sentence is not
reached yet, and there is a corresponding effect in the second and third line.
Thus the reader is invited to dwell on and look for rewarding effects not
only from the sentence as a whole, but also from each part separately, as
divided by the verse boundaries.
Conversely, though the fourth line is not linked to the other parts of the
poem by any particular syntactic or logical function, it is presented as an
integral part of it by its conformity to the formal structure, and it is left to
the audience to explore its relationship to the content of those other parts.
To illustrate further that poetic effects do indeed depend essentially on
this freedom to explore, let us briefly digress to the discussion of one of
Hebbel's poems by E. Staiger (1956:37ff.). The poem begins as follows:

Komm, wir wollen Erdbeeren pflucken,

1st es doch nicht weit zum Waldo
Wollenjunge Rosen brechen,
Sie verwelkenja so bald!
Drobenjene Wetterwolke,
Die dich angstigt, mrcht ich nicht;
Nein, sie ist mir sehr willkommen,
Denn die Mittagssonne sticht.
(Italics my own)

Prose translation:

Come, let us pick strawberries,

After all, it's not far to the woods.
Let's pick young roses,
Since they wither so soon!
That stormy cloud above,
Which makes you afraid, I do not fear;
No, it is very welcome to me,
For the midday sun bums hot.
(Translation and italics my own)
166 Translation and Relevance

Staiger comments that the poem gives a "frosty impression", and seems
"educative" rather than creative, and he suggests that this is to be blamed
mainly on the seemingly harmless words 'doch', 'ja', 'nein', 'denn'. Once
they are dropped, these educative verses become much more like a song:

Wir wollen Erdbeeren pflucken,

Es ist nicht weit zum Wald,
Und junge Rosen brechen,
Rosen verwelken so bald ...
(Staiger 1956:37f.)

Prose translation:

We want to pick strawberries,

It is not far to the woods,
And pick young roses,
Roses wither so soon ...
(Translation my own)

Again, in terms of relevance theory this observation is natural enough: most

of the expressions blamed by Staiger fall in the class of 'pragmatic connec-
tives', and as we saw earlier on, the main task of pragmatic connectives is
to narrow down the number of possible interpretations of an utterance by
specifying how the proposition expressed is intended to be relevant.
Returning to the phonological patterns of rhyme and rhythm, our notion
of restricted interpretive resemblance would allow for them to be taken into
account in direct translation: they provide important communicative clues
for the interpretation of poetic texts. Whether any given receptor language
has the means of providing similar clues is a matter of statistical probability.
Note, however, that our current definition of direct translation does not
provide for a rendering that tries to achieve the poetic effects of the original
by spelling out or 'explicating' some of the weak implicatures of the origi-
nal. As mentioned earlier (cf. p. 143), such implicatures involve the use
of contextual information and hence go beyond the clues provided by the
communicator.

Conclusion

Here we come to the end of our survey of what kinds of phenomena would
be covered by the notion of communicative clues. While this survey could
Translating What was Expressed 167

not be exhaustive, it did show the possibility of setting up a narrowly de-

fined notion of direct translation that allows for the preservation of a wide
range of stylistic elements. In doing so we were also able to show that rel-
evance theory offers the explicit treatment of a number of features and
characteristics of texts that have sometimes been viewed as "elusive tonali-
ties of speech". (Chukovskii 1984:48, cf. p. 132 above)21
However, useful as it may seem, defining direct translation in terms of
shared communicative clues is unsatisfactory in a number of respects. There-
fore, in the next and final chapter we shall see that there is, in fact, a better
way of handling this notion that allows a unified account of both direct and
indirect translation at the same time.

21 Perhaps a word of explanation is in order here for those readers who might have
expected a category of communicative clues that relate to 'discourse features'. The rea-
son why no such category has been set up is not that these features are not considered
important, but rather the belief that the phenomena commonly referred to under this
heading do not form a domain of their own, but arise from the interplay of various
linguistic properties of utterances with considerations of relevance. For proposals along
these lines see Blass (1986), (1988, chapters 1 and 2), (1990), and Blakemore (1988b).
Some 'discourse features', however, are probably cultural, i.e. contextual, hence do not
come under communicative clues: e.g. the tendency to present the main point of a text
at its begiuning (cf. Gutt 1982).
7. A Unified Account of Translation

In the last two chapters we outlined two notions of translation that seemed
rather different from each other. In chapter 5 we considered translation
defined as 'interlingual interpretive use', defined in terms of shared explica-
tures and implicatures. This notion of translation is very flexible and highly
context-sensitive - but pays the price of allowing a wide range of receptor
language texts to qualify as translations. In chapter 6 we therefore devel-
oped the additional concept of 'direct translation', defined in terms of shared
communicative clues. It constitutes a fixed, context-independent notion.
While this state of affairs may be acceptable from the practical point of
view, from a theoretical perspective the situation is unsatisfactory in sev-
eral respects.
Firstly, while the notion of 'interlingual interpretive use' fits straightfor-
wardly into the framework of relevance theory, matters seem much less clear
with 'direct translation': we introduced the notion via direct quotation - but
we did not discuss how either direct quotation or direct translation relate,
for example, to interpretive use.
Secondly, our discussion of direct translation in chapter 6 remained
largely on an intuitive level; we did not propose any relevance-based
technical definition of 'communicative clue' nor of the notion of 'direct
translation' itself.
Thirdly, given that we have arrived at two alternative ways of viewing
translation - how do the two relate to one another? And could there not be a
number of other options? In other words - is it not possible that there are
not just two, but perhaps five or scores of other notions of translation that
need to be considered?
Fourthly, if there are different approaches to translation, how is the
translator to know which one to follow on any given occasion? And what
effect does his choice have on the audience who may have expectations of
their own?
The first three questions will occupy us in the first three sections of this
chapter. The fourth section will deal with matters relating to question four.
Finally, I shall try to draw out the main results of this quest for a relevance-
based account of translation.
A Unified Account of Translation 169

Direct Translation - a Special Case

of Interpretive Use

In order to clarify how direct translation relates to other relevance-theoretic

notions, let us first take a closer look at direct quotation. While Sperber and
Wilson do not deal with direct quotation as a topic on its own, they refer to
it in a number of their writings, usually in connection with resemblance-
based uses of verbal stimuli. In Sperber and Wilson direct quotations are
presented as "the most obvious examples of utterances used to represent
not what they describe but what they resemble" (1986a:228). More specifi-
cally, a direct quotation resembles its original "because it is a token of the
same sentence" (1986a:227). Similarly, in Wilson and Sperber (1988a) di-
rect quotations, together with parody, are said to be based "on resemblances
in syntactic and lexical form" (19881a:136). And, as already quoted in chap-
ter 6, (p. 133) in the same article Wilson and Sperber state that, "Direct
quotations are chosen not for their propositional form but for their superfi-
ciallinguistic properties" (1988a:137).
Since direct quotations rely on shared linguistic properties rather than
shared assumptions, they do not automatically constitute instances of inter-
pretive use. However, they are nevertheless of interest for interpretive use
because there is a causal interdependence between stimulus, context and
interpretation: that is, if two stimuli with identical properties are processed
in the same context, they will lead to the same interpretation because they
causally interact by the same principle - the principle of relevance. Thus if
both stimuli have the same properties and the contexts in which they are
processed are also identical, then their propositional forms and/or de-
scription, their contextual effects, the processing effort required, and the
evaluation of optimal relevance will all be the same, leading to identical
interpretations. 1
It is this fact which makes direct quotations of special interest for inter-
pretive use; since they preserve all the linguistic properties of the original,
they give the audience the possibility of reconstructing for itself the mean-
ing intended by the original communicator, provided it uses the contextual
assumptions envisaged for the original act of communication.
However, as mentioned above, the fact that direct quotations can be
employed in this way, is not a defining characteristic. An utterance can be

1 This claim obviously requires some ceteris paribus hedging; for example, mental fa-

tigue may lead to distortion of the interpretation.

170 Translation and Relevance

quoted directly for purposes that may have nothing to do with the intended
interpretation of the original; for example, an utterance may be quoted di-
rectly in a linguistic article not in virtue of the interpretation it was intended
to convey originally, but perhaps in virtue of some remarkable grammatical
feature it displays.
Turning now to direct translation, when we introduced this notion in
chapter 6, one of the important points made was that direct translation can-
not be understood in terms of resemblance in actual linguistic properties,
for the simple reason that languages differ in those properties. In fact, if the
linguistic properties of the original were reproduced, the result would not
be a translation at all, but an actual quote from the source text in the source
language. For this reason we suggested a more abstract basis for direct
translation; we argued that what mattered were not the actual linguistic
properties of the source language utterance, but rather the 'communicative
clues' they provided to the intended interpretation. What we did not bring
out then is that that argument does, in effect, make the notion of direct trans-
lation a case of interpretive use. The following considerations will show
why this is so.
According to or proposal in chapter 6, for an utterance in the receptor
language to qualify as a direct translation of some original utterance in the
source language, it needs to share all the 'communicative clues' ofthat origi-
nal. But how can we know whether two utterances share a communicative
clue? We argued that clefting in a Silt'e sentence, for example, correspond-
ed to contrastive stress in an English sentence in that both had 'the same
effect' on the interpretation of the sentence. How does one know whether
clefting and contrastive stress have the same effect in these two languages?
The crucial point here is that the sharing of these properties is necessary to
obtain the same interpretation in both cases. Other things being equal, a
Silt' e sentence without clefting would yield an interpretation different from
that intended by the English original, whereas a Silt' e sentence that did
have it would convey the same interpretation, if processed in the originally
envisaged context.
On this argument, the way we can know whether two utterances in lan-
guage A and B share all their communicative clues is by checking whether
they give rise to the same interpretation when processed in the same con-
text. This in tum means that the notion of direct translation is dependent on
interpretive use: it relies, in effect, on a relationship of complete interpre-
tive resemblance between the original and its translation.
A Unified Account of Translation 171

On this recognition we can now consider a definition of direct transla-

tion that is fully integrated into the relevance-theoretic framework:

(1) A receptor language utterance is a direct translation of a source

language utterance if and only if it purports to interpretively re-
semble the original completely in the context envisaged for the
original.

The various stipulations of this definition as well as some of their rami-

fications will be considered in the remainder of this chapter. Before turning
to matters of detail, let us consider the significance of its most basic claim,
that is, that direct translation is a case of interpretive use.
First of all, this claim provides answers to the first three questions raised
at the beginning of this chapter. With regard to the relationship of direct
translation to relevance theory, the answer is that direct translation is cov-
ered in the relevance-theoretic framework as an instance of interpretive use.
Since indirect translation, too, falls under interpretive use, we find that we
have, in fact, arrived at a unified account of translation: both direct and
indirect translation are instances of interlingual interpretive use. Thus de-
spite first appearances to the contrary, direct translation is not of a kind
altogether different from indirect translation - relevance theory offers a uni-
fied account of both.
At the same time, we now have a better understanding of the similarities
and differences between direct quotation and direct translation. The two
differ in that the resemblance relationship found in direct quotations relies
on the sharing of linguistic properties only, without necessary reference to
the intended interpretation of the original; direct quotation can therefore be
defined apart from interpretive use; direct translation, however, relies on
interpretive resemblance and can therefore only be defined within inter-
pretive use.
What the two have in common is that both can be used to give the audi-
ence access to the originally intended interpretation. In this respect we might
say that direct translation is the interlingual 'simulation' of a direct quota-
tion: if processed in the original context, a direct translation purports to
allow the recovery of the originally intended interpretation interlingually,
just as direct quotations purport to do intralingually. In the case of direct
quotation this possibility is incidental - direct quotations can be used for
other purposes as well; for direct translation it is the crucial property - it
defines this form of interlingual communication.
172 Translation and Relevance

The answer to the second question is clear, too: we have just proposed a
technical definition of direct translation. As it turned out, we were able to
do so without reference to 'communicative clue', since this concept does
not have any theoretical status of its own but is, in fact, derived from the
notion of 'interpretive use'. However, it may well be that the concept 'com-
municative clue' will prove of some value in the practice of translation: it
might help the translator to identify and talk about features in the source
and target language utterances that affect their interpretations.
As to our third question, why there should be two rather than three or
more different approaches, the answer is that this is natural on the assump-
tion that interpretive resemblance is a graded notion that has complete
resemblance as its limiting case: indirect translation covers most of the con-
tinuum, and direct translation picks out the limiting case. Since interpretive
resemblance extends along a cline, there is no reason why there should be
other principled ways of distinguishing between cases of interlingual inter-
pretive use.
Furthermore, the claim that translation generally falls under interpretive
use is significant in that it offers an explanation for one of the most basic
demands standardly made in the literature on translation - that is, that a
thorough understanding of the original text is a necessary precondition for
making a good translation. This is naturally entailed if translation is based
on interpretive use: in order to produce such a translation, the translator
obviously needs to know the interpretation of the original, and in the case
of direct translation, aiming at complete interpretive resemblance, his knowl-
edge of the original interpretation would have to be very good indeed. 2
This also implies that translation is dependent on the translator's inter-
pretation of the original, or to put it more correctly, on what the translator
believes to be the intended interpretation of the original. In all cases where
the interpretation of the original is not obvious this opens the possibility of
error: if the translator misinterprets the original, then his translation is likely
to misrepresent it, too. This is another facet of the difference between direct

2 This has some interesting implications for attempts to simulate translation in artificial
intelligence, because it implies that truly adequate simulations will have to deal not just
with sentences and their linguistic representations but with their interpretations in con-
text. If relevance theory is right, this will further necessitate ways of simulating relevance
in artificial intelligence, which may posit challenges not considered so far; for example,
it would need to take into account the comparative degree of accessibility of informa-
tion stored in memory and the ability of humans to make judgements about the adequacy
of contextual effects.
A Unified Account of Translation 173

quotation and direct translation: direct quotation is possible without a proper

understanding of the intended interpretation of the original. Recitations of
memorized religious texts may be a case in point. They are possible even if
the meaning of the text is obscure to the speaker - perhaps because it is in a
foreign language. Direct translation, however, presupposes an understand-
ing of the original.
Finally, direct translation answers the question we raised at the begin-
ning of chapter 4 (p. 69): it shows how the translator can succeed, after all,
in communicating the originally intended interpretation to the receptors.
However, the crucial point is that direct translation presumes to do this only
in the context envisaged by the original communicator and not in any con-
text the receptor audience may happen to bring to the translation.
This observation brings us to the two main points of our definition that
set direct translation offfrom indirect translation, which are the demand for
the translation to be interpreted with regard to the original context on the
one hand, and the presumption of complete interpretive resemblance in that
context on the other.

On the Use of the Original Context

The requirement that direct translations should be processed with respect to

the original context may seem surprising and perhaps artificial at first -
especially against the background of popular demands that translations
should not require the receptors to know the original culture. However, a
closer look will show quickly that this is not an unreasonable demand at all.
As mentioned earlier, this requirement applies straightforwardly to the
interpretation of direct speech quotations, the intralingual equivalent of di-
rect translation. If one wants to understand what a direct quotation was
originally intended to convey one has to try to take into account its original
setting and interpret it in that light; otherwise one could go seriously wrong,
especially in the case of non-literal uses oflanguage or irony. This point is
not only common sense but well recognized in literary studies; one of the
preconditions of authentic literary interpretation is a reconstruction of the
historical, cultural and sociological background in which the piece oflitera-
ture in question was written.
Looked at in this light, what seems hard to understand is not the imposi-
tion of this condition but rather its widespread neglect in translation. Not
only have there been calls that translations should generally be spontane-
ously intelligible for the receptors, no matter how different their background,
174 Translation and Relevance

but translated works are regularly criticized for failing to convey what are,
in fact, context-dependent implicatures. As will be recalled, this seemed to
be the case with Yuasa's evaluation of some of the translations of Bash6's
haiku when he felt that "it was the responsibility of a translator to say more
than 'an old pond' to give a little more sense of the presence of the poet by
the pond" (Yuasa 1987:234). If the same standards were applied to direct
quotation, then someone wanting to quote from Shakespeare today should
word the 'quotation' in such a way that the audience could interpret it cor-
rectly, no matter how different their background might be from that of
Shakespeare's original audience. The most likely reason for this somewhat
absurd situation seems to lie in the code-model view of language and com-
munication. On that view the successful communication of the original
message would mainly depend on the proper use of the code (except for
'noise' in the channel), and so, if the translation led to misunderstandings,
the cause would naturally be thought to be a coding mistake, that is, an
error on the translator's part.
Considered from the point of view of relevance theory itself, the refer-
ence to the original context is motivated by the need to make the originally
intended interpretation communicable, that is, to overcome the basic
problem over which the 'communicative approaches' stumbled. On the as-
sumption that the audience will use the contextual information envisaged
by the original communicator, it is reasonable to expect that they will be
able to identity the originally intended interpretation by the criterion of con-
sistency with the principle of relevance. 3
In secondary communication situations this implies the need for the tar-
get audience to familiarize themselves with the context assumed by the
original communicator. Depending on how accessible this information is to
them, this may not be an easy task. However, as pointed out above, this task
is neither unusual nor arbitrary but needs to be carried out by anybody inter-
ested in communications not intended for himself, even in intralingual
situations. 4

3 In Gutt (1990) the stipulation 'in the context envisaged for the original' is omitted

from the definition of direct translation; it is seen to be entailed by the stipulation of

complete interpretive resemblance in the sense that the recovery ofthe intended inter-
pretation in consistency with the principle of relevance can be guaranteed only with
regard to the original context. There is no a priori reason why the original interpretation
should be communicable in any other context. However, for the clarity of exposition it
seems better to have this condition expressed in the definition, especially since it has
been overlooked in the communicative approaches (Cf. our discussion of communica-
bility in chapter 4, pp. 103-4).
4 The requirement that the readers of the translation should familiarize themselves with
A Unified Account of Translation 175

For the translator, one of the important consequences of this is that it

makes the explication of implicatures both unnecessary and undesirable. It
makes it unnecessary because the reason for such explication was mismatches
in contextual information in the cognitive environment ofthe receptors. Since
in direct translation it is the audience's responsibility to make up for such
differences, the translator need not be concerned with them. It also makes
such explication undesirable because it would be likely to have a distorting
influence on the intended interpretation. This can be illustrated simply with
one of the examples of figurative language from Larson that we discussed
in chapter 4 (p. 88):

(2) John eats like a pig. (Larson 1984:248, reproduced here for con-
venience)

Larson observed that this simile allows for various "points of similarity" to
be seen, such as "the pig eats too much" or "the pig eats sloppily" (1984:248).
I argued that, contrary to Larson's assumption, it is not necessarily the case
that this figure was intended to communicate strongly one of these two pos-
sible interpretations, and that, in fact, non-literal language typically relies
on conveying a wider range of implicatures more weakly. Against this back-
ground it can be seen that the explication of any particular implicature or
set of implicatures would have a distorting influence. Thus none of the ut-
terances in (3), each explicating some implicature(s), conveys the same
interpretation as (2) above:

(3) (a) John eats too much, like a pig.

(b) John eats sloppily like a pig.

Even (4) does not necessarily mean the same as (2):

(4) John eats too much and sloppily, like a pig.

the historical and cultural setting of the original seems to be implicit also in the view of
translation which suggests that "the translator can ... leave the writer [ofthe original] in
peace as much as possible and bring the reader to him" (Schleiermacher 1838:219;
trans-lation from Wilss 1982:33; cf. also Goethe 1813). As Schleiermacher points out,
this means that "the translator tries to let his own work substitute for the reader's lack
of comprehension of the original language" , but in doing so the translator transports the
reader to the 'location' ofthe original "which, in all reality, is foreign to him" (Schlei-
ermacher 1838:219; translation from Wilss, loc.cit.; italics my own). Schleiermacher
does not stress that this may mean considerable work on the part of the receptor lan-
guage audience.
176 Translation and Relevance

The reason for this is that (2) also allows for other implicatures, like:

(5) (a) John eats noisily like a pig.

(b) John eats greedily like a pig.

As relevance theory points out, the communicator can intend to convey all
of these assumptions weakly, without intending to convey anyone of them
strongly. This simply cannot be captured by explication - it would tend to
distort the communicator's intention either with regard to the range or the
strength of the implicatures, or both.
This distorting influence is also the deeper reason why the explication
considered by Yuasa in his translation of Bash6' s haiku (cf. chapter 6, pp.
139-43) is undesirable in direct translation. By explicating the word 'silence',
for example, Yuasa directs the audience's interpretation in a particular di-
rection and thus precludes them from considering the full range of
implicatures that the original would have allowed in the original cognitive
environment. All this confirms Adams' observation that "one word with
twelve important overtones just isn't the equivalent of twelve words" (1973:9).5
In fact, this brings out afresh the general link-up between the require-
ment of preserving all and only the 'communicative clues' of the original
and that of preserving the original interpretation. The very nature of' com-
municative clues' is to guide the audience to the intended interpretation,
hence different clues will lead to different interpretations - if the contextual
information used is the same.
Having discussed these points relating to the requirement that direct trans-
lations need to be interpreted with regard to the originally envisaged context,
we still need to consider the other basic condition: that is, the presumption
of complete interpretive resemblance with the original. As we shall see, this
question is dealt with best as part of a more general question about how
translation generally can be a successful means of communication. However,

5 Adams illustrates this point beautifully with a discussion of the title of Stendhal's

novel La chartreuse de Parme and the possibility of translating it into English (1973:9f.).
Having surveyed its rich nuances at some length, he concludes: "Now it's no great
problem to write a paragraph explaining, more or less after this fashion, the implica-
tions of 'Chartreuse de Parme', but to find three or four English words which will, in
combination, produce something like this effect, is the literary translator's overwhelm-
ing dilemma - and all the choices open to him are in various ways and for various
reasons impossible" (1973: 10). Adams does not take into account that the problem here
is one not only of language differences, but of differences in context.
A Unified Account of Translation 177

before turning to those wider issues, let us look at a group of translational

phenomena that may not seem to fit very well under either indirect or direct
translation.

Partial Resemblance in Linguistic Properties as

Translation?

The most representative class of the translational phenomena we want to

look at here are glosses or 'back-translations'. We have used them several
times in this study, as in the following Greek example from the beginning
of chapter 6 (cf. p. 130):

kai heteron epesen epi teen geen teen agatheen

[and other it-fell on the ground the good]

The question is: how does this fit into our relevance-theoretic framework?
It does not seem to fall under either indirect or direct translation in that
whatever explicatures or implicatures of the original it might convey, they
are clearly not conveyed in the least costly way, even if processed in the
original context. In fact, depending on the degree of difference in the struc-
ture of the language, a gloss may be quite unintelligible by itself, and
therefore glosses tend be used mostly in conjunction with freer renderings.
More importantly, it is not clear that the purpose of the above gloss had
much to do with the originally intended interpretation at all. As will be re-
called from the discussion there, its point was rather to convey a point of
syntactic structure to readers not familiar with the Greek language, espe-
cially to make perspicuous to them the linear ordering of the article teen
and the adjective agatheen after the noun geen. There is no reason to be-
lieve that it was part of Luke's informative intention to communicate these
syntactic points, that is, these points were not part of the interpretation of
the original - though they were used to convey it - and therefore do not fall
under interpretive use between utterances. Yet, clearly, resemblance did play
a crucial role - except that it was resemblance in linguistic properties, in
this case resemblance in word order, that counted: by arranging the English
words in the same sequence as that found in the Greek text I drew attention
to the word order in the original.
Of course, the resemblance does not have to be in word order; often
178 Translation and Relevance

glosses reflect the morphological structure of the original, sometimes also

the lexical structure of idiomatic expressions. Zukofsky and Zukofsky's
(1969) 'phonemic translation' ofCatullus, mentioned in chapter 5, (p. 136),
also falls under this category, being based mainly on resemblance in phono-
logical features. The fact that this case is a rare exception rather than the
rule follows naturally from the principle of relevance. There would presum-
ably be few readers for whom similarity with the sounds of an unknown
language would achieve adequate contextual effects.
These kinds of translations are therefore comparable to direct quota-
tions, which are also defined in terms of resemblance, in that case complete
resemblance, in linguistic properties. (Cf. chapter 6, p. 132 and pp. 169-71,
this chapter). This exploitation of resemblances in non-interpretive proper-
ties does not pose a problem for relevance theory; as Sperber and Wilson
(1986a) have pointed out, "In appropriate conditions, any natural or artifi-
cial phenomenon in the world can be used as a representation of some other
phenomenon which it resembles in some respect" (Sperber and Wilson
1986a:277).
Furthermore, an interlingual communicator can make use of both inter-
pretive and non-interpretive resemblance in the same utterance. This
approach is often followed in translations designed to help students learn
the original language. Savory characterizes this situation as follows:

The student is best helped by the most literal translation that can be
made in accurate English; it helps him to grasp the implications of
the different constructions of the language that he is studying ...
(Savory 1957:58f.)

It is clear that the feasibility of this approach will be strongly determined

by the degree of structural similarity between the two languages in ques-
tion. The more they differ in structure, the less it will be possible to combine
the demand for resemblance in structure with that for intelligibility.
We find, then, that the interlingual exploitation of resemblance in lin-
guistic properties is provided for in the relevance-theoretic framework. A
receptor language utterance can be used in virtue of resembling the original
utterance in linguistic properties, and such linguistic resemblance can also
be used in combination with interpretive resemblance.
A Unified Account of Translation 179

Translation, Relevance and Successful

Communication

In this section we want to address the issue of how translation relates to

successful communication. Looking into the literature one is struck by wide-
spread scepticism about how successful translation is or even can be. Thus
in Newmark's view "the main reason for formulating a translation theory,
for proposing methods of translation ... is the appalling badness of so many
published translations ... Literary or non-literary translations without mis-
takes are rare" (1988:5). And he quotes the article on translation in the
Encyclopaedia Britannica (of 1911) as saying that, "Most versions of mod-
em foreign writers are mere hackwork carelessly executed by incompetent
hands". Adams' book is significantly entitled Proteus, his lies, his truth,6
and in the preface he refers to "whatever falsehood a translator may be forced
to put on his title page - Homer's Iliad, Flaubert's Madame Bovary, the
Bible" (1973:ix). This skepticism is epitomized in the widely-quoted Italian
proverb 'traduttore, traditore' - 'translator, traitor'.
We shall first highlight some of the factors that endanger the success of
translations, and then consider what help relevance theory can give in deal-
ing with such problems.

Risks offailure in translation

The causes of failure in translation are varied, ranging from misunderstand-

ings of the original to insufficient mastery of the receptor language. In the
literature by far the greatest attention seems to be given to matters of lan-
guage difference, that is, to problems that arise from lexical and structural
differences between languages. Textbooks on translation thrive on exam-
ples where translations failed because the differences between the languages
had not been observed properly by the translator, and there is little point in
either reproducing those examples or adding new ones to the list. Rather,
what I should like to do here is to bring out some sources of miscommuni-
cation related to the distinction between direct and indirect translation.
It follows from the causal interdependence of stimulus, context and in-
terpretation that the greater the difference between the original and receptor

6 The following titles on translation are also suggestive: The manipulation a/literature

by Hermans (1985), and Ubersetzen als Hochstapelei [Translation as a fraud] by Willson

(1982).
180 Translation and Relevance

context, the more difference there will be between a direct translation and
an indirect translation of the same text.
A clear example of this is provided by a comparison of a translation of
Matthew 9:6 into the Ifugao language, reported in Hohulin (1979:33), with
its rendering in the Revised Standard Version (RSV).7 This passage con-
cerns the incident with a man who is paralyzed where Jesus is challenged
about his right to forgive sins, and where he responds by healing that man.

RSV Ifugao
"But that you may know "But I will prove my
speech to you. You know
that it is God alone who re-
moves sickness. You also
know that it is God alone
who forgives sin. And so,
if I remove the sickness of
this person and he walks,
that's the proof
that the Son of man has that I, the Elder sibling of
authority on earth to for- all people, I also have the
give sin" ability to forgive sin."
Jesus turned toward the
- he then said to the para- paralytic and said, "Get up,
lytic - "Rise, take up your take your stretcher and go
bed and go home." to your (pI) house!"
(Hohulin 1979:33)

The rendering of the Revised Standard Version can be taken here as an

instance of direct translation, relying as it does on the audience to supply
the contextual information necessary for understanding this passage.
Hohulin's rendering seems to follow the lines of indirect translation. On
the ground that the lfugao audience is unable to supply all that is implied
here and hence has great difficulties in understanding this passage, Hohulin
explicates a number of implicatures, mostly implicated assumptions. As can
be seen from the parallel presentation above, the resultant translations dif-
fer considerably from each other - for example, in the fact that in the Ifugao
rendering Jesus seems to be explaining more to the people than in the Re-
vised Standard Version.

7 For a more detailed discussion ofthis example see Gutt (l987a).

A Unified Account of Translation 181

It is not difficult to see that miscommunication is likely to arise when

these two kinds of translation are mistaken for one another. The cases that
seem to have drawn the most attention are those where direct translations
have been interpreted as if they were indirect translation. Part of the reason
for this is probably that this kind of misinterpretation is particularly com-
mon. Unless told otherwise, an audience will naturally assume that a
communication addressed to it will yield the intended interpretation by op-
timal processing. Hence it will go ahead and interpret the text in question
with the most highly accessible context available to it. In cases where the
cognitive environment ofthe receptor language audience differs significantly
from that of the source language audience, a direct translation interpreted
along these lines will quite naturally be misunderstood. The misinterpreta-
tions of Mark 2:4 (p. 94) and Ruth 1:22 (pp. 94-5) discussed in chapter 4
above are cases in point.
The converse problem is addressed less often; that is, the case where
problems of interpretation arise because an indirect translation is mistaken
for a direct one. One voice that has raised this issue in Bible translation is
Carson in his criticism of the 'dynamic equivalence' approach. Carson
observes:

Ifwe followTEV's [Today English Version] 'police' orNEB's [New

English Bible] 'constable' in Matthew 5:25, are we not unwittingly
fostering images of a gun-toting officer or an English bobby? Per-
haps these cases don't matter. But many cases have stings in the tail.
If for instance we replace 'recline at food' or 'recline at table' with
'sit down to eat', we are going to have a tough job imagining how
John managed to get his head on Jesus' breast. Preservation of what
is to us an alien custom, reclining at tables, makes it possible to un-
derstand a later action, John placing his head on Jesus' breast (Carson
1985:209).8

Apart from this problem of giving rise to potential inconsistencies within

the translated text, Carson points out that there is the important issue of
historical reliability when dealing with cultural differences that are obsta-
cles to ease of comprehension:

8This also creates at least an oddity of expression in the Good News Bible in Luke 7:38.
Having said that "Jesus sat down to eat", GNB goes on to say that a woman came and
"stood behind Jesus, by his feet" - which makes good sense if one knew that Jesus was
actually lying down, but seems rather strange with regard to someone sitting at table.
182 Translation and Relevance

Row such problems are resolved may depend to some extent on the
literary stage of development of the receptor group, but even if the
group is coming across the printed page for the first time, and enjoys
virtually no comprehension of cultures other than their own, it must
be remembered that this receptor group will likely use this new trans-
lation of the Bible for decades to come, maybe a century or two.
During all that time, an increasing number of this receptor people
will be exposed to new cultures and education. Row well will the
Bible translation serve then? Christianity is a religion whose roots
are deeply embedded in the particularities of history, and our trans-
lations must not obscure that fact. (Carson 1985:209f.)

In other words, what will the audience think of the reliability of a transla-
tion that presents people as sitting when they were actually lying down?
While this case may be oflittle significance in itself, dealing with a point
of marginal interest to the average reader, the Ifugao example (p. 180 above)
raises more serious questions. Here the readers will almost certainly treat
the statements "You know that it is God alone who removes sickness. You
also know that it is God alone who forgives sin" as sayings of Jesus attested
by Matthew, and on a level of authority and authenticity with any other of
his attested sayings, like "I am the light of the world". One wonders what
their reaction will be once they discover that these statements were placed
in Jesus' mouth by the translator. 9
Barnwell documents problems of acceptability that seem to have a bear-
ing on this issue:

For example, in Nigeria until recently, the versions used in churches

have been the KJV [King James Version] or RSV [Revised Standard
Version], or the rather literal translations in Igbo, Yoruba, or Rausa.
Because people have already become used to the form of the Scripture
in one of these more literal versions, when an idiomatic (meaning-
focused) translation is made into the local language there is often a

9 Interestingly this point is conceded by Deibler in his critical response to Gutt (1987a),

and he suggests that "it would now be considered better to put this information into the
editor's words" (1988:32). This differentiation between the words of Jesus and the words
of Matthew is surprising because according to Deibler the real question in explication
of implicit information is "is it part of what the author intended and expected to be
understood as part of his message or isn't it? That is the question which needs to be
considered" (1988:30). Ifthis were the whole issue, then surely the explication should
become part of Jesus' speech since it is he who must have intended to convey those
implicatures in the first place.
A Unified Account of Translation 183

negative reaction. It seems as if the Bible has been changed. If any

part of the meaning is made explicit, people feel that something has
been added to the text. They feel strongly that the form of the origi-
nal should not be changed. (1983: 19)

In her analysis of the situation, Barnwell lists six different possible

causes of the problem. Interestingly, four of them refer to an inadequate
understanding of the 'form-meaning' distinction assumed by the 'idiomatic
approach'. In three cases the audience is seen to lack an adequate grasp of
this distinction, in one case the translator. 1o
The first cause concerns the people's reverence for the Bible; Barnwell
notes this as a positive factor but says that "The problem comes when 'ac-
curacy' is interpreted to mean 'identity of form' rather than 'identity of
meaning'" (1983:20).
The second cause is identified as the people's emotional attachment to a
particular version.

The problem here is that people have become attached to the famil-
iar written form of the words rather than to the message itself. They
are not able to distinguish between the message and the grammatical
and lexical form by which it is expressed. (Bamwell1983 :20, italics
as in original)

In a recent article, Nida makes a similar observation:

One of the greatest surprises for Bible translators is to find that a

perfectly intelligible translation of the Scriptures may not be accept-
able ... In fact, many people prefer a translation of the Scriptures
which they only partially understand. For example, the archaic and
obscure words and grammatical forms of the King James Version
seem to many people to fit the esoteric nature of the contents and to
lend authority to the text. (1988:301)

The third cause is "a lack of understanding of the principles which un-
derlie the apparent changes. People are not consciously aware of linguistic
differences between languages. They assume that what can be said in a cer-
tain way in one language can be said in the same way in any other language.

10 The remaining two causes are the misunderstanding of the Bible as a "magic" book

not actually meant to be intelligible and the assumption that "it is the pastor's job to
explain the Bible" (Bamwelll983:20).
184 Translation and Relevance

The distinction between the meaning and the form oflanguage is often con-
fused even by some with high educational background" (Barnwell 1983 :20).
The last cause concerns actual mistranslations:

Sometimes translations are too free and are not accurate in commu-
nicating the original meaning. In aiming for natural expression of
the message in the receptor language, the translator may have lost
the essential focus on exact equivalence of meaning. Sometimes
changes of form are made unnecessarily. (Barnwell 1983 :20, italics
as in original)

Now having been involved myself for about ten years in Bible transla-
tion in the Third World and in the training of Bible translators there, I am
aware of the problems of interference between languages that lead to the
carry-over of source language structures into the receptor language. And
perhaps there are people who reject modem English versions of the Bible
simply because they do not sound obscure enough. However, in view of the
inadequacies inherent in the 'form-meaning' distinction that we looked at
earlier on, I should be reluctant to see the problem of acceptability as lying
quite so much on the receptor side as both Barnwell and Nida seem to sug-
gest. Could it not be that the receptor language audience reacts to 'changes
of form' that are, in fact, 'changes of meaning'?
Given that people are fully capable ifmaking good use of the difference
between 'implicit' and 'explicit' information in everyday communication,
it does not seem surprising that in matters of translation, too, they have
clear intuitions that the two are not the same. They know that having meant
something is not necessarily the same as having said it, and vice versa.
In view of this, I should take up Barnwell's third cause (see above), but
with a different emphasis: yes, on the receptor language audience's part
"there is a lack of understanding of the principles of translation which un-
derlie the apparent changes". However, rather than holding responsible for
this an inadequate understanding ofthe 'form-meaning' distinction, I should,
more neutrally, begin with the observation that there is a mismatch in ex-
pectations about the translation: the receptor language audience expected a
translation of one kind, and they received one of another kind. Put more spe-
cifically, the audience expected the translation to show resemblances of one
kind, and it did not show those resemblances, or perhaps not enough of them.
Conflicts about acceptability are, naturally, unpleasant, but perhaps
more serious are those instances where the receptor language audience is
A Unified Account of Translation 185

monolingual and unable to detect this difference in expectations, because

the translation is the only access it has to the original. Such situations can
give rise to latent misinterpretations that may come to light only in the course
of time, as people obtain other translations of the same work, or learn the
source language.
Thus it may come as a surprise to the readers of a number of different
translations of the New Testament that Jesus had one major theme in his
preaching and teaching ministry, basileia tau theou, 'the Kingdom of God',
this expression occurring more than a hundred times in the four gospels. ll
Thus in the German 'dynamic equivalence' translation, three distinct ren-
derings are used, depending on the context: in "contexts where the emphasis
is on something happening at the present time, the phrase 'God wills to
establish his rule now' is used .... In contexts where the completion of the
event is seen as future ... the words 'When God completes his work, then'
are used. In passages concerned with seeing the basileia or entering into the
basileia we have translated by 'God's new world'" (Kassuhlke 1974:237).
In a draft translation of Luke into Gurung, Glover (1978:231) uses a still
larger number of different renderings, which 'fall into four main groupings':

l. God's power at work in the world.

2. The personal response to God, in obedience and receiving
blessing.
3. God's future open ruling of the world.
4. The ultimate blessings of God's rule in heaven.

The following is an example of grouping 2 (Luke 7:28):

RSV [Revised Standard Version]: Among those born of women none

is greater than John: yet he who is least in the kingdom of God is
greater than he.

Gurung: John is the greatest of all people. Even though he is, who-
ever, after seeing my deeds and hearing my words, obeys God
according to what I say, even though that person be small, he will
receive greater blessing than John. (Glover 1978:232, italics as in
original)

Two renderings in grouping 4 read like this:

11 According to Ott (1984:2).

186 Translation and Relevance

RSV: Blessed is he who shall eat bread in the kingdom of God.

Gurung: Those who get to eat God's feast in heaven will be very
happy.
(Glover 1978:235, italics as in original; Luke 14:15)

RSV: Blessed are you poor, for yours is the kingdom of God.

Gurung: Blessing has come to you poor people, because since you
look to God, he will give you his full blessing.
(Glover 1978:236, italics as in original; Luke 6:20)

With regard to our discussion here, the following part from Glover's
conclusion is particularly interesting; having noted that to make the mean-
ing clearer, the Gurung renderings tended to be considerably longer than
those in the Revised Standard Version, Glover asks:

But is this not necessary? "It is quite certain that the Kingdom of
God was the central message and proclamation of Jesus" (Barclay,
page 63), yet it has remained a puzzle to many people with centuries
of Christian heritage and the benefit of years of preaching and teach-
ing. It is more than time for translators to grapple with this central
message of Jesus, all the more so when their prospective readers do
not have commentaries and other theological works to consult.
(Glover 1978:236)

One can understand Glover's concern and sympathize with him, espe-
cially with regard to his last point. However, the question that remains is:
what does the receptor language audience expect - is it a direct translation
that is not easy to understand but preserves, for example, the fact that Jesus
had as his central message 'the Kingdom of God', or is it an indirect trans-
lation, that is easier to understand, but loses the 'Kingdom of God' as a
focal concept? It is only too easy to see how mismatches in expectations
can provide fuel for those who hold the view 'traduttore, traditore.'
This problem of clashes of expectations shows itself with particular clarity
in the case of Bible translation, because here the urge to communicate as
clearly as possible is equally strong as the need to give the receptor lan-
guage audience access to the authentic meaning of the original, unaffected
by the translator's own interpretation effort. Since the differences in cogni-
tive environment between the source language and the receptor language
audiences are generally great, these two objectives are bound to clash. If
A Unified Account of Translation 187

the translator wants to make the translation clear to the receptors in their
particular cognitive environment, he will need to bring out explicitly (parts
of) his own interpretation or that of some other authority; if the translator
aims at authenticity and hence is concerned to keep the influence of his own
interpretation on the translated text to a minimum, his translation will in
many cases prove difficult to the receptor language audience. 12
While this problem presents itself in a very crystallized form in Bible
translation, it is by no means limited to this area. Examples are found in
secular translation as well. Thus consider the following translation into
German ofC. Doyle's A scandal in Bohemia. The passage is taken from the
beginning of the story, where Doyle describes Holmes' relationship to Irene
Adler; part of this discourse deals with Holmes' contempt for feelings. For
ease of comparison the original text is also given.

1. He was, I take it, the Erwar die perfektionierteste

most perfect reasoning and Denk- und Beobachtungs-
observng machinve that maschine, die die Welt
the world has seen, but as gesehen hat, aber als
a lover he would have Liebhaber hiitte er seme
placed himself in a false Rolle verfehlt.
position.
2. He never spoke of the
softer passions, save with
a give and a sneer.
3. They were admirable
things for the observer -
excellent for drawing the
veil from men's motives
and actions.
4. But for the trained
reasoner to admit such in-
trusions into his own
delicate and finely adjusted
temperament was to intro-
duce a distracting factor
which might throw doubt
upon all his mental results.

12 The translator can make the interpretation of a direct translation easier by providing

part ofthe contextual information needed through supplementary channels of commu-

nication, such as introductions, notes, pictures, glossaries and the like. A strong advocate
of this view in Bible translation is Peacock: "The day is gone when the bare text can be
put in the hands of the reader ifhe is expected to nnderstand its message. He needs help
in bridging the temporal and cultural gap between the biblical period and today .... This
means Bibles with notes, ... " (1981:8). For further proposals cf. also Gutt (1988b).
188 Translation and Relevance

5. Grit in a sensitive instru-

ment, or a crack in one of
his own high-power lenses,
would not be more disturb-
ing than a strong emotion
in a nature such as his.
6. And yet there was but Es gab nur eine Frau fUr
one woman to him, and that ihn, und diese Frau war
woman was the late Irene Irene Adler. 14
Adler, of dubious and
questionable memory. 13

In a way the German translation here is a counter-example to the Ifugao

passage we looked at above (cf. p. 180); as the blanks in the German text
column show, four whole sentences have been simply omitted in the trans-
lation, and part of sentence 6 has also remained untranslated. The point that
makes this example particularly pertinent to our discussion is that this par-
ticular version is expressly presented as an 'unabbreviated edition'
['ungeklirzte Ausgabe']' It is easy to see that this could give rise to disap-
pointments on the part of the receptor audience, especially since similar
omissions occur in other parts of the translated text. Apart from the loss in
content, this translation also misrepresents Doyles' verbose styleY
To take another example from literary translation, this time involving con-
siderable differences in cognitive environment, let us look at von Tschamer' s
discussion of translations of Chinese poetry into German. For one song,
taken from the Shi-king (Kuo Feng VI, 8), von Tschamer compares three
different German renderings. The rendering that is found most adequate by
von Tschamer is that of Victor von Strauss, who expressed his main objec-
tive as follows:

13 Sir Arthur Conan Doyle, A scandal in Bohemia. In The Penguin complete Sherlock

Holmes. Penguin Books, London, 1981, pp. 161-75, p. 161.

14 Sir Arthur Conan Doyle, Skandal in Bohmen. In Sherlock Holmes' Abenteuer. Ullstein,

Frankfurt, 1967, pp. 9-30; p. 9. 'Aus dem Englischen ubersetzt von Hans Herlin.
Ungekfuzte Ausgabe.'
15 The omissions in themselves are easily explained in terms of relevance: verbosity

requires more linguistic processing than is justified by the contextual effects achieved.
However, the translator would have done well to inform the audience that he was pre-
senting a shortened version to them.
A Unified Account of Translation 189

Of course my first desire was to translate everywhere faithfully to

the meaning, but then also as literally as possible. (von Tscharner
1969:249f., translation my own)

That is, Strauss was apparently concerned mainly with a direct translation.
Von Tscharner contrasts this with a reworking ofthis song by Ruckert, which
he evaluates as follows:

It is a perfectly German poem of the subjective-concrete variety,

written in Ruckert's smooth well-known way. A free poetic develop-
ment of the given external motif, which also appears to have been
altered ... The added pictorial apostrophes 'mein Licht' ['my light']
and '0 Freudenstern' ['oh star of joy'] addressed to the absent friend,
beloved or spouse, having nothing Chinese to them either. (von
Tscharner 1969:252, translation my own)

Thus von Tscharner, whose own preference is clearly for direct transla-
tion, reacts negatively to a fairly loose interpretive rendering - precisely
because it fails to preserve properties of the original regarded as typical of
this kind of Chinese poetry, and hence worth preserving. 16 More specifi-
cally, von Tscharner reacts to interpretive translations ofthese songs because
he feels that it is a strong characteristic of the 'spirit' of Chinese poetry to
leave the interpretation as open and indetenninate as possible, and to prefer
"subjectless indetenninacy, the ambivalence latent in a concise spoken word
as well as in an unspoken 'nothing'" (1969:270, translation my own). Von
Tscharner feels that "we deny this very spirit when in the transfer of Chinese
lyric we "put into the lines" that which we should read "between the lines",
when we spell out with an enthusiasm for words and pictures the potential
contained in a single, concise word" (1969:270, translation my own).
Whatever we think of von Tscharner's personal viewpoint, the point
here is that considerable misunderstandings can occur when it is not clear
what kind of resemblance is intended. Thus a reader taking Ruckert's ren-
dering as a direct translation might come to the conclusion that Chinese
poetry is not very different in style from Gennan poetry, or, to go back to
our first point, a reader expecting a rendering that gives him access to the

16 Von Tscharner shows his preference when he says that "for us only the second of

Goethe's maxims of translation can really be valid. The ideal, to express ourselves in
the German language as the Chinese in his, is, however, uureachable" (von Tscharner
1963:270, translation my own). The maxim referred to is the one that requires "that we
move over to the stranger and make ourselves familiar with his situation, his way of
expression and his peculiarities" (Goethe 1813:705, translation my own).
190 Translation and Relevance

interpretation of the poem might be disappointed because it seems lacking

in coherence and clarity of expression.
Thus we see that the problem of misinterpretation created by miscon-
ceptions about the kind of resemblance actually intended is a general one.
The question is: what, if anything, can be done about it?

Making intentions and expectations meet

In relevance theory the need for the audience to recognize the degree of
resemblance intended in interpretive use is part of the general requirement
for successful communication - which is "to have the communicator's in-
formative intention recognised by the audience" (Sperber and Wilson
1986a:16l). How can the audience recognize the communicator's informa-
tive intention?
As will be recalled from chapter 2, (p. 34), ostensive communication is
an asymmetrical process - it is the communicator's responsibility to ensure
that the audience can recognize his informative intention. Since the transla-
tor is an ostensive communicator, too, this means that the responsibility for
avoiding misunderstandings and other commmunication breakdowns in
translation lies with him. What does this responsibility entail?
Crucially, the translator's responsibility begins with the formation of his
informative intention. As we argued in chapter 4, (pp. 103), the translator
needs to clarify for himself whether his informative intention is, in fact,
communicable, that is, whether he can reasonably expect the audience to
derive this interpretation in consistency with the principle of relevance. Thus,
the translator is confronted not only with the question of how he should
communicate, but what he can reasonably expect to convey by means of his
translation.
The answer to this question will be determined by his view of the cog-
nitive environment of the target audience, and it will affect some basic
decisions. It will, for example, have a bearing on whether he should engage
in interpretive use at all whether descriptive use would be more appro-
priate. As we saw in chapter 3, there are instances where the fact that the
information given in the receptor language text interpretively resembles an
original in another language is quite irrelevant - in which case the 'trans-
lator' would simply engage in descriptive use. This, in tum, may have
consequences for many decisions he will have to make later in the produc-
tion of the receptor language text.
A Unified Account of Translation 191

In other cases the translator may judge it relevant to the audience to

recognize that the receptor language text is presented in virtue of its re-
semblance with an original in another language. In such situations he will
have to consider further what degree of resemblance he could aim for, be-
ing aware that communicability requires that the receptor language text
resemble the original "closely enough in relevant respects" (Wilson and
Sperber 1988a:137). To determine what is close enough resemblance in
relevant respects, the translator needs to look at both the likely benefits,
that is, the contextual effects, and also at the processing effort involved for
the audience. Thus he will have to choose between indirect and direct trans-
lation, and also to decide whether resemblance in linguistic properties should
be included as well.
A good example of the need to carefully consider processing cost in
translation is provided by the experience of o 'Flaherty (1987), who at one
point experimented with a new way of translating Sanskrit literature into
English. She reports:

One of the first articles I published, in 1968, was 'A New Approach
to Sanskrit Translation', in which I called upon the reader to do a
great deal of work indeed (1987: 123).17

Essentially this new approach was "a very literal, word-for-word trans-
lation, retaining the long, multiple compounds, and bracketing pairs of words
to represent the puns and doubles entendres with which Sanskrit abounds"
(1987:123). She also added "elaborate notes", and she "took pains to justify
this approach" in the introduction (1987:123). One ofthe examples she gives
is the following:

The moon, grasping with rays like fingers the hair-mass-darkness,

kisses the night

night -face with closed-

- mouth bud-made-
evening

Lotus eyes. (O'F1aherty 1987:123)

17 The author here refers to 0' Flaherty (1971).

192 Translation and Relevance

o 'Flaherty was fully aware that the result was "a highly unorthodox fonn
of English verse", but she argued that "the English language can certainly
strain to accommodate it" (1987: 124, italics as in original). The following
part of her reasoning is particularly interesting:

I had reasoned that the people who were likely to read translations
of Sanskrit poetry were not the same people who read the sort of
novels that one bought in airports; they were people who were genu-
inely interested in a foreign culture and who were willing to make a
major investment of their intellectual energy in this enterprise.
(1987:124)

However, she came to realize that this reasoning was mistaken:

I failed to realize two things: that anyone who was interested in fight-
ing through that sort of translation would be likely to go ahead and
learn the original language; and that people in airports were quite
capable of doing that, too. (O'Flaherty 1987: 124, italics as in original)

We shall not go on to trace the further development of o 'Flaherty's notion

of translation. Her intention was to get the receptors as close to the original
as possible - what she had to learn was that people may not be willing to
pay the price for this. Since her translation violated grammaticality, the au-
dience could not take full advantage of their linguistic competence that
provides a very fast and efficient way of developing schemas for propositional
fonns from linguistic input. As 0 'Flaherty herself realized, the increase in
processing effort this caused approximated that of actually learning another
language, and for most readers this was far above the amount of effort they
were willing to invest. The point here is that her experience brings out the
importance of assessing the processing-cost element in translation.
Most importantly, in order for the communication to succeed, the as-
sumptions about the nature ofthe communication act - for example, whether
it is meant to be understood descriptively or interpretively, and, in interpre-
tive use, what the intended degree of resemblance is - must be shared by
the translator and his audience. Furthennore, because of the asymmetrical
distribution of responsibilities in ostensive communication, the burden will
be on the translator: he has to ensure that it is clear to his audience what his
intentions in this respect are, and also that his intentions meet the audi-
ence's expectations.
In some cases these assumptions may be clear from the audience's request
A Unified Account of Translation 193

for the translator's assistance. Thus Honig and Kui3maul consider cases where
the translator is told, for example, to 'translate' a business letter but telling
only "under what conditions our offer will be accepted" (1984:27), or where
the translation of a paper is to omit the embellishments of the original and
to concentrate on reporting what the paper said about a new production
process.
In other cases these assumptions may be clear from the 'label' with which
the receptor text is presented. Thus the fact that the receptor language text
is labelled 'translation' may suffice to make clear what degree of resem-
blance is intended. However, such labelling can be relied on only under the
condition that the assumptions conveyed by this label are, in fact, the same
for both translator and audience. Given the divergence of ideas among both
experts and laymen about what translation should be like, reliance on this
label alone for this important part of the communication process seems risky
indeed.
Therefore, in many cases and especially when addressing a wide or var-
ied audience, the translator will do well to make his intentions explicit. Thus
the practice of translators to explain their 'translation principles' in a fore-
word makes good sense in our relevance-theoretic framework and could
probably be used more widely to make translations successful.
The importance of ensuring that the intended resemblance be known to
both parties, and the danger of relying on tacit assumptions in this matter,
can hardly be overemphasized. I believe that insufficient awareness in this
area has contributed greatly to the misunderstandings, unjustified criticism,
confusion and frustration that tend to accompany translation. Mismatches
in these expectations do matter, sometimes only a little, but sometimes very
much so. One of the clearest examples that I know of has recently been
documented by Dooley (1989) regarding a translation of the New Testa-
ment into the Guarani language of Brazil. 18 A draft translation following
the idiomatic approach had been completed in 1982, and a number of cop-
ies were given out to be tested on a limited scale. After a year's testing, the
result was that the church decided that the translation had to be changed.
Dooley reports that, "The changes were so extensive that virtually every-
thing had to be translated and keyboarded again" (1989:51).

18 This concerns the Mbya dialect, of which Dooley reports: "Speakers of the Mbya

dialect number around 7,000 and live in eastern Paraguay, northern Argentina, and south-
ern Brazil. The 3,000 Brazilian Guarani, the principal audience ofthe New Testament
translation in question, live in at least 35 locations in six states" (1989:49).
194 Translation and Relevance

Dooley explains that "From the Guarani point of view, the rationale be-
hind the changes in translation style was that the Scriptures in Guarani should
be clearly seen as a faithful translation of the high-prestige Portuguese ver-
sion ... What the Guarani expect is that the meaning in their translation
correspond, in a fairly self-evident way, with what they find in the Portu-
guese" (1989:53). More specifically, the correspondence they look for "in
each passage involves those levels [of grammatical structure] at which com-
parison with the Portuguese is most likely" (1989:53), and Dooley illustrates
that this may sometimes be on the clause level, at other times on the word
level and so forth. Interestingly, among other changes, "much implicit in-
formation that had been made explicit in the text was relegated to a footnote,
a picture, the glossary or eliminated altogether. Such implicit information,
when it was made explicit in the text, came to be viewed as 'explanation' of
the text per se" (1989:52).
Looked at from the perspective of relevance theory, it seems that there
was a discrepancy between the resemblance assumed by the translator's
earlier version and the resemblance expected by the Guarani audience.
Whereas the translator assumed that a kind of indirect translation would be
appropriate, that is, a translation that would communicate implicatures with
ease to Guaranis with little knowledge of the original background, the
Guaranis themselves seemed to have looked for a translation more along
the lines of direct translation, possibly accompanied by resemblance in lin-
guistic properties as well. This mismatch in expectations led to a breakdown
in communication in that "the use of the Guarani translation was falling
behind the use of the Portuguese" (1989:54) - that is, there was a tendency
to discontinue the use of the Guarani translation.
This example is interesting not only because of the extremity of the situ-
ation - requiring a re-translation rather than a revision - but also because it
brings out again the need to consider both processing effort and benefits
together: the earlier, 'idiomatic' version would no doubt have involved much
less processing effort for the audience; however, it failed because it was not
felt to resemble the 'original' - in this case the Portuguese translation -
closely enough in those respects that the receptors judged to be relevant.
Interestingly, such differences in expectations matter not only with re-
gard to texts to which particular importance is attached, such as sacred texts
or great literary works. Thus the publisher of the German version of the
Asterix comic series notes in the introduction to the tenth anniversary edi-
tion of Asterix der Gallier that his publishing house was surprised by the
A Unified Account of Translation 195

criticisms made of the translations used in earlier editions:

We would never have thought that one would apply to a comic the
same standards of translation as to Moliere or Proust. (Kabatek
1988:3, translation my own)

However, they took these criticisms seriously enough to incorporate a number

of changes into the tenth anniversary edition.
An alternative strategy is to try to change the audience's expectations.
Thus one of the solutions recommended, for example, by Barnwell is to
"provide opportunities for people to come to understand the principles which
are being used in making the translation" (1983:21). Envisaged are such
means as discussion, training courses and seminars. This strategy has been
used with success in some situations, as reported by Payne (1988), for ex-
ample. However, the use ofthis strategy requires rather special circumstances
and may not be practicable as a general solution. It is likely that it will more
often be the translator who has to change ifhe wants to be successful in his
communication effort.
Now all this may seem very clear in principle: for communicative suc-
cess, the translator has to ensure that his intentions and the expectations of
the audience will be in line; so he chooses a suitable approach and makes
sure that the audience is aware of and in agreement with his choice. But
what if it turns out to be impossible to follow the approach selected consis-
tently throughout the translated text? In fact, given the 'messy' reality of
languages, is it not unrealistic to expect that any particular approach will
work consistently for even one text? Will not grammatical differences and
lexical mismatches frustrate any such attempt sooner or later, even between
closely related languages?
These concerns are particularly strong in the case of direct translation
with its objective of achieving complete interpretive resemblance - which
brings us back to its second defining characteristic: when, if ever, can trans-
lations consistently achieve complete interpretive resemblance with the
original? There is no a priori reason why from a purely linguistic point of
view it should be possible to compose a receptor language utterance or text
that will be able to consistently direct the receptors to the originally in-
tended interpretation without distortion or loss. But if this is so - is not our
account of direct translation open to the charge of relying on an ideal and
saying little about the realities of interlingual communication?
Here several points need to be made. The first is that the defining
196 Translation and Relevance

characteristic of direct translation is not that it achieves complete interpre-

tive resemblance but rather that it purports to achieve it, that is, that it creates
a presumption of complete interpretive resemblance. This difference is not
just a fine point of philological interest, but rather basic in relevance theory,
paralleling, in fact, the difference between achieving and presuming opti-
mal relevance. 19 The presumption of complete interpretive resemblance in
direct translation gives the receptors important information about the in-
formative intention of the communicator. It entitles them to consider all the
explicatures and implicatures which they can recover with respect to the
original context as having been part of the intended interpretation of the
original. Thus this presumption is important for coordinating the audience's
expectations with the communicator's intentions.
Secondly, this presumption is of value not only where complete resem-
blance is, in fact, achieved, but also where the translator knows that it is not
achieved - for example, because linguistic differences between the two lan-
guages make this impossible. Being aware that there his translation will not
meet the expectations of the audience and hence mislead them, the translator
can consider strategies for preventing communicative failure, for example
by alerting the audience to the problem and correcting the difference by some
appropriate means, such as footnotes, comments on the text or the like.
Thirdly, there is no a priori reason why a translation should aim at com-
plete interpretive resemblance throughout. There can, of course, be cases
where this is required, but this would be a matter of circumstance rather
than of principle or theoretical necessity - the only principle to be kept in
mind throughout is the principle of relevance. Thus consistency throughout
a translated work may make the coordination between the translator's in-
tentions and the audience's expectations easier, but again the translator may
be able to achieve this coordination by other means, such as supplementary
notes and the like.
However, may the use of such supplementary means not lead to a trans-
lation abounding with footnotes, many of which bore the audience because
they are concerned with insignificant detail? This can, of course, happen
but alertness to consistency with the principle of relevance can again help
the translator avoid this problem; he should consider in each case whether
the correction will be adequately relevant to his audience, whether its ben-
efits will outweigh the processing effort it requires.

19 Cf. the discussion ofthis point in Sperber and Wilson (l986a: 158ff.).
A Unified Account of Translation 197

These same points apply to translation in general. By monitoring the

agreement or disagreement between the audience's expectations and what
his translation is likely to achieve, the translator can anticipate mismatches;
he can then consider further whether these mismatches are relevant enough
to require explicit treatment and if so, how they will be best treated. In all
cases he will measure the success of his translation by whether it enables
the receptors to recognize his informative intention.
Before leaving this topic of successful communication by translation, it
may be worthwhile briefly to address a phenomenon found in translation
that on the face of it seems to be at odds with our definition of successful
communication. This is the fact that translations can be, and often are, read
with enj oyment on the audience's part without them realizing that they are,
in fact, reading a translation rather than an originaPO In terms of our defini-
tion of successful communication, such instances cannot be considered
fully successful in that one of the intentions of the translator, that is, that his
work is to achieve relevance in virtue of its resemblance with an original,
has not been recognized by the audience. On the other hand, from the audi-
ence's point of view, they might well be satisfied with the result of this act
of communication.
This apparent ambivalence of the notion of 'successful communication'
can be accounted for readily in the relevance-theoretic framework. It fol-
lows straightforwardly from its basic claim that the only way in which an
audience can identity the intended interpretation of a stimulus is by the cri-
terion of consistency with the principle of relevance - it has no other, more
direct means of knowing what the speaker meant. This allows for the possi-
bility that, due to mismatches in context, the audience may arrive at an
interpretation which it finds consistent with the principle of relevance, and
hence assumes to be the speaker-intended one, but which, in fact, differs
from the speaker's informative intention. What this means is that the crite-
rion of consistency with the principle of relevance does not necessarily
reveal all instances of miscommunication. This seems true to our experience:

20 What I have in mind here differs from Mossop's point that even when overt markings
show that one is dealing with a translation, "their presence tends to disappear from
consciousness as the reader becomes absorbed in the text ... As long as one is looking
only at the translation, its English graphic appearance conjures up a picture of an Eng-
lish narrator" (1987:4). It is not necessary for the success of a translation that the target
audience be constantly aware that the original was written in a different language; what
is necessary, however, is that it takes the interpretation it derives to resemble that in-
tended by the original author.
198 Translation and Relevance

misunderstandings can go unrecognized for quite some time, not only in trans-
lation, but in communication in general - which is just another illustration
of the fact that the principle of relevance does not guarantee communi-
cative success.

Conclusion

Having surveyed the relationship between translation and successful com-

munication, we are now in a position to draw out some of the main points of
this study.
In this study I have tried to explore the possibility of accounting for the
range of phenomena commonly considered as translation in terms of the
general psychological endowments that, according to relevance theory, en-
able human beings to communicate with one another. Starting with instances
of translation where the relationship to the original is incidental rather than
crucial for the success of the translation, we found that these are naturally
accounted for in relevance theory in terms of descriptive use. Turning to
other instances of translation where resemblance with the original is impor-
tant, I argued that most of these instances are covered under the notion of
interpretive use already present in the theory. I also developed the idea that
interpretive use can account for the concern to preserve both meaning con-
tent and style. Realizing that ostensive stimuli also have intrinsic properties
quite apart from their use to convey meaning, we briefly considered the role
that resemblance in linguistic properties can play in translation. Again, we
found that this is a possibility provided and accounted for in the relevance-
theoretic framework.
In all instances we were able to account for the phenomena in question
without reliance on descriptive-classificatory theory, and indeed without
reliance on any translation principles or rules. In fact, I tried to show that
the principles, rules and guidelines of translation are applications of the
principle of relevance; thus the proposal is that all the aspects of translation
surveyed, including matters of evaluation, are explicable in terms of the
interaction of context, stimulus and interpretation through the principle of
relevance, a universal principle believed to represent a psychological char-
acteristic of our human nature. Thus the main contribution of this book is a
reductionist one on the theoretical level - issues of translation are shown to
be at heart issues of communication.
What I have not attempted to do is to show how all possible translation
A Unified Account of Translation 199

problems should be analysed or solved. Such an undertaking seems neither

possible nor necessary. It seems impossible because translation problems
are as varied as the languages, informative intentions and contexts that give
rise to them. It is unnecessary because the analyses follow from our un-
derstanding of communication in general; the better our insight into
human communication becomes, the better we shall understand transla-
tion problems. As far as solutions are concerned, they are determined by
context-specific considerations of relevance.
On the assumption that the findings of this exploratory study are indeed
valid for translation in general, the following important points can be drawn out.
Firstly, the translator must be seen and must see himself clearly as a
communicator addressing the receptor language audience: whatever his view
of translation, even if it is that of a 'phonemic translation' after the fashion
of Zukofsky and Zukofsky (1969), he always has an informative intention
which the translated text is to convey to the receptor language audience.
This recognition should help to raise the often bemoaned, low status of
translators as copyists, and prevent misunderstandings that arise from the
pretense that there is a direct act of communication between the original
source and the receptor language audienceY
Secondly, another crucial point that every translator should grasp is that
whatever he does in his translation matters primarily not because it agrees
with or violates some principle or theory of translation, but because of the
causal interdependence of cognitive environment, stimulus and interpreta-
tion. Furthermore, assumptions about what translation is or should be can
influence the success of a given translation - not in virtue of any special
translation-theoretic status, but simply as contextual assumptions used in
the interpretation process.
Thirdly, since the phenomena of translation can be accounted for by this
general theory of ostensive-inferential communication, there is no need to
develop a separate theory of translation. The success or failure of transla-
tions, like that of other instances of ostensive-inferential communication,
depends causally on consistency with the principle of relevance. 22 Indeed,

21The existence of such a direct relationship is also denied by ReiB and Vermeer 1984.
22Note that this accounts also for the claim made e.g. by Harris and Sherwood 1978 that
translation is an 'iunate skill'. Cf. also the suggestion by Massaro with regard to the
mental processing taking place in simultaneous translation: "Our analysis of the lan-
guage interpretation and communication situation would seem to imply that no unique
or novel skills are required, as long as the interpreter knows the two relevant languages
as well as the person on the street knows one" (1978:300). However, this does not
200 Translation and Relevance

if the relevance-theoretic account of communication is right, then it will be

impossible to give an adequate account of translation without reference to
the principle of relevance.
It may be worth pointing out that on the psychological level this pro-
posal entails the claim that translation requires no special mental equipment
not used in inferential communication generally. Utterances are translated
from one language to another via the comparison of their interpretations,
and the comparison of interpretations is a process inherently involved in all
inferential communication. The fact that the interpretations compared are
derived from utterances in different natural languages is incidental to the
overall communication process; it obviously presupposes competence in
more than one natural language, but this faculty for multilingualism needs
to be accounted for in terms of linguistic competence in general; it is not
peculiar to translation.
Fourthly, the choice of a particular approach to translation, such as di-
rect or indirect translation, is not theoretically significant; both kinds of
translations are processed by the same principles of communication, the
distinction between the two approaches is purely theory-internal. This im-
plies that there is no theoretical necessity for a translator to follow either of
the two approaches consistently. What he has to remember, though, is that
unexpected deviations from a given approach can lead to mismatches in the
cognitive environment and are therefore likely to put communicative suc-
cess at risk.
Fifthly and more generally, the account of translation given here is nei-
ther descriptive nor prescriptive in its thrust, but explanatory. It's aim is not
to give a systematic account of what people do in translation nor to tell
them what they ought to do. It rather tries to understand what causal inter-
dependencies are at work in translation, and hence to bring out what its
conditions for success are. For example, this relevance-theoretic account
may be seen to throw new light on the century-old debate of literal versus
free translation: it is possible that the notions of direct and indirect transla-
tion spell out the intuitions underlying these two traditional notions. However,
as I see it, the main contribution of this study is not to arbitrate in the debate

necessarily contradict the claim that translation can require expert skills beyond the
mastery of the source and receptor language (cf. e.g. Newmark 1988, Snell-Hornby
1988). For example, in literary translation a good background in literary studies will be
necessary both for interpreting the original authentically, and for composing a transla-
tion that will meet the expectations of a literary audience.
A Unified Account of Translation 201

by coming down on one side or the other, but to make explicit the commu-
nication-theoretic roots of this controversy and to help people understand
the natural strengths and limitations of each approach.
The account given here is also explanatory in that it explains why trans-
lation occurs at all; it occurs where consistency with the principle of relevance
requires that the utterance representing another utterance be expressed in a
language different from that of the original.
Finally, due to its exploratory nature this study has tried to deal with a
fairly wide range of phenomena - which means that many of them have
only been touched on. I trust that this will stimulate others to research in
greater depth the numerous facets of this fascinating faculty of our nature-
the ability to open our world of thought to one another, even when we do
not speak the same language.
 Postscript
A statement may be pseudoscientific even if it is
eminently 'plausible' and everybody believes in
it, and it may be scientifically valuable even if it
is unbelievable and nobody believes in it. A
theory may even be of supreme scientific value
even if no one understands it, let alone believes it.
Imre Lakatos l

A Decade Later

About one decade has passed since I carried out the re-
search reported in this book - I submitted the results in
June 1989. There have been various kinds of reactions:
appreciation, criticism, comments, suggestions, and also
applications of this framework to various areas of trans-
lation. (See 'Reviews and literature with comments on
Gutt 1991', p. 240 below). I had hoped to be able to re-
spond to them all in detail but have not managed to do so
up to now. Perhaps I shall get around to it one day. For
the purpose of this chapter, I shall try to focus on some
core issues that I feel to be of special importance.
Approaches to Some of the comments received suggest that it may
translation versus
accounts of be helpful to draw a distinction between approaches to
translation translating and (theoretical) accounts of the phenomenon
of translation. In this terminology, approaches to trans-
lating favour (or reject) particular modes of translating,
such as 'literal translation', 'free translation' (whatever
one might mean by those terms), 'formal correspondence
translation', 'dynamic equivalence', 'idiomatic transla-
tion', 'functional equivalence translation' or the like.
By contrast, accounts of translation are interested
in claritying what this phenomenon is all about, what its
nature and characteristics are. In this sense, Catford's

1 lmre Lakatos (1978) The methodology ofscientific research programmes, Cambridge:

Cambridge University Press, p. 1.

Postscript 203

linguistic model, Descriptive Translation Studies, action-

theoretic investigations would all be examples of accounts
of translation. 2
Against this background, the relevance-theoretic study
of translation presented in this book intends to be a (theo-
retical) account of translation; its focus is to explain how
the phenomenon of translation works. It does not consti-
tute or advocate a particular way of translating.
This distinction has sometimes been missed, and some
readers have - wrongly - assumed that this book is ad-
vocating a particular mode or ideology of translation. In
order to counteract this misunderstanding, in this chap-
ter I shall use the terms 'approach (to translation), and
'account (of translation)' with the distinct meanings just
explained.

The Research Programme Shift

(Paradigm Shift)

A sharp divide Reactions to this book (as far as they have come to my
attention), appear to divide fairly sharply into two groups:
reviewers either appreciate it as a major step forward in
the endeavour of understanding translation, or they are
quite disappointed, some even worried by it. There ap-
pears to be little middle ground between these two
reactions.
A fundamental Why should this be so? Closer examination suggests
shift that the most decisive factor is what people make of the
first chapter of the book: whether they have appreciated
or missed the fundamental importance of the call for an-
other shift in research programme. 3 Part of the problem

2 I realize that in the past this distinction has often been blurred, in that certain accounts
of translation have tended to go together with certain approaches, or certain approaches
have tended to rely on certain accounts. For example, 'dynamic equivalence' advocates
a certain approach to translating, but also relies on a code-based account oftranslation.
However, in principle these two dimensions can vary independently.
3 Earlier programme shifts in this century may be seen in the transition from philologi-

cal to communication-oriented translation in the 1960s and 1970s, then Toury's proposal
to shift from prescriptive to descriptive studies of translation in the 1970s and 1980s
(see Hermans 1995).
204 Translation and Relevance

is that the paradigm shift suggested here goes far deeper

than matters of translation and affects some fundamental
but often subconscious beliefs about language and com-
munication. Without a revision of those beliefs it is
difficult to understand and appreciate the account of trans-
lation presented here. Since I obviously did not succeed
in getting this point across to everybody, let me give it
another try, from a different angle.

The input-output research programme

Translation The currently most familiar and favoured paradigm for

research by investigating phenomena of translation is what one may
comparing out- call the 'input-output' account of translation. 4 Its most
put with input ...
central axiom appears to be that translation is best stud-
ied by systematic comparisons of the observable input
and output of the translation process: 'input' being the
original text, 'output' the translated or target text. Hence,
the object domain in this framework consists essentially
of texts, mostly source-text / target-text pairs. The prop-
erties of each text pair are carefully studied, described
and compared.
From this body of descriptive comparisons, which con-
... in relation to
stitutes the basic platform from which all further scientific
the target culture
interpretation proceeds, generalizations of various sorts
can then be made: usually groupings or classifications
on the basis of shared characteristics, but also attempts
at explanation, which build on observed correlations be-
tween the proposed groupings and, for example, socio-
and historico-cultural factors, especially those found in
the realm of literary traditions, action-theoretical con-
cepts, like 'skopos', or other external factors that seem
to have a bearing on the input-output relation. Further-
more, the input-output account also attempts to make

4 I am introducing this new term because it should be free from the connotations and
associations of familiar terms; it is meant to be fairly wide, and one can try to determine
on the basis of the characteristics proposed whether any given established view of trans-
lation would be affected by its arguments. I expect this to be a question of matter of
degree, with grey areas, rather than a binary 'inside' or 'outside' distinction.
Postscript 205

generalizations about the process and procedures which

lead translators to produce the target text on the basis of
the source text.
Translation as This research programme typically aims at the sys-
a relationship tematic compilation of statements about regularities in
between pairs the relationships between source and target text, in the
a/texts
more sophisticated accounts attempting to relate these
regularities further to relevant external factors, such as
those just mentioned. The concept of translation is thus
based on or is an abstraction from the relationships be-
tween pairs of texts. The research in this paradigm is - to
a large extent - still programmatic, and leading propo-
nents predict that the road ahead is a long one, promising
many, many years of exciting, if hard work. 5

The competence-oriented research programme

Why can people The research programme followed in this book has a radi-
translate? cally different starting point. It starts from the observation
that human beings have the remarkable ability of telling
in one language what was first told in another language.
Given this fact, the programme endeavours to find out
what this human ability consists in. Thus, this framework
seeks to understand translation through understanding
the communicative competence that makes it possible,
both for the translator and his/her audience. In line with
this goal, competence-oriented research of translation
(CORT) embeds itselfin scientific investigations of the
competence ofhuman beings to communicate with each
other.
Focus on the The object domain of competence-oriented transla-
mind tion research is, therefore, very different from that of
input-output-oriented research; since communicative
competence is most likely located in the human mind,
that is where the focus of the research lies. Source and
translated text obviously play important roles, too, but
primarily as data that allow conclusions about the mental

5 See Toury (1995) for an attempt to formulate two provisional 'laws' oftranslation.
206 Translation and Relevance

faculties concerned. Thus the aim of CORT is to under-

stand and explicate the mental faculties that enable human
beings to translate in the sense of expressing in one lan-
guage what has been expressed in another. The idea is
that once these faculties are understood, it is possible
to understand not only the relation between input and
output, but also, and perhaps more importantly, the
communicative effects they have on the audience. Af-
ter all, the raison d' etre of all translation appears to be to
communicate.
In the competence-oriented paradigm, one can retro-
Explanation and
prediction in spectively explain not only why a certain input led to a
CORT certain output, but also why the output had the com-
municative effects it did, and how those related to the
communicative effects of the input. Prospectively, one
can attempt to predict not only what output a certain in-
put is likely to lead to, but also what the communicative
effects of the output are likely to be, and how they will
compare to those of the input.

Small wonder different evaluations

In the light of these basic differences in framework, the

split in evaluations is hardly surprising: adherents of the
input-output programme obviously look in vain for what
in their view is the 'meat' in theories of translation - that
is, detailed statements about input-output relations fitted
into complex text-typological and functional systems.
Against those expectations the principle of relevance will
not unnaturally appear to be rather meagre and vague.
Furthermore, since they consider relevance theory as a-
largely unnecessary - addition rather than as the back-
bone of the whole endeavour, they feel irritated rather
than enlightened by it.
What people often do not realize is that CORT does
Cultural factors
have their place take full account of all the detail, including the socio-
too and historico-cultural factors that influence what trans-
lators do and audiences expect. The main difference is
Postscript 207

that these factors are not treated through the establish-

ment of classificatory systems or sub-systems of the
theory, but rather as natural elements of the contextual
background that translator and audience bring to the com-
munication act. The notion or notions of 'translation'
favoured by a particular cultural group or sub-group, no-
tions of 'genre' that might determine which notion of
translation is considered appropriate for certain kinds
of texts, conventions or ideas about what makes a text
'literary' and the like - they all enter into the relevance-
theoretic account of translation as contextual assumptions
held by translator and audience.

The same in the end?

While most readers would probably agree that the two

frameworks for translation research are clearly very dif-
ferent from each other, some, however, might wonder
whether the differences are mainly on the surface - that
ultimately both will lead to the same results. Let me give
a few reasons why, to the best of my understanding, this
seems very unlikely.
A Sisyphean First, the input-output framework attempts to make
enterprise statements about an unlimited number of concrete input
entities, an unlimited number of concrete output enti-
ties, and a presumably also unlimited number of external
influential factors, relating them all in ever new combi-
nations. Chances are that the research process will be
very protracted, possibly endless. It may also be men-
tioned here that this paradigm faces the questions raised
in pragmatics about the existence of texts as valid units
of scientific analysis (cf. e.g. Blass 1990, Reboul and
Moeschler 1997).
Second, the framework faces all the problems outlined
What about
chapter 1? in chapter 1 above, including that of multi-disciplinary
disintegration. To my knowledge, none of the reviews I
have seen addresses those problems in any serious man-
ner, let alone offer any realistic solutions for them. Unless
such solutions are found, the prospects for valid results
along these lines are dim.
208 Translation and Relevance

Text-production Third, the input-output framework appears to have

or
very little, if any, concern for the communicative effects
communication?
and effectiveness of translational activities; for the most
part, it is preoccupied with textual aspects of translation,
augmented by relations to cultural factors. This seems,
to say the least, very strange, since the raison d' etre and
purpose of translation appears to be not to produce cer-
tain kinds of textual phenomena, but to communicate
across language boundaries. To be sure, the most easily
observed parts of the translation event are the original
text and the target text, yet these are not ends in them-
selves but instruments designed for the purpose of
communication. What hope can there be to understand
the whole phenomenon if one neglects its central aspect?
(Previous accounts that have included communicative
concerns have been seriously hampered by the inadequa-
cies of the code-model on which they - expressly or
tacitly - built, as I have tried to show in chapters 3-6.)

The Unique Mandate of Translation

Most importantly, perhaps, the relevance-theoretic ac-

count has been able to bring out with new clarity the truly
unique mandate of translation which distinguishes it from
other modes of interlingual communication. This should
help to clear up a major area of confusion that has bur-
dened translation theory and practice for centuries. In fact,
it is only the development of relevance theory that has
made it possible to clarify the uniqueness of translation
to such an unprecedented degree. Since this appears to
be a point that has been missed by many I shall try to
briefly highlight it here.

Equivalence ofmessage or function is not enough

The criticism that translation is not simply conveying the

same message as the original or even carrying out an
equivalent function has been made before, usually in the
sense that the purpose of the translated text could require
Postscript 209

drastic changes compared to the original. The point to be

made in this section, however, has nothing to do with the
extent of the changes a target situation may call for. Its
thrust is quite different and may be illustrated with the
following example.
Suppose in Sweden a road accident (of a third party)
Saying the same
thing in two has been witnessed by both a local person and a foreign
languages isn't tourist. The police come, and take down the eyewitness
yet translation accounts of both persons: the Swede's account is taken
down in Swedish, that of the foreign tourist in English.
For the sake of argument, let's assume that the two ac-
counts show good agreement: they basically' say the same
thing' - the accounts are, in effect, pretty much identical
both in 'message' and 'function'.6 However, despite this
high degree of correspondence the two accounts would
not normally be considered translations of each other.
On the other hand, there are instances where the simi-
Not saying the
same thing - and 1arity between two texts in different languages might not
yet translation be as close, yet many people would be quite happy to
call one a 'translation' of the other. Many of the 'freer'
or more 'dynamic' translations of literature would be
cases in point.
These observations suggest that the question of
Correspondence whether one text is a translation of another does not
relations cannot depend on the existence of correspondence relation-
provide the
answer
ships between the two. Correspondence relations appear
to be neither sufficient nor necessary conditions for de-
fining translation. If this is so, then what, if anything,
does define translation and distinguish it from other
modes of interlingual communication?

Translation as interlingual quotation

Intuitively, the point seems to be that in translation,

one does not simply express the same ideas that some-
one expressed, but that one presents those ideas as an

6 I know such close agreement of eyewitness accounts doesn't happen all that often -

but for argument's sake let's assume that rare case.

210 Translation and Relevance

Translation - expression ofwhat that person expressed. Moditying the

telling what
road-accident example, if someone expressed in Swedish
someone ex-
pressed in - for the sake of monolingual bystanders, perhaps - what
another language the foreign tourist had said in English, then, in most peo-
pie's mind, this would quality as an incident of translation.
What is there behind this apparent difference between
'saying the same thing as someone else' and 'saying what
someone else said'?
To the best of my knowledge, only relevance theory
as developed by Sperber and Wilson offers an explicit
answer to this question, with a claim to some cognitive
reality. It does so with the concept of the 'interpretive
use' oflanguage, as distinct from 'descriptive use'.
It would be redundant to repeat an explanation of this
distinction here; the reader can tum to the last section of
chapter 2 for a fairly brief account, or to Sperber and
Wilson (1986, chapter 4, section 7) for a more com-
prehensive explanation. At the same time, since this
distinction seems to have been one of the most easily
missed or misunderstood concepts of relevance theory,
some further comments seem to be called for.
The issue of in- The fundamental characteristic of the interpretive use
tended relevance of language is not just the fact that two utterances inter-
pretively resemble one another, but that one of them is
intended to be relevant in virtue of its resemblance with
the other utterance. In general terms, in 'reported speech'
interpretively used utterances "achieve relevance by in-
forming the hearer of the fact that so-and-so has said
something or thinks something" (Sperber and Wilson
1986:238).
The relevance of Translation, as a case of interlingual reported speech
what someone or quotation, therefore, achieves relevance by informing
said or wrote in the target audience of what the original author said or
another language
wrote in the source text. This brings out a crucial point:
the distinction between translation and non-translation
hinges first and foremost on the way the target text is
intended to achieve relevance. This statement has a
number of very important consequences that have not
been paid much attention to generally because accounts
Postscript 211

of communication other than relevance theory have

lacked adequate conceptual tools for dealing with them.

Some Important Consequences of a

Quotation-Based Concept of
Translation

Translation is an act of communication rather

than a text genre

If it is correct that the distinction between translation

and non-translation is a matter of the communicator's
intention, that is, of how it is intended to achieve rel-
evance, then attempts of looking for a definition of
translation in terms of intrinsic structural properties of
the text, as many accounts of translation aim to do, must
be of questionable validity.
The wrong In a way, the question of whether a text is a transla-
question to ask tion or not seems to be the wrong kind of question to
ask; it is not texts as such, as structured compositions in
a particular language, that are translations or otherwise,
it is the use of such texts with a particular intention that
constitutes translation. In other words, translation - in
the primary sense - is an act of communication, more
precisely, an act of interpretive use across language
boundaries. It is only in a secondary, derived sense that
the term 'translation' applies to texts (or classes oftexts),
that is, to texts that have been produced in such acts of
communication.

An aside on intentions

Communicators At this point it may be helpful to address an objection

intend to sometimes made to the focus of relevance theory on in-
communicate
tentions. The suspicion of some would go so far as to
something
doubt their very existence. As an author, all I can say to
that is that the struggle and effort authors invest in trying
to express their thoughts seem very difficult to explain
212 Translation and Relevance

unless they have an intention to communicate certain

thoughts or ideas. They are not contented with putting
black marks on paper that will somehow entertain the
audience or show conformity to some structural scheme
of the culture. The audience, too, seems to start on the
assumption that communicators are not just making noises
or producing structured objects but that they intend to
communicate something by them. Intentions may seem
philosophically debatable, but in real life they appear to
have a pretty robust existence.
Are intentions Less radical critics point out that it is rather naIve to
recoverable? believe in the recoverability of other people's intentions-
assuming they exist. The answer here is quite simply that
relevance theory does not hold the view that intentions
are recoverable in any direct sense. Sperber and Wilson
are very clear on this point. They certainly do not hold
that communication opens up the communicator's inten-
tions to inspection by others. Intentions cannot usually
even be recovered by means of' demonstrable inference':

The addressee can neither decode nor deduce the

communicator's communicative intention. The
best they can do is construct an assumption on
the basis of the evidence provided by the com-
municator's ostensive behaviour. For such an
assumption, there may be confirmation but no
proof. (Sperber and Wilson 1986:65)

A hypothesis, not Thus Sperber and Wilson clearly dissociate themselves

a reconstruction from any naIve view of the recoverability of intentions.
Their claim is far more cautious and, I believe, also quite
realistic: it certainly is an everyday reality that we make
inferences about people's intentions on the basis of their
behaviour. 7 Thus, what the audience comes up with is
not a 'reproduction', but an 'interpretive assumption

7 To invalidate this claim, one would have to show that it is impossible in principle -

not for circumstantial reasons - to ever construct such an assumption with any prospect
of validity.
Postscript 213

aboutthe speaker's [or writer's EAG] informative inten-

tion' (Sperber and Wilson 1986:230tV
Even in quite successful cases of communication, the
audience's hypothesis will hardly agree one hundred per
cent with the communicator's thoughts, yet this limita-
tion does not obviate the point ofhuman communication:

It seems to us neither paradoxical nor counter-

intuitive that there are thoughts we cannot exactly
share, and that communication can be successful
without resulting in an exact duplication of
thoughts in communicator and audience. We see
communication as a matter of enlarging mutu-
al cognitive environments, not of duplicating
thoughts. (Sperber and Wilson 1986:193)9

Translation is communication between

translator and target audience

Another interesting fallout of the relevance-theoretic

For the target
audience, the account is that de facto translation is an act of commu-
translator is the nication between translator and target audience only.lo
communicator This is true even in simultaneous interpretation, where
the original author may be physically together with the

8 This is also where the Quinean concerns about translation seem to be moot: not much
appears to depend on whether A's concept of 'rabbit' is actually identical to B's con-
cept of 'rabbit' and whether that can ever be proved (Quine 1959). The point rather is
that B can make intelligent guesses about what A might be thinking about - even if the
concepts of the words they use are never identical. While proofs of identity are not
possible, these guesses are subject to confirmation or otherwise, as Sperber and Wilson
suggested. This probably explains why - despite Quine'S problems - we have not each
come to live in totally different worlds, where for one person rabbits are animals and for
another they are a specific kind of stellar matter in a faraway galaxy, and nobody would
ever notice the difference.
9 Another objection sometimes made against dealing with intentions is that communica-

tors caunot always be aware, let alone sure of their own intentions. This objection would
only be valid if one assumed that only thoughts one can be either aware or sure of can
playa role in communication - an assumption that seems rather naIve and arbitrary.
Modern psychology appears to be quite strong in its recognition that consciousness is
not a prerequisite for the reality of mental processes.
10 This may help explain the intuition behind the strong insistence in Descriptive Trans-

lation Studies that translations somehow 'belong' to the target culture than anywhere
else. (cf. esp. Toury 1995:29)
214 Translation and Relevance

target audience. As far as the target audience is concerned,

what they are actually confronted with is the target text
produced by the translator; that text comes with a prom-
ise or presumption of interpretively resembling the original
text, but the original text does not reach the target audi-
ence (unless, of course, they are bilingual- in which case
they won't need the translation in the first place).
The following diagram which locates translation in
the relevance-theoretic framework, may help to show this.

the translated text

The propositional form of
an utterance

is an intepretation of

a thought of the
translator
a thought of
the speaker

which is

an interpretation of a description of

an attributed a desirable an actual a desirable

thought thought state of state of affairs
affairs

a thought attributed to
someone who expressed it in
another language, i.e. the
original author

Figure 1
The place of translation in relevance theory
(based on Sperber and Wilson 1986:232)
Postscript 215

The bolded lines (going down the centre and then to

the left) indicate that in translation the propositional form
of the utterance (~ 'translated text') is an interpretation
of a thought of the speaker ( ~ translator) which is an
interpretation ofa thought attributed to someone who ex-
pressed it in another language (~ the original author)Y
Thus the communicator whose utterance the target audi-
ence is actually dealing with is that of the translator.
More than
So this recognition ofthe translator as the actual com-
wishful thinking municator to the target audience is not a claim born
perhaps from a desire to raise the status of translators.
Rather, it is a plain fact of the realities of communica-
tion. It holds true even if the target audience misses it or
chooses to ignore it: they inevitably listen to or read the
translator's interpretation of the original - whether they
realize it or not (cf. pp. 221-2 below).

Translation cannot be covert

The other side of the coin is that, in a very real sense,

translation cannot be covert: since one of its defining
characteristics is that it comes with the intention of in-
forming the target audience that the original author has
said or written such-and-such, it cannot achieve its ob-
jective without that intention being recognized by the
audience. 12
This exclusion of 'covert translation' from the realm
of translation proper has met with a certain amount of
scepticism and opposition. Much of the criticism, how-
ever, seems to be due to misconceptions of what this
exclusion means. Let me try to clarify some ofthem here.

11 Note that this account also takes care of the phenomenon of 'pseudo-translations'

observed by Toury (e.g. 1995): they are cases where the (explicit or implicit) presump-
tion of the target language communicator to represent thoughts expressed in another
language is untrue.
12 This goes in the same direction but further than Toury's 'source-text postulate'; the

"assumption that [a source-text] must have existed" alone is not enough (Toury 1995:34);
this assumption must have been part of the communicator's informative intention, ei-
ther explicitly or implicitly.
216 Translation and Relevance

The production of 'covert translations' involves

translation but ...
Not entirely As has been pointed out by some reviewers, no matter
arbitrary what relevance theory might say, the fact of the matter is
that people do and no doubt will continue to talk about
'covert translations '. The reason for this is not entirely
arbitrary either: an act of translation does occur in such
processes of communication, even though this transla-
tion act does not involve the target audience directly. The
following diagram may help to clarify the complexities
of what actually takes place.

Step 1

translator

German English translation

instruction
manual

Step 2

English instruction
manual

Figure 2
The process of 'covert translation'
Postscript 217

This figure shows a 2-step process: at the preparatory

stage, step 1, the instruction manual (here how to oper-
ate a computer) is translated from German into English.
Thus, at this stage, there is translation proper. However,
at the main communication stage, step 2, the user reads
the instruction manual to discover how the machine op-
erates. S/he has no interest in whether there was a version
of this manual in any other language nor what such a
manual might have said. Nor does the communicator,
which is the company selling the machine, have any
intention of informing the user of what some foreign
engineer wrote; it rather intends to give the customer ad-
equate instructions about how to operate the machine.
Thus, although a process of translation did take place
in the production of the English manual, as far as both
company and user are concerned the relevance of the
manual does not lie in its interpretive resemblance with
the original, but in its giving an accurate description of
how to operate the machineY Nevertheless, it is easy to
see that an input-output account of translation with its
focus on texts rather than the communication act would
call the target manual a translation.

Does it matter?
If the production of 'covert translations' does involve a
process of translation proper - then why not simply in-
clude it under that concept? Is this not just a matter of
terminological pedantry? No, it is not. There is an im-
portant distinction here which has a number of significant
consequences for the translator, his/her client and the
translation theorist. Three of these consequences are the
following.

13 The fact that neither the original nor the translator are normally mentioned in in-

stances of 'covert translation' underlines their irrelevance to the communication act.

Note, however, that this omission of an explicit declaration as translation in itself does
not make a translation 'covert'! (See p. 202.)
218 Translation and Relevance

Effectiveness, not 1. If the translator does take on a job of 'covert trans-

resemblance lation', slhe will do well to realize that s/he is now in a
ball game quite different from translation proper: in
the last resort, what is called for is not interpretive re-
semblance, but descriptive accuracy and adequacy and,
especially in the field of advertising, effectiveness.
What ultimately counts for hislher client is the effec-
tiveness of the target text: does it tell the user clearly
and effectively all slhe needs to know about the prod-
uct? In the view of cultural differences, 'faithfulness'
or 'equivalence' may not be reliable guides to success,
and slhe should not, therefore, look to translation
theory for help or guidance. This realization has been
the - unfortunate - motivation for some translation
theorists to introduce and give precedence to notions
like 'purpose' ('skopos') and adequacy (see below).

2. However, what has not normally been recognized

'Covert
translations' as is that there is nothing in 'covert translation' that makes
shortcuts to a process of translation intrinsically necessary; typi-
target language cally, any 'covert translation' could be equally well
originals produced without translation, and often with better re-
sults. (By contrast, if! want to find out what a Chinese
poet wrote, then only translation will do.)
Suppose a company needs to produce an English in-
For best results ... struction manual for a complicated German agricultural
machine to be used in a Third World country. Ifthe
company commissioned an engineer in that country -
with no knowledge of German, but trained on the ma-
chine and with good communication skills - to write a
manual, this person would have a number of advan-
tages over a translator: slhe would be free from
language and culture interference from the German
manual and could fully concentrate on the needs of
the future users overseas; with hislher genuine under-
standing of the machine, the engineer could take into
account their special circumstances (degree oftechni-
cal competence, relevant environmental factors etc.)
and thus produce a manual optimally suited to the tar-
get situation.
Postscript 219

A second- best By comparison, a translator who lacked this

engineering competence would not be in a position
to produce an equally effective manual, regardless of
how good hislher language competence might be.
Thus, when faced with a request for 'covert transla-
tion', the translator could seriously consider whether
it is both in his/her own as well as his/her client's
best interest to tackle this job by 'translation'.
Chances are that, due to traditional practice, more
often than not the client is not even aware of the
disadvantages of the translational approach. Mini-
mally, the translator could explain to the client the
existence of non-translational alteruatives, with their
pros and cons, and leave the decision to him.14
Resemblance as Again to avoid misunderstanding: if a client does
a 'crutch' commission a translator to produce a 'covert transla-
tion', such as in the example of the agricultural machine
just given, then for the translator without engineering
competence the only hope for achieving descriptive
accuracy in the English version of the manual (i.e. give
the user correct information about how to operate the
machine) lies in close meaning resemblance between
target text and original. However, the reason for this
reliance on meaning resemblance is primarily that the
translator has no independent understanding of how
the machine operates, and therefore has no other way
of judging whether the information s/he gives is cor-
rect or not. If s/he had that knowledge, meaning
resemblance would be of no importance.
Convenient Thus, in the last resort, 'covert translations' are
shortcuts mainly convenient shortcuts to producing target lan-
guage originals, taken out of economic interests: it is
usually easier and cheaper to find bilinguals who can
translate the instructions for a product into another lan-
guage rather than to find an engineer in the target
language who understands the product well enough and
has the skill of composing those instructions from
scratch.

14 I realize that this might mean that the translator is talking himself out of a job; but
then, s/he may not feel too attracted by a task that others could probably do much better.
220 Translation and Relevance

Eliminating 3. Lastly, regarding the scientific study of translation,

confusion in the exclusion of 'covert translation' will eliminate the
translation
considerable confusion which attempts to accommo-
research
date this different mode of communication within
translation theory have caused. One case in point is
the admission of a - totally undefined - relation of
'functional change' between source text and original
by Honig and KuBmaul, rendering the relationship be-
tween translation and original virtually void of any
meaning (1984; see discussion on pp. 55-6 above).
Another instance is the attempt to make the pur-
pose, or 'skopos', the primary concern oftranslation. As
Chesterman observes, "this approach has been less
successful in pinpointing the effects of the skopos on
the translator's actual textual choices, and in propos-
ing testable generalisations about the relationship
between skopos and translation-text" (Chesterman
forthcoming: 8). 15

'Covert transla- Whether or not translators should engage in 'covert

tion' marches on translation' is a different question altogether. As just
shown, translators can have a part in the process, and no
doubt they will continue to be asked to translate tourist
brochures, advertisements, instruction manuals etc. Re-
alistically, not least for economic reasons, in many cases
they will continue to oblige. Nonetheless, translators -
and, perhaps more importantly, their clients - should be
fully aware of the different nature of their task and that
their client might be served better by a target language
original than a translation.

15 The main problem is that the 'skopos' theory oftranslation is 'hollow at the core'.

Condition (3) of the general translation theory attempts to zero in on what differentiates
translation from other forms of giving information about a text in the source language.
It reads: '(3) Das Informationsangebot einer Translation wird als abbildender Transfer
eines Ausgangsangebots dargestellt' (ReiB and Vermeer 1984: 105; italics my own. While
the authors give an essentially semiotic definition of 'Transfer' (ReiB and Vermeer
1984:88), they give no explanation or definition of what they mean by 'abbildend',
though that is the central concept that would differentiate translation from other forms
of interlingual information transfer.
Postscript 221

Manuals, tourist brochures, advertisements can be ob-

jects of translation proper
Before leaving the topic of 'covert translation' altogether,
it may be worth pointing out that the question of whether
one is dealing with an instance of 'covert translation' or
a 'translation proper' cannot be answered on the basis of
text type or subject matter.
What did the Suppose a machine purchased from a foreign manu-
original manual facturer breaks down; the purchaser, who followed the
say? original instruction manual, claims the malfunctioning is
due to incorrect instructions in the manual. If the matter
goes to court, the court not being conversant with the
original language may well ask for an exact translation
of the original manual in order to be able to evaluate its
adequacy or otherwise. Note that in this situation, a
'covert translation' which aimed at the best possible de-
scription ofthe machinery regardless of what the original
said, would not serve the court at all. (Cf. also the adver-
tising examples in chapter 3, p. 57)
Thus there can be instances where even in the case of
instruction manuals, advertisements, or tourist brochures
translation proper is called for. This shows again the limi-
tations of basing accounts oftranslation on text typology.
As pointed out above, the core of the matter lies in the
way in which a text is intended to achieve relevance.

Translations can be interpreted 'incognito' but ...

Another point to clarify is that translated texts, including

translations proper, can and often are read without any
realization that they are intended to be translations rather
than originals. This observation does not conflict with
the statement just made that translation proper cannot be
covert. What it does mean is that, like any other commu-
nicator, the translator has no guarantee for getting the
target audience to infer hislher informative intention with
one hundred per cent success. Thus an audience may fail to
realize that the translator intended the text to be relevant
222 Translation and Relevance

as an interpretation of what someone else, the original

writer, wrote. This need not lessen the audience's enjoy-
ment ofthe story or text, but it may cause comprehension
problems.
Being led astray Without an awareness of the fact of translation, the
audience would naturally assume that the text was de-
signed for them and would use the information it could
assume to be part of the cognitive environment mutually
shared with the communicator, as consistent with the
principle of relevance. It would fail to realize that the
text was written against a different context, which they
would first have to familiarize themselves with. Depend-
ing on the nature and degree of differences between the
original context and their own, this could lead to consid-
erable misinterpretation. 16
Not necessarily However, it does not therefore follow that translation
an explicit label proper must be marked as such in an explicit way. The
relevance-theoretic definition of translation implies that
the translator communicates hislher interpretive intention,
but it does not necessarily require that it be communi-
cated explicitly; depending on the situation, it may be
communicated just as well by implicature. Thus if I ask
someone to tell me in English what a speaker at a confer-
ence says in Arabic and the person complies with my
request, the reply will by implication be a translation.
Hence while translation proper cannot be 'covert' - in
the defined sense - it can be implicit.

Translation proper does not depend on the exist-

ence of a concept of 'translation' in the target
culture

Lastly, it may be helpful to point out that in the relevance-

theoretic framework, the question of whether one is
dealing with an instance of translation proper or not does

16 Note here, for example, the observation by Weizman and Blum-Kulka that the recog-

nition of a translated text as a translation '''protects' the reader, as it were, from

misinterpreting the writer's intentions" (1987:72).
Postscript 223

not depend on whether the communicator - or anybody

else - applies the term 'translation' to it - or even whether
the target culture has any concept of 'translation'. That
question is settled entirely by whether the act of commu-
nication concerned fulfils the definitional criteria of
translation: that is, whether the text or utterance in ques-
tion is intended to be relevant as an interpretation of what
someone expressed in a different language. This is a fun-
damental difference to some other accounts of translation,
notably that of Descriptive Translation Studies which
builds on the assumption that the culture has some con-
cept of translation, even ifit has no word for it. As far as
relevance theory is concerned, translation occurs regard-
less of whether the target culture has conceptualized it in
any way or not. 17

The Perspective of Translation Practice

The primary concern of this book has been theoretical;

as stated in the preface, the research underlying it was
originally directed towards formulating a general theory
of translation. In the light of this orientation, it may be in
order to examine what, if anything, the account presented
here has to offer to the practising translator. 18

17 I am referring here - as well as on p. 7, chapter 1, not to the English word 'transla-

tion'; as Toury points out, that would indeed be a strange position to take (Toury
1995:33); I rather refer to the concept(s) for which this word stands in the English
culture and literary tradition. Similarly, when I speak about the Amharic 'ti'rgum', the
concept is meant, rather than the word, the point being that there is no a priori reason to
assume that the concepts in the English and Amharic culture respectively have much in
common - unless the two terms are themselves appropriate translations of each other,
as Hermans (1995:221) correctly observes. However, if one pursues this question fur-
ther, one runs into problems: are they supposed to be 'translations' of each other or
'ti'rgums' - which could make a difference! It seems that, taken to its logical conclusion,
the extension of Descriptive Translation Studies to any other culture than English in-
volves an infinite looping or regress.
18 Since the publication of this book, the author himself has written some practice-

oriented applications of the relevance-theoretic account: Gutt (1992) presents some crucial
insights for Bible translators; Gutt (1996a) addresses the issue of assessing translation
quality (see also Gutt 1996b). To address the task of translation training, a course
book - focusing on Bible translation - is being prepared (Sim and Gutt in preparation).
224 Translation and Relevance

No practical Some critics have claimed that there is little of inter-

help?
est to practitioners oftranslation. Thus Malmkj <er advises
translators that "if they want direct help with their every-
day concerns, they should not expect to find it here"
(Malmkj<er 1992:306; similarly Tirkkonen-Condit
1992:243).
Where are the The main complaint is that the relevance-theoretic ac-
rules? count remains too general. Thus noting that the account
does not provide the translator with specific rules and
principles that tell him/her what to do with specific prob-
lems, Malmkj<er alleges that an understanding of
relevance theory will not by itself enable translators to
predict the relevance of any particular turn of phrase to
those individuals which they might see as the projected
audience for their translations. Ergo, translators cannot
expect to be able to apply relevance theory directly when
translating (Malmkj<er 1992).
How relevance The validity of this criticism has to be judged first of
theory primes the all against the fact that translators have been able to ap-
translator for ply the framework to their work (cf. e.g. Muschard 1995,
his/her task
Navarro 1995, Weber 1998) as well as against the find-
ings of other reviewers, who point out the helpfulness of
the relevance-theoretic account for the translator (cf. e.g.
Dolitsky 1992, Frawley 1992, Hymes 1992, Salkie 1991
and Stark 1993).
Secondly, it will be noted that this criticism reflects
the expectations of the input-output programme of trans-
lation research discussed above: it looks for detailed
statements about relationships between source text
and target text renderings, and failing to find them it con-
cludes that the relevance-theoretic framework has nothing
to say to them. Such critics miss the fact that a good grasp
of the relevance-theoretic framework not only deepens
the translator's understanding of the problems s/he en-
counters, but is indeed a prerequisite for the proper
application of any rules and principles of translation s/he
might come across.
Postscript 225

Toward a deeper understanding of 'meaning'

A better grasp of To start with, relevance theory will give the translator a
the 'meaning' of better idea of the complexities of meaning of the origi-
the original nal, making him/her aware of

the difference between the (linguistically) expressed

meaning and the intended interpretation;
explicatures, implicatures and varying degrees of
strength of communication;
the crucial dependence of that interpretation on the
availability of the right contextual information;
the fact that hislher own interpretation of the origi-
nal will only be adequate to the degree that it is based
on a thorough understanding of the original context;

Regarding the target text, too, s/he will recognize that

Special concern
for contextual the interpretation of the target audience will strongly de-
factors of the pend on the particular contextual information they bring
target culture to the translated text. Slhe will realize that one crucial
part of that contextual information is what the audience
expects from the translated text - that is, what kind and
degree of resemblance they expect between the interpre-
tations of translation and original. Any label with which
the translated text may be presented - e.g. as a 'transla-
tion', as a 'paraphrase', as an 'abridgement' etc. - and
the meaning it may carry for that particular audience can
be very influential here. The particular interests of the
target audience, whether, for example, they have 'liter-
ary interests' or simply want information, are another
important part of the contextual background. This is, in
fact, the area where the socio-cultural concerns on which
Descriptive Translation Studies focus are taken care of,
such as literary systems and traditions; they enter the pic-
ture as contextual factors and as such relevance theory
naturally draws attention to them and to their importance
for the translation.
Well-prepared Thus a translator who - in addition to a good knowl-
edge of the two languages involved - understands the
226 Translation and Relevance

relevance-theoretic framework is well-equipped to re-

search the meaning expressed in the original text, the
contextual background in which it was intended to be
interpreted, and the interpretation resulting from the com-
bination of the two; likewise, s/he will be well-prepared
to research the target audience's background, including
their particular interest in the translated text.
Equipped with that knowledge, and realizing that the
audience will be guided in their interpretation efforts by
the search for optimal relevance, the translator will -
pace Malrnkj<er - be in a good position "to predict the
relevance of any particular tum of phrase to those indi-
viduals" (Malrnkj<er 1992:308).

Applying the rules

Rules and Returning to the matter of specific translation rules and

guidelines are guidelines, I would like to repeat here that they can be of
useful and
valuable
great help to the translator. 19 After all, those rules and
guidelines sometimes express generalizations about quite
complex phenomena and reflect a considerable amount
of experience, collected by competent translators over
longer periods of time. In many cases these generaliza-
tions reflect what particular features or class of features
are likely to be of special relevance to a particular audi-
ence. Such generalizations are very useful indeed because
not every audience is completely different in all aspects
from any other audience and so it is entirely reasonable
for the translator to use these observations when facing
choices and decisions at the translation desk. It would be
a waste of time if the translator ignored this treasure of
experience and started from scratch, re-inventing the
wheel for every task s/he faces.

19 Despite the claims of some critics to the contrary, in this book I have clearly ex-

pressed my appreciation of literature on translation that provides specific translation

rules and guidelines, pointing out that "it is extremely useful in making the translator
sensitive to the importance of the assumptions present in the cognitive environment in
which he produces his translation" (p. 124).
Postscript 227

Knowing rules However, having said this, it must be emphasized that

cannot replace
a knowledge of these rules and guidelines alone is not
understanding
communication sufficient and cannot replace the need for an understand-
ing of the nature of communication. What the translator
must fully understand in order to do competent work is
that these rules and guidelines have no validity in them-
selves, but only insofar as they are adequate applications
of the principle of relevance to particular texts, particu-
lar groups of people, particular sets of circumstances etc.
Does the rule This is important in two distinct respects. First, the
apply? value of the rule or guideline itself depends on how clear
an indication it gives ofthe contextual situation for which
it was developed. Often times rules and guidelines lack
this contextual information, thereby leading to apparent
contradictions and requiring numerous exceptions (cf.
Savory's provocative collection of translation guidelines,
cited in chapter 7, p. 127 above).
The second respect in which a relevance-theoretic un-
derstanding is important concerns the application of any
such rule. Supposing that the rule or guideline in ques-
tion has in fact been stated explicitly enough, before
applying it to a specific instance of translation, the trans-
lator will need to check out how well the contextual
conditions ofhislher target audience agree with those for
which the rule was drawn up in the first place. To the
degree that these circumstances differ from those assumed
by the rule maker, the rule will be inappropriate and is
likely to lead to bad results. Hence the translator will
better be wary about too quick applications of transla-
tion rules and guidelines.
Furthermore, and this is where I believe lies one of
Where there are
no rules and the greatest contributions of relevance theory to the prac-
guidelines ... tical work of translation, with the understanding that the
primary principle guiding and co-ordinating communi-
cation is the search for optimal relevance relative to a
certain contextual background, the translator knows how
to make a decision in all those instances for which no
rules or guidelines exist. Since texts, people and circum-
stances can vary in - for all practical purposes - infinite
228 Translation and Relevance

ways, it is unlikely that translation rules and guidelines

will ever be able to cover all possible translation prob-
lems. Here the relevance-theoretic account frees the
translator from dependence on existing translation rules.
It enables him/her to be confident about the context-
specific solutions found to the degree that s/he can be
confident ofhislher understanding of the whole commu-
nication situation.

Promises and Perils of Translation

The simplicity of the definition of translation as inter-

lingual quotation should not deceive one about the
complexity and intricacy ofthe phenomenon itself. Trans-
lation offers unique opportunities for cross-cultural
communication, but it also has important limitations
which it would be perilous to ignore. This section is meant
to highlight some ofthe more important promises as well
as risks involved in translation.

Interpretation with a unique claim of authenticity

As authentic as One of the special promises of translation as interlingual

can be ... quotation is that of authenticity. The higher the presumed
degree of resemblance to the original, the higher its
claimed authenticity will also be. The highest possible
claim of authenticity is found in direct translation with
its presumption of complete interpretive resemblance.
Bearing in mind possible limitations imposed by linguis-
tic differences, the fallibility ofthe translator and the need
for familiarity with the original background, the target
audience can expect to gain from direct translation as
authentic an understanding ofthe original as it ever could
across language boundaries. 20

20The concern here is with real authenticity in the sense of giving access to the meaning
ofthe original text. This view differs essentially from Anderson (1998). He proposes
'perceived authenticity' as a practical compromise, which accepts a loss in accuracy for
the gain of increased acceptability of a translation to a conservative audience.
Postscript 229

An authentic experience as well as an authentic

message

Communication One of the constraints of verbal communication is that

is something to of linearity: many of the structural elements of language
be experienced
can only be presented in temporal sequence, though a
certain amount of simultaneity is allowed for by so-called
'supra-segmental' features, such as intonation. This con-
straint limits communication in some ways, but it also
enriches it in others. Thus the communicator can exploit
the time dimension for special effects, creating experi-
ences of suspense, development and progression. If we
could somehow take in the plot of a novel in an instant,
as a four dimensional picture, so to speak, we would be
missing a lot. Mysteries would lose much of their appeal
altogether. Often the wayan author unfolds his or her
thoughts is as rewarding and enjoyable as the thoughts
themselves.
Access not only These experiences are closely linked to the way the
to a message, but author expresses his/her thoughts in the text, and since
to an experience
translation as interlingual quotation aims at preserving
how the author expressed himiherself, translation offers
a sharing - though hardly ever an exact reduplication -
of these experiences as well.

Translation (alone) cannot necessarily

guarantee successful communication

Having dealt with some promises of translation, we

should now tum to some of the limitations which - if
ignored - may endanger the success of translation.
Better warn them The most obvious and most commonly acknowledged
limitations are those imposed by the linguistic differences
between source and target language. These differences
may make it very difficult, if not impossible in places, to
achieve the intended degree of interpretive resemblance.
Hence the actual resemblance may fall short of expecta-
tions and may also vary at different points in the text. As
230 Translation and Relevance

I have suggested in chapter 7 above (pp. 192-5), one of

the consequences of these difficulties is that the transla-
tor will do well to inform - where relevant - both his/her
client and the target audience about such limitations in
appropriate ways.
Evaluation - in The other major difficulty is that of contextual back-
the right context! ground differences; though it often affects the success of
the translation at least as much as language differences,
its full significance seems to have been recognized by
few practitioners and theoreticians alike. Yet, as should
be amply clear by now, no matter what the translator does
in the translated text itself, the understanding of the tar-
get audience will crucially depend on the context in which
it processes it. This context-dependence is built-into hu-
man nature and affects translation as much as it does any
other form of human communication.
Strangely, when evaluating translations in terms of
success or failure, this fact is largely ignored. 21 Translators
are often indiscriminately blamed for miscommunica-
tions that arise from the audience's failure to familiarize
themselves with the intended contextual background as
well as for actual translation mistakes. However, like any
other form ofhuman communication, translation can only
be expected to be successful when processed in the in-
tended context. Therefore, translation critics need to pay
careful attention to potential distortions due to contex-
tual differences.
More than However, it seems that not only translation critics need
translation to be alerted here. Since the distinctness of contextually
required conditioned communication problems from linguistically
conditioned ones has been largely overlooked by most

21 A significant part ofthe problem may be the common idea that communicators pro-
vide all the necessary contextual knowledge in the co-text, that is, in the surrounding
text. What is often not realized is that communicators will only provide such informa-
tion which they believe not to be (readily) available to the particular audience they have
in mind. When the audience changes, as is often the case in translation, there is no
guarantee whatsoever that the co-text will include all the background knowledge neces-
sary for the new target audience.
Postscript 231

previous and current accounts of translation, an aware-

ness needs to be created more generally that translation
is confronted not only with one barrier - the all too fa-
miliar language barrier - but also with the distinct second
barrier of contextual differences.
Authenticity and In line with the notion of translation as interlingual
freedom to adapt quotation, the primary responsibility of the translator is
are inverse to the mastery of the first barrier. Concerning the second
each other
barrier, it seems that the lower the degree of resemblance
required, the freer the translator is to adapt to the actual
context of the target audience. On the other hand, the
higher the degree of intended resemblance gets and the
greater the claim of authenticity, the more responsibility
falls on the audience to familiarize themselves with the
original context.
The translator's One of the consequences of this realization is that the
responsibility deployment of translation may require additional meas-
with regard to ures which lie outside of but are complementary to the
contextual
translation effort itself and which are designed to adjust
problems
the target audience's context as necessary. In simple
cases, the translator may consider providing this com-
plementary material him/herself. Where the contextual
differences are great, requiring the provision of exten-
sive information about the socio-cultural and historical
setting in which the original was written, this task may
go far beyond the translator's resources. It may require
the development of additional channels of communica-
tion accompanying the translation. Yet even where the
translator cannot himlherself provide the solution to these
contextual problems, they must be addressed by some-
one and s/he is in a unique position to recognize and draw
attention to them. 22
At the very least, the target audience must be made

22 Hence I do see the background knowledge of the target audience not as something
given or unchangeable - as e.g. Tirkkonen-Condit seems to assume, pointing out that
"it is not easy to envisage translation tasks in which one can reasonably expect the
context of reception to be identical" (1992:241) - but as a variable that may need to be
adjusted by adequate supplementary strategies in order for the translation to succeed.
232 Translation and Relevance

aware of these problems so that they will not naively as-

sume that the interpretation they derive whatever context
they happen to process it in is the intended meaning of
the original. If the translator is carrying out the job for
someone else, then s/he would do well to inform the client
of these problems and perhaps give advice on possible
solutions.

The subtle complexity of the translator's task

No word-by-word The relevance-theoretic notion of translation as inter-

translation lingual quotation may conjure up in some readers' minds
the abhorrent image of someone changing words, phrases,
sentences, paragraphs and larger units of speech or writ-
ing from one language to another. Indeed, there have been
those who have suspected that the relevance-theoretic
account would favour 'formal correspondence transla-
tion' or 'literal translation'. This suspicion has applied
especially to direct translation as an interlingual parallel
to direct quotation; after all, since direct quotation con-
sists of a word-by-word reproduction of the original
utterance, would not direct translation consist in a word-
by-word transfer into another language?
A brief look at the section entitled 'Direct transla-
tion - A special case of interpretive use' in chapter 7
will show that such fears are baseless. 23 As pointed out
there, direct translation cannot ultimately be defined by
correspondences between textual properties, not even by
correspondences between the more abstract 'communi-
cative clues'. Rather, "the notion of direct translation is
dependent on interpretive use: it relies ... on a relation-
ship of complete interpretive resemblance between the
original and its translation"(p. 170).24
Immersion in the What this means in practice is that competent trans-
original context lation necessarily involves the following components.
Beyond sufficient mastery of the original language, the

23 See also Gutt( 1998b).

24 Cf. also Gutt (forthcoming).
Postscript 233

translator needs to thoroughly familiarize himlherself with

the contextual background knowledge assumed by the
original communicator for the text in question. (This in
itself can be a formidable task, for example when one is
dealing with historical documents much of whose set-
ting is only partially known.)
Appreciating the Then, working hislher way through the text, the trans-
contextual effects lator needs to understand at each point what contextual
effects were intended in the original context and thereby
form a comprehensive hypothesis of the originally in-
tended interpretation of the original, consisting of both
explicatures and implicatures.
Aiming at close N ext, for each source language utterance, s/he needs
enough resem- to construct a target language utterance, selecting for it
blance in properties of the target language that will lead to an in-
interpretation
terpretation that resembles the original interpretation
closely enough to make it consistent with the principle
of relevance for the target audience.
In the case of direct translation, which presumes and
aims at complete interpretive resemblance with the origi-
nal, the translator will aim to design the target language
utterances in such a way that - in the ideal case - they
will lead to the same explicatures and implicatures as the
original, that is, to the same overall interpretation, when
processed in the originally intended context.
Just imagine ... ! This means, in effect, that the translator produces a
target language expression that seems suitable, then - in
hislher mind - imagines processing that expression in the
originally intended context, and compares the resulting
interpretation to that of the original. If satisfied with the
resemblance achieved, s/he will accept the target lan-
guage expression; otherwise slhe will attempt to improve
resemblance by changing the properties of the target
expreSSIOn.
Out olone's From this it should be clear that the focus of relevance-
mind? theory based translation is on the comparison of inter-
pretations, not on the reproduction of words, linguistic
constructions, or textual features.
234 Translation and Relevance

Viewed in this way, translation is indeed a tall order

and it draws the researcher's attention to a number of
interesting questions about the capabilities of the human
mind. For example, considering the potential complexity
of contextual background knowledge, it would appear
that attempting to process an utterance in an imagined
context, different from the background knowledge one
actually has, would make enormous demands on mental
processing capabilities. How do our minds cope with
these demands? 25 The bulk of the mental activities of
translation (as of communication in general) takes place
below the level of consciousness; this raises the question
to what degree translators can make themselves aware of
these processes?26
Beyond Furthermore, how does relevance as an experience
intuition? enter into this process? Are intuitions of relevance the
same regardless of whether we process an utterance in
an imagined context or in the real, actually available con-
text? Constructing a context different from the one readily
available would naturally seem to require more effort than
simply using what is there, and if relevance is sensitive
to processing effort - as the current version of relevance
theory assumes, but see Gutt (l998a) - then one would
expect differences in the intuitive experience of relevance.
What would this mean for the translator's reliance on
intuitions of relevance?
Not so strange These questions are, of course, not limited to translation
after all but affect the processing of utterances in communication
in general: it is a crucial part of any communicator's task
to anticipate the context actually available to the audi-
ence and to design the stimulus accordingly. Experience
seems to indicate that human beings can cope with this
fairly well, though with considerable individual variation.
It will be interesting to find out how such complex tasks
are achieved.

25Wilson (1998) approaches some of these questions under the heading 'metarepresentations' .
26The largely subconscious nature of mental processes in communication appear to be
one reason why think-aloud-protocols have remained of limited significance for trans-
lation research and practice.
Postscript 235

A Word on Cultural Translation Studies

Some critics have alleged that the relevance-theoretic

account claims "that translation theory and translation
studies should have been made redundant by relevance
theory" (Malmkj<er 1992:307; similarly Tirkkonen-
Condit 1992).
No needfor a To get this issue clear, we must first consider the un-
separate derstanding of the term 'translation theory'. The declared
explanation of aim of this book is to propose an explanatory theory in
translation as
the sense of a cause-effect account of translation as a
communication
phenomenon of communication. This theory is intended
to explain how it is possible for a human communicator
to convey to an audience in language B what someone
expressed in a different language A, what chain of cau-
sality in the human mind makes that possible, what factors
contribute to its success or failure, and how. If that is
what one expects a 'translation theory' to do, then my
claim is indeed that there is no need for developing a
separate theory of translation, with concepts and a theo-
retical framework of its own. In that particular sense I
do indeed claim that there is no need for a general theory
of translation (cf. p. 199 above).
... but for inte- If, however, one understands the term 'translation
grating its theory' in a wide, encyclopaedic sense as a synonym of
communicative 'translation studies', that is, as an organized investiga-
essence
tion into any phenomena associated in some way with
translating, translators, and translations, then there obvi-
ously is a place for a 'translation theory' - in that sense.
For example, I find the Routledge Encyclopedia of Trans-
lation Studies (Baker 1998) very useful. However, it still
seems to me that if such a general theory is to be com-
prehensive, it needs to address the communicative essence
of translation and in that regard "it will be impossible to
give an adequate account of translation without reference
to the principle of relevance"(p. 200 above).
Hopefully these remarks help to prove false the alle-
gation that this book declares or implies the redundancy
of translation studies in general.
236 Translation and Relevance

Conclusion

Overall, I see the relevance-theoretic account of transla-

tion offered in this book as having the following major
points in its favour:

1. In historical perspective, it seems that the relevance-

theoretic account has brought us a very significant step
closer to an understanding of the essence of transla-
tion. Despite longstanding and deeply ingrained
scepticism in translation circles about the possibility
of ever being able to define translation, relevance
theory has made it possible to propose a definition that
is simple, explicit- as part of an explicit general theory
of communication - and intuitively attractive: transla-
tion consists in interlingual quotation, that is it is an
instance of quotation - direct or indirect - where the
quote is in a different language from the original. In
my view, the relevance-theoretic account demonstrates
that, pace Hermans, attempts of defining translation
in an "essentialist, immanent way" need not be aban-
doned a priori after all and that they can lead to valid
and helpful results, unhampered by the prescriptive
views of how some "armchair critic thought [transla-
tions] should have turued out" (Hermans 1995:215f.).
2. It is competence-based.
3. It reveals the distinctness of translation, without
being Procrustean. Even phenomena that have been
called 'translation' by some but fall outside the scope
of the definition proposed here - e.g. phonemic trans-
lation, interlinear translation, covert translation - are
not simply ignored or discarded; they are still ac-
counted for within the wider framework of relevance
theory, and their perceived 'oddity' is explained.
4. The definition is communicatively well-motivated:
quotation - both indirect and direct - is very common
as a phenomenon of intra-lingual communication; it is
therefore understandable that human beings should em-
ploy similar modes of communication across language
boundaries.
Postscript 237

5. Through the relevance-theoretic framework, the ac-

count is empirical, rooted fmnly in cognitive psychology.
6. Through the relevance-theoretic framework, the
account is explanatory in terms of communicative
cause and effect, allowing also empirical predictions
with regard to the likely success or failure oftransla-
tion efforts.
7. As a bonus, this account also helps to understand
why earlier attempts at developing an explicit theory
of translation have kept running into serious difficul-
ties - mainly because of some inherent weaknesses of
the code-model.

It's all in From the point of view of theory construction, the ac-
the general count is particularly attractive for two reasons: first, given
framework
the general framework of relevance theory, no special,
additional concepts or theoretical tools are needed to ac-
commodate translation. In the development of science,
the explanation of seemingly special, complex phenom-
ena in terms of more general empirical factors already
known is a major step ahead. Secondly, it is attractive in
that the complexities lie in the general framework, the
concept of translation itself is simple.
Guidance where From the practitioner's point of view, an important
there is no break-through achieved by this account is that it allows
precedence the translator to understand and deal with the specifics
of translation wherever they are found. While it affirms
and validates the usefulness of rules and principles of
translation, it does not leave the translator alone in un-
precedented situations, where those tools do not exist or
fail to apply: the translator can always rely on the guid-
ance provided by the search for optimal relevance, taking
into account the specific contextual background of the
audience s/he is working for.
Assessing one's Since practising translators are usually concerned
task ahead of about the communicative results of their activities, the
time understanding of the communicative cause-effect relation
between original thoughts, original stimulus, translation
and the interpretation arrived at by the target audience
238 Translation and Relevance

helps them to realistically assess ahead of time what they

mayor may not be able to achieve.
A realistic idea Most importantly, they realize that language differ-
of the challenge ences are only one of the barriers that stand in the way of
communication across languages; the other, and some-
times more formidable, barrier is that of differences in
contextual background knowledge. Within the framework
of relevance theory, the practising translator will further-
more understand very clearly that such differences can
a) seriously disrupt the communication process, b) that-
as a matter of principle - they are problems that lie outside
the text to be translated, and that c) depending on the
degree of contextual differences, translations may need
to be accompanied by supplementary means of commu-
nication that take care of those contextual differences, if
there is to be communicative success worth speaking of.
Postscript 239

General References in Postscript

(Details for the references in chapters 1-7 are given in the 'General Bibliog-
raphy' (pp. 243-9).

Anderson, T. David (1998) 'Perceived Authenticity: The fourth criterion of good

translation', Notes on Translation 12(3): 1-13.
Baker, Mona (ed.) (1998) Routledge Encyclopedia of Translation Studies, Lon-
don & New York: Routledge.
Blass, Regina (1990) Relevance Relations in Discourse: A study with special
reference to Sissala, Cambridge: Cambridge University Press.
Chesterman, Andrew (forthcoming) 'What Constitutes "Progress" in Transla-
tion Studies?', Paper presented ASLA conference, Stockholm, 1998.
Dolitsky, M. (1992) Review of Hatim and Mason 1990 and Gutt 1991,Journal
of Pragmatics 17(2): 175-78.
Frawley, William (1992) 'Ernst-August Gutt, Translation and Relevance: cog-
nition and context', Review in Journal of Linguistics 28(2): 516-19.
Gutt Ernst-August (1992) Relevance Theory: A guide to successful communi-
cation in translation, Dallas, Summer Institute of Linguistics, and New York,
United Bible Societies.
------ (1996a) 'Relevance: A key to Quality Asessment in Translation', in B.
Lewandowska-Tomaszczyk and M. Thelen (eds) Translation and Meaning
Part 4. Proceedings of the Lodz Session of the 2nd Maastricht-Lodz Duo
Colloquium on 'Translation and Meaning', held in Lodz, Poland, 22-24
Sept. 1995,531-542.
------ (1996b) 'Implicit Information in Literary Translation: A relevance-theo-
retic perspective', Target: International Journal of Translation Studies 8(2):
241-56.
------ (1998a) 'Relevance and Effort - A paper for discussion', Paper presented
at the Relevance theory workshop, 8-10 September 1998, University ofLuton.
Unpublished manuscript.
------ (1998b) 'Pragmatic Aspects of Translation: some relevance-theoretic ob-
servations', in L. Hickey (ed) The pragmatics of Translation, Clevedon:
Multilingual Matters, 41-53.
------ (forthcoming) 'When is a Textual Property a Communicative Clue?', Pa-
per presented at the Susanne Hubner Seminar, December 1998, at the
University of Zaragoza.
Hermans, Theo (1995) 'Toury's Empiricism Version One', The Translator 1(2):
215-223.
Honig, Hans G. and Paul KuBmaul (1984) Strategie der Ubersetzung, Tubingen:
Gunter Narr Verlag.
Hymes, Dell (1992) Review ofGutt 1991, Language in Society 21(2): 316-17.
Malmkjrer, Kirsten (1992) 'Review: Translation and Relevance: cognition and
240 Translation and Relevance

context. by E.A. Gutt', Mind and Language 7(3): 298-309.

Muschard, Jutta (1995) Relevant Translations: history, presentation, criticism,
application (European University Studies, Series XXI, Linguistics, vol. 163),
FrankfurtlM.: Peter Lang.
Navarro-Errasti, Maria Pilar (1995) 'Communicative Clues in Sir Gawain and
the Green Knight', in Andreas H. Jucker (ed.) Historical Pragmatics. Prag-
matic developments in the history ofEnglish , Amsterdam: John Benjamins,
187-94.
Quine, W. V. (1959/1966) 'Meaning and Translation', inR. A. Brower (ed.) On
Translation, N ew York: Oxford University Press, 148-72.
Reboul, Anne and Jacques Moeschler (1997) 'Reduction and Contextualization
in Pragmatics and Discourse Analysis', Linguistische Berichte (Sonderheft
'Implicatures') , 283-95.
Reiss, Katharina and Hans J. Vermeer (1984) Grundlegung einer allgemeinen
Ubersetzungstheorie, Tubingen: Niemeyer.
Salkie, Raphael (1991) 'Deviations from the Literal', Review of Emst-August
Gutt, in Times Higher Education Supplement 26.7.1991.
Sim, Ronald J. and E.-A. Gutt (in preparation) Telling Translation: A coursebook.
To be published by Summer Institute of Linguistics.
Sperber, Dan and Deirdre Wilson (1986) Relevance: communication and cog-
nition, Oxford: Basil Blackwell (2 nd revised ed. 1995).
Stark, Detlef(1993) Review ofGutt 1991, Language 69(4): 857-58.
Tirkkonen-Condit, Sonja (1992) 'A Theoretical Account of Translation - with-
out translation theory?', Target 4(2): 237-45.
Toury, Gideon (1995) Descriptive Translation Studies and Beyond, Amster-
dam: John Benjamins.
Weber, David (1998) 'Jesus' Use of Echoic Utterances', Notes on Translation
12(2): 1-10.
Weizman, Elda and Shoshana Blum-Kulka (1987) 'Identifying and Interpreting
Translated Texts: On the role of pragmatic adjustment', in Gideon Toury
(ed.) Translation Across Cultures, New Delhi: Bahri Publications, 61-73.
Wilson, Deirdre (1998) 'Linguistics Metarepresentations', Paper presented at
the Relevance Theory Workshop, 8-10 September 1998, University ofLuton.
Unpublished manuscript.

Reviews and Literature with Comments on Gutt 1991

(This includes reviews and applications of the account of translation pro-

posed in Gutt 1991)

Anonymous (1992) Review of Gutt 1991, Language International 4(5): 41.

Abplanalp, Laure (1998) La pertinence et la traduction, Lausanne: Centre de
Traduction Litteraire, Universite de Lausanne.
Postscript 241

Baker, Mona (ed.) (1998) Routledge Encyclopedia of Translation Studies, Lon-

don & New York: Routledge.
Dolitsky, M. (1992) Review of Hatim and Mason 1990 and Gutt 1991, Journal
of Pragmatics 17(2): 175-78.
Frawley, William (1992) 'Ernst-August Gutt, Translation and Relevance: cog-
nition and context', Review in Journal of Linguistics 28(2): 516-19.
Hatim, Basil (1998) 'Pragmatics and Translation', in Mona Baker (ed.)
Routledge Encyclopedia of Translation Studies, London & New York:
Routledge, 179-83.
Hymes, Dell (1992) Review ofGutt 1991, Language in Society 21(2): 316-17.
Jurasek, B. S. (1993) Review ofGutt 1991, Modem Language Journal 77(2): 272.
Mackenzie, I. (1994) Review of Gutt 1991, Journal ofLiterary Semantics 23(3):
239-42.
Malmkjrer, Kirsten (1992) 'Review: Translation and Relevance: cognition and
context. by E.A. Gutt', Mind and Language 7(3): 298-309.
Mason, Ian (1998) 'Communicative/Functional Approaches', in Mona Baker
(ed.) Routledge Encyclopedia of Trans lation Studies, London & New York:
Routledge, 29-33.
Mikkelsen, Hans Kristian (1994) 'Book Review Ernst-August Gutt: Transla-
tion and Relevance: cognition and context', in Arnt Lykke Jakobsen (ed.)
Translating LSP texts: Some theoretical considerations (Copenhagen Stud-
ies in Language 16), Copenhagen: Copenhagen Business School, 143-49.
Muschard, Jutta (1995) Relevant Translations: history, presentation, criticism,
application (European University Studies, Series XXI, Linguistics, vol. 163),
Frankfurt/M.: Peter Lang.
Navarro-Errasti, Maria Pilar (1995) 'Communicative Clues in Sir Gawain and
the Green Knight', in Andreas H. Jucker (ed.) Historical Pragmatics. Prag-
matic developments in the history ofEnglish, Amsterdam: John Benjamins,
187-94.
Romney, C. (1993) Review of Gutt 1991, Canadian Journal ofLinguistics 38(1):
llO-l13.
Salkie, Raphael (1991) 'Deviations from the Literal', Review of Ernst-August
Gutt, in Times Higher Education Supplement 26.7 .199l.
Stark, Detlef(1993) Review ofGutt 1991, Language 69(4): 857-58.
Steiner, E. (1993) Review ofGutt 1991, Linguistics 31(3): 577-83.
Thomas, S. (1994) 'Relevance and Translation', Turjuman 3(2): 37-49.
Tirkkonen-Condit, Sonja (1992) 'A Theoretical Account of Translation - with-
out translation theory?', Target 4(2): 237-45.
van Grootheest, Dave (1996) Relevance Theory and Bible Translation: an ex-
ploratory study. M.A. thesis, Free University of Amsterdam.
Vermeer, Hans J. (1993) 'Rezension (Entwurf): Sperber, Dan and Wilson,
Deirdre: Relevance. Communication and cognition and Gutt, Ernst-Au-
gust: Translation and relevance: cognition and context', Heidelberg,
Unpublished manuscript.
242 Translation and Relevance

Wendland, Ernst R. (1996) 'Relevance of "Relevance Theory"', The Bible Trans-

lator 47(1): 126-37.
------ (1996) 'A Review of "Relevance Theory" in relation to Bible Translation
in South-Central Africa', Journal of Northwest Semitic Languages 22(1):
91-106.
Wilss, Wolfram (1993) 'Ernst-August Gutt, Translation and Relevance: cogni-
tion and context', Review, Lebende Sprachen 1(93): 38-39.
Winckler, W.K, van der Merwe, C.R.J. (1993) 'Training Tomorrow's Bible Trans-
lators: Some theoretical pointers', Journal of Northwest Semitic Languages
19: 41-58.
General Bibliography
Adams, Robert M. (1973) Proteus, his Lies, his Truth: Discussions ofLiterary
Translation, New York: Norton.
Arnold, Matthew (1861) 'On Translating Homer', in On the Study of Celtic
Literature and on Translating Homer, New York, 1906, 141-300.
Arrowsmith, William and Roger Shattnck (1964) The Craft and Context ofTrans-
lation: A critical symposium, New York: Anchor Books.
Barnwell, Katharine (1983) 'Towards acceptable translations', Notes on Trans-
lation 95: 19-25.
Bassnett-McGuire, Susan (1980) Translation Studies, London: Methuen.
Bausch, K.-R., J. Klegraf, and W. Wilss (1970172) The Science of Translation:
An analytical bibliography, 2 vols., Tubingen: Narr.
Beekman, John and John Callow (1974) Translating the Word of God, Grand
Rapids, Mich.: Zondervan.
Bell, Roger T. (1986) 'Why Translation Theory is in a Mess and What We can
do about it', in Proceedings of the GALLA Congress, Thessaloniki 1985,
GALLAIAILA,280-87.
Blakemore, Diane (1987) Semantic Constraints on Relevance, Oxford:
Blackwell.
------ (1988a) "So' as a Constraint on Relevance', In Ruth Kempson (ed.) Men-
tal Representations: The interface between language and reality, Cambridge:
Cambridge University Press, 183-95.
------ (1988b) 'The Organization of Discourse', In F. J. Newmeyer (ed.) Lin-
guistics: The Cambridge survey, vol. 4 (Language: The socio-cultural
context), Cambridge: Cambridge University Press, 229-50.
Blass, F. and A. Debrnnner (1961) A Greek grammar of the New Testament and
other early Christian literature, Chicago: University of Chicago Press.
Blass, Regina (1986) 'Cohesion, Coherence and Relevance', Notes on Linguis-
tics 34: 41-64.
------ (1988) Discourse Connectivity and Constraints on Relevance in Sissala,
University College London doctoral dissertation.
------ (1990) Relevance Relations in Disourse: A study with special reference to
Sissala, Cambridge: Cambridge University Press.
Brislin, Richard W. (1976) 'Introduction', In R. W. Brislin (ed.) Translation:
Applications and research, N ew York: Gardner Press, 1-43.
------ (1978) 'Contributions of Cross-Cultural Orientation Programs and Power
Analysis to Translation/Interpretation', In D. Gerver and H. W. Sinaiko (eds)
Language interpretation and communication. Proceedings of the NATO
Symposium on Language and Communication, Venice, 26.9. -1.10 .1977 , New
York: Plenum Press, 205-216.
van den Broeck, Raymond (1980) 'Toward a Text-type-oriented Theory of Trans-
lation', In S.-O. Poulsen and W. Wilss (eds) Angewandte Ubersetzungswissen-
schaft. Internationales Ubersetzungswissenschaftliches Kolloquium an der
Wirtschaftsuniversitat ArhuslDanemark, 19.- 21. Juni 1980,82-96.
244 Translation and Relevance

Callow, Kathleen (1998) Man and Message, Lanham, MD/Dallas: University

Press of America and Summer Institute of Linguistics.
Carson, D. A. (1985) 'The Limits of Dynamic Equivalence in Bible Transla-
tion', Evangelical Review of Theology 9(3): 200-213.
Carston, Robyn (1988a) 'Implicature, Explicature, and Truth-Theoretic Seman-
tics' , in Ruth Kempson (ed.) Mental Representations: The interface between
language and reality, Cambridge: Cambridge University Press, 155-8l.
------ (1988b) 'Language and Cognition', In F. J. Newmeyer (ed.) Linguistics:
The Cambridge survey, vol. 3 (Language: Psychological and biological as-
pects), Cambridge: Cambridge University Press, 38-68.
------ (1996) 'Enrichment and Loosening: complementary processes in deriving
the proposition expressed?', UCLWorking Papers in Linguistics 8: Univer-
sity College London, Dept. of Phonetics & Linguistics, 205-232.
Chomsky, Noam (1966) Current Issues in Linguistic Theory, The Hague:
Mouton.
------ (1980) Rules and Representations, Oxford: Blackwell.
------ (1986) Knowledge of language: Its nature, origin and use, New York:
Praeger.
Chukovskii, Kornei (Trl. & ed. Lanren G. Leighton) (1984) The Art ofTransla-
tion, Knoxville: University of Tennessee Press.
Deibler, Ellis W. jr. (1988) 'Adding Contextual Information', Notes on Trans-
lation 2(3): 24-34.
Dooley, Robert A. (1989) 'Style and Acceptability: the Gnrani New Testament',
Notes on Translation 3(1): 49-57.
Dye, Wayne T. (1980) The Bible Translation Strategy, Dallas: Wycliffe Bible
Translators.
Flores d' Arcais, G. B. (1978) 'The Contribution of Cognitive Psychology to the
Study of Interpretation', in D. Gerver and H. W. Sinaiko (eds) Language
interpretation and communication, New York: Plenum Press, 385-402.
Fodor, Jerry A. (1983) The Modularity ofMind, Cambridge, Mass.: MIT Press.
France, Richard T. (1981) 'The Formula-Quotations of Matthew 2 and the Prob-
lem of Communication' ,New Testament Studies 27: 233-5l.
Frawley, William (1984) 'Prolegomenon to a Theory of Translation' , in William
Frawley (ed.) Translation: Literary, linguistic and philosophical perspec-
tives, London: Associated University Press, 159-75.
Glover, Warren W. (1978) 'The Kingdom of God' in Luke', The Bible Translator
29(2): 231-37.
von Goethe, Johann Wolfgang (1813) 'Zu Briiderlichem Andenken an Wieland',
reprinted in H. J. StOrig (ed.) Das Problem des Ubersetzens, Darmstadt:
Wissenschaftliche Buchgesellschaft, 1969,34.
Grassegger, Hans (1985) Sprachspiel und Ubersetzung: Eine Studie anhand
der Comics-Serie "Asterix ", Tubingen: Stauffenbnrg.
Grice, H. P. (1975) Logic and Conversation, reprinted in A. P. Martinich (ed.),
The philosophy oflanguage. Oxford University Press, Oxford, 1985, 159-70.
Gutt, Ernst-August (1982) 'Information Sequencing in Mark, chapters 7 and 8',
General Bibliography 245

Notes on Translation 92: 6-27.

------ (1983) 'Studies in the Phonology of Silti', Journal of Ethiopian Studies
16: 37-73.
------ (1985) 'Relevance Theory and Increased Accuracy in Translation', Notes
on Translation 107: 29-3l.
------ (1986) 'Matthew 9:4-17 in the Light of Relevance Theory', Notes on Trans-
lation 113: 13-20.
------ (1987a) 'What is the Meaning We Translate?', Occasional Papers in Trans-
lation and Textlinguistics 1: 31-58.
------ (1987b) 'The Dependence of Translation on Linguistic Semantics'. Un-
published manuscript.
------ (1988a) 'Towards an Analysis of Pragmatic Connectives in Silt'i', inPro-
ceedings ofthe Eighth International Conference ofEthiopian Studies, Addis
Ababa University, 26-30 November 1984, Vol. 1, Huntingdon: ELM Publi-
cations, 665-78.
------ (1988b) 'From Translation to Effective Communication', Notes on Trans-
lation 2(1): 24-40.
------ (1998) 'Relevance and Effort - a paper for discussion'. Paper presented at
the Relevance Theory Workshop, 8-10 September 1998, University of Luton.
Harris, Brian and Bianca Sherwood (1978) 'Translating as an Innate Skill', In
D. Gerver and H. W. Sinaiko (eds) Language Interpretation and Communi-
cation, New York: Plenum Press, 155-70.
Headland, Thomas (1981) 'Information Overload and Communication Problems
in the Casiguran Dumagat New Testament', Notes on Translation 83: 19-27.
Hebel, Johann Peter (1972) 'Kindesdank und Undank', in H. Steinhauer (ed.)
First German eader, New York: Bantam Books, 6-10.
Honig, Hans G. and Paul KuBmaul (1984) Strategie der Ubersetzung, Tubingen:
Gunter Narr Verlag.
Hofmann, Norbert (1980) Redundanz und .A'quivalenz in der Literarischen
Ubersetzung: dargestellt an funf deutschen Ubersetzungen des Hamlet,
Tubingen: Niemeyer.
Hohulin, Richard (1979) 'The Logic of Logic in Translation' , Notes on Trans-
lation 77: 29-40.
Holz-Manttari, Justa (1984) Translatorisches Handeln. Theorie und Methode,
Helsinki: Suomalainen Tiedeakatemia.
House, Juliane (1981) A Modelfor Translation Quality Assessment, Tubingen:
Gunter Narr.
Kabatek, Adolf(1988) 'Was ist neu am neuen Asterix?', in R. Goscinny and A.
Uderzo Asterix der Gallier, Asterix Band I, Jubilaumsausgabe 1968-1988,
Stuttgart: Delta, 3.
Kade, O. (1968a) Zufall und GesetzmCiJ3igkeit in der Ubersetzung, Leipzig: VEB
Verlag Enzyklopadie.
------ (1968b) 'Kommunikationswissenschaftliche Probleme der Translation',
in A. Neubert (ed.) Grundfragen der UbersetzungswissenschaJt, Beihefte
zur Zeitschrift Fremdsprachen II, Leipzig: VEB Verlag Enzyklopadie, 3-20.
246 Translation and Relevance

Kassuhlke, Rudolf (1974) 'An Attempt at a Dynamic Equivalent Translation of

Basileia tou Theou', The Bible Translator 25(2): 236-38.
Kelly, Louis Gerard (1979) The True Interpreter: A history oftranslation theory
and practice, Oxford: Blackwell.
Kloepfer, W. (1967) Die Theorie der Literarischen Ubersetzung, Munich: Fink.
Koller, W. (1972) Grundprobleme der Ubersetzungstheorie: Unter besonderer
Berucksichtigung schwedisch-deutscher Ubersetzungsfalle, Berne: Franke.
------ (1983) Einfuhrung in die Ubersetzungswissenschaji, Heidelberg: Quelle
& Meyer.
Krings, Hans P. (1986) Was in den Kopfen von Ubersetzern vorgeht, Tubingen:
Gunter N arr.
KuBmaul, Paul (1986) 'Ubersetzen als EntscheidungsprozeB. Die Rolle der
F ehleranalyse in der Ubersetzungsdidaktik', In M. Snell-Hornby (ed.)
Ubersetzungswissenschaji - eine Neuorientierung. Zur Integrierung von
Theorie und Praxis, Tubingen: Francke, 206-29.
Lakatos, Imre (1978) The Methodology of Scientific Research Programmes,
Philosophical Papers, vol. 1 (edited by John Worrall and Gregory Currie),
Cambridge: Cambridge University Press.
Larson, Mildred L. (1984) A Guide to Cross-Language Equivalence, New York:
University Press of America.
Lefevere, Andre (1975) Translating Poetry: Seven strategies and a blueprint,
Amsterdam: Van Gorcum.
Lehmann, D. (1982) Arbeitsbibliographie Ubersetzen: Interdisziplinare Aspekte
der Sprach- und Ubersetzungswissenschaji sowie der Ubersetzungspraxis,
Laut-Paper 83, Series B. Trier.
Levy, Jii'i (1967) 'Translation as a Decision Process', in To Honor Roman
Jakobson. Essays on the occasion of his seventieth birthday, 11 th October
1966, vol. II, ParislDen Haag, 1171-82.
------ (1969) Die literarische Ubersetzung: Theorie einer Kunstgattung, Frank-
furt: Athenaum.
Lyons, John (1969) Introduction to Theoretical Linguistics, Cambridge: Cam-
bridge University Press.
Massaro, Dominic W. (1978) 'An Information-Processing Model of Understand-
ing speech', in D. Gerver and H. W. Sinaiko (eds) Language interpretation
and communication, New York: Plenum Press, 299-314.
Moser, Barbara (1978) 'Simultaneous Interpretation: A hypothetical model and
its practical application', in D. Gerver and H. W. Sinaiko (eds) Language
interpretation and communication, New York: Plenum Press, 353-68.
Mounin, George (1967) Die Ubersetzung: Geschichte, Theorie, Anwendung,
Munchen: Nymphenburger.
Namy, Claude (1978) 'Reflections on the Training of Simultaneous Interpreters:
A metalinguistic approach', in D. Gerver and H. W. Sinaiko (eds) Language
interpretation and communication, New York: Plenum Press, 25-34.
Neubert, A. (1968) 'Pragmatische Aspekte der Ubersetzung', in Neubert (ed.)
Grundfragen der Ubersetzungswissenschaft, Beihefte zur Zeitschrift
General Bibliography 247

Fremdsprachen II, Leipzig: VEB Verlag Enzyklopadie, 21-33.

Newmark, Peter (1988) Approaches to Translation, Hemel Hempstead:
Prentice Hall.
Nida, Eugene A. (1964) Toward a Science of Translating: With special reference
to principles and procedures involved in Bible Translating, Leiden: Brill.
------ (1988) 'Intelligibility and Acceptability in Bible Translating', The Bible
Translator 39(3): 301-308.
Nida, Eugene A. and Charles Taber (1969) The Theory and Practice ofTrans-
lation, Leiden: Brill.
O'Flaherty, Wendy Doniger (1971) 'A New Approach to Sanskrit Translation,
Applied to Kalidasa's Kumarasambhava Canto VIII', MAHFIL: A Quar-
terly of South Asian Literature 7(3-4): 129-42.
------ (1987) 'On Translating Sanskrit Myths', in W. Radice and B. Reynolds
(eds) The translator's art: Essays in honour of Betty Radice, Penguin:
Harmondsworth,121-28.
Ott, Willis (1984) 'A New Look at the Concept of the Kingdom of God', Notes
on Translation (Special edition no.2).
Payne, David (1988) 'On Promoting Vernacular Scriptures where a National
Language Translation is Venerated', Notes on Scripture in Use 17: 18-23.
Peacock, H. F. (1981) 'Current Trends in Scripture Translation', Bulletin United
Bible Societies 1241125: 5-9.
Plummer, Alfred (1922) A Critical and Exegetical Commentary on the Gospel
According to S. Luke, Edinburgh: Clark (5th edition).
Reiling, J. and J. R. Swellengrebel (1971)A Translator's Handbook on the Gos-
pel ofLuke, Leiden: Brill.
Reiss, Katharina (1981) 'Type, Kind and Individuality of Text. Decision mak-
ing in translation', Poetics Today: Theory and Analysis of Literature and
Communication 2(4): 121-3l.
Reiss, Katharina and Hans J. Vermeer (1984) Grundlegung einer allgemeinen
Ubersetzungstheorie, Tubingen: Niemeyer.
Savory, Theodore (1957) The Art of Translation, London: Cape.
Schleiermacher, Friedrich (1838) 'Ueber die verschiedenen Methoden des
U ebersetzens', reprinted in H. J. StOrig (ed) Das Problem des Ubersetzens,
Darmstadt: Wissenschaftliche Buchgesellschaft, 1969,38-70.
Schulte, Rainer (1987) 'Translation Theory: A Challenge for the Future', Trans-
lation Review 23: 1-2.
Smith, Neil V. and Deirdre Wilson (1979) Modern Linguistics: The results of
Chomsky's revolution, Harmondsworth: Penguin.
Snell, B. M. and P. Crampton (1983) 'Types of Translation' , in Catriona Picken
(ed.) The Translator's Handbook, London: Aslib, 109-120.
Sperber, Dan and Deirdre Wilson (1986a) Relevance: Communication and Cog-
nition, Oxford: Blackwell.
------ (1986b) 'Loose Talk', Proceedings of the Aristotelian Society 86: 153-7l.
------ (1987) 'Precis of Relevance: Communication and Cognition', Behavioural
248 Translation and Relevance

And Brain Sciences 10: 697-754.

Staiger, E. (1956) Grundbegriffe der Poetik, Zurich: Atlantis.
Steiner, George (1975) After Babel: Aspects ofLanguage and Translation, Ox-
ford: Oxford University Press.
De Sua, William J. (1964) Dante into English, Chapel Hill:University of North
Carolina Press.
Svejcer, A. D. (1981) 'Levels of Equivalence or Translation Models?', in W.
Kuhlwein, G. Thome and W. Wilss (eds) Kontrastive Linguistik und
Ubersetzungswissenschaft. Akten des Internationalen Kolloquiums Trier/
Saarbrucken, 25.-30.9.1978, Munchen: Fink, 320-23.
Toury, Gideon (1985) 'A Rationale for Descriptive Translation Studies', In Theo
Hermans (ed.) The manipulation of literature: Studies in literary transla-
tion, London: Croom Helm, 16-41.
von Tscharner, Eduard Horst (1969) 'Chinesische Gedichte in deutscher
Sprache', in H. J. StOrig (ed.) Das Problem des Ubersetzens, Darmstadt:
Wissenschaftliche Buchgesellschaft, 242-72.
Turner, Nigel (1976) Style, Vol. 4 of James H. Moulton (ed.) A grammar of
New Testament Greek, Edinburgh: Clark.
Vernay, Heuri (1974) 'Elemente einerUbersetzungswissenschaft', in Wolfram
Wilss (ed.) Ubersetzungswissenschaft, Darmstadt: Wissenschaftliche
Buchgesellschaft,236-49.
de Waard, Jan and Eugene A. Nida (1986) From one Language to Another:
Functional equivalence in Bible translating, Nashville: Nelson.
Wendland, Ernst R. (1987) The Cultural Factor in Bible Translation: A study
of communicating the Word of God in a Central African cultural context,
London: United Bible Societies.
Wilson, Deirdre and Dan Sperber (1985) 'An Outline of Relevance Theory', in
Encontro de Linguistas Aetas, Unidade Cientifico-Pedag6gica de Letras e
Artes, Minho, Portugal: Universidade do Minho, 21-4l.
------ (1988a) 'Representation and Relevance', in Ruth M. Kempson (ed.) Men-
tal Representations: The interface between language and reality, Cambridge:
Cambridge University Press, l33-53.
------ (198 8b) 'Mood and the Analysis of non-Declarative Sentences', in J. Dancy,
J. Moravcsik and C. C. W. Taylor (eds) Human agency: language, duty and
value, Stanford: Stanford University Press, 77-101.
Wilss, Wolfram (1982) The Science of Translation: Problems and Methods,
Tubingen: Gunter Narr.
------ (1988) Kognition und Ubersetzen: Zu Theorie und Praxis der mensch lichen
und der maschinellen Ubersetzung, Tubingen: Niemeyer.
Winer, G. B. (1882) A Treatise on the Grammar of New Testament Greek, re-
garded as a sure basis for New Testament exegesis, Edinburgh: Clark.
Wonderly, William L. (1968) Bible Translations for Popular Use, London:
United Bible Societies,.
General Bibliography 249

Yuasa, Nobuyuki (1987) "'The Sound of Water": Different versions of a hokku

by Basha', in W. Radice and B. Reynolds (eds) The Translator's Art: Es-
says in honour ofBetty Radice, Hannondsworth: Penguin, 231-40.
Zukofsky, Celia and Louis Zukofsky (1969) Catullus, London: Cape Goliard
Press.
Index

abridgement 225
acceptability 93, 182ff, 228
accuracy
descriptive 59, 217-9
and form/meaning distinction 183-4
and style 131
action theory 16, 19,23,203-4
Adams, R.M. 107-11, 176, 179
adaptations 49, 82-3, 123
adequacy 19, 106,218,221
of contextual effect 32-5, 40, 67, 96, 101-3, 106-8, 110-1, 116, 118,
122, 158, 172, 178
and equivalence 12-3, 16
of performance 63
adj ective, postposition 147, 149
advertising 12,54-9,64,66-7,121,218,220-1
ambiguity 39, 76-7, 123
Amharic 156,223
analysis, componential 80, 82
approximation 97-8
arbitrariness 91, 101
art, translation as 2, 9
artificial intelligence (AI) 172
assumptions, contextual 27,29,33,39,43,45-6, 76-79, 122, 124, 139,
142,199,207
accessibility 27-8,30,34, 118, 153
communicated/non-communicated 40-2,46, 102, 117
and direct quotation 169
elimination 29-31
explication 80, 93-101, 121, 180
implicated 46, 116, 180
andmeaningoforiginal72, 76-7,95-100,120,142,169
and memory 158, 162
modification 28
Index 251

relevance 31, 66, 102, 106, 108, 116-7

schemas 26, 135, 138, 157
strength 29-31, 90, 176
audience
awareness of translation 199,231-2
competence 205
expectations 11,33,93,110,149,168,181-2,184,186-8,190-7,199,
206,225
and implicit information 40,84,87,91, 164, 180
interests 11-2,60,69, 100, 116, 118, 121, 123, 192,234
and meaning of original 102-4, 123, 171, 174, 186,235
original 75-9,81,99, 102-4, 140
response 10, 70-1, 82, 96, 110-1, 184,206
responsibility 34, 143, 174-5, 180, 191-2,231
authentic
meaning 186-7, 200, 228-9, 231
experience 229

background, contextual (see context)

back-translation 177-79
Barnwell, K. 182-4, 195
Bassnett-McGuire, Susan 1, 123
Bausch, K.-R., K1egraf, J. and Wi1ss, W. 1
Beekman, J. and Callow, J.
on dynamics of translation 95
on equivalence 119-20
on historical fidelity 120
idiomatic approach 71
on implicit information 83-5, 89,92, 94
on meaning 72, 85, 102
on metaphor 89
behaviour, and communication 21-3
Bell, Roger T. 2-3
benefits (see effort, processing)
Bible 62, 69-71, 100-2, 125
direct/indirect translations 180-6, 193-4
original and translation 65-6, 72-5
252 Translation and Relevance

and secondary communication 78-83, 90-4, 96-8

and style of writer 130-1, 146-9
Blakemore, D. 26, 43, 151, 153
Blass, F. and Debrunner, A. 147, 148-9
Blass, R. 43, 151,207
Bobrov, S.P. and Bogos1ovskaja, M.P. 145
Braine, John 11
Brislin, R.W. 62, 78
Van den Broeck, R. 5,21

c
causation (cause-and-effect) 235, 237
Chinese poetry 188-9
CORT (see research programme, competence-oriented)
Callow, J. n
Callow, K. 85
Carson, D.A. 181-2
Catford, J.C. 202
Catullus, Gaius Valerius 136, 178
change
linguistic 82
syntactic 144-5
see also form; function; meaning
Chesterman, A. 220
Chomsky, N. 17-8,21,25
Chukovskii, K. 131-2, ,145,161,167
classification 12, 17-21,207
clefting 135, 170
client, of translator 63,219-20,230,232
clues, communicative 132-36, 166-7
and direct translation 168, 170, In, 176,232
and formulaic expressions 155-9
and onomatopoeia 159-60
and phonetic properties 150-1
and semantic constraints on relevance 151-5
and semantic representations 13 8-44
and sound-based poetic properties 162-6
and stylistic value of words 160-2
Index 253

and syntactic properties 144-50

cognition, and relevance theory 21-2,25-6,27,237 (see also mind)
commentator, role 69
communicability, of contents 103, 174, 190-1
communication
code mode122, 174,208,237
competence 21-2, 205-6, 236
as context-dependent 104, 230
effectiveness 208, 218
expressive 70
failure 92, 118, 180-90, 196
imperative 70
as inferential 22, 24-5, 28, 100, 104, 199-200
interlingua1 55, 56, 61, 64-8, 171, 178, 195,208-9,228
intralingua1 61, 66, 102, 132, 171, 174
non-verba124-5, 42-3, 164
in relevance theory 98, 227
successful179-98, 213, 218, 221, 229-30, 235, 237-8
verba122, 24-6, 39, 76, 136, 155, 161-2, 163,229
translation as 22-3, 100, 140
weak 9-1, 175-6
communication situations
primary 76, 82, 98, 103
secondary 76-97,99-100, 174
communicator
responsibility 34, 66
see also translator
companson
in translation 87-8, 121
of translation input and output 204
of translations 13-14, 130, 180, 194
compensation 49-50
competence
communicative 21, 22, 205
engineering 219
linguistic 192,200,219,225,232
translational 19,22,205,232-4
concept
in relevance theory 141-3, 155-9
254 Translation and Relevance

of word 159, 160

connectives, pragmatic 43-4, 151, 154, 164-6
consciousness 234
constancy, functional 55 (see also function)
content
communicability 103, 174
source/translation 56-61, 63-7, 93
context
adjustment 174-5,231
and content 142-3
cultural 27-8, 48-50, 57, 63-4, 67, 70, 78-9, 80, 82, 84, 94-5, 101, 108,
173,181-2,192,204,206,208,225,231
definition 26
difference (see difference, contextual)
and implicit information 88
intended 225-6
and misinterpretation 76-7, 94, 222
literary 204, 225
modification 29 (see also effects, contextual; implications, contextual;
effect, contextual)
original 173-7,225
and relevance theory 26-35,37-8, 76-7
receptor 207, 226, 237
situational (external) 204
translator 207
see also co-text; effects, contextual
co-text (as special kind of context) 230
culture 19,57,64,67, 173, 181-2, 192
and covert translation 47-50
and definition of translation 5-7, 212
and dynamic equivalence approach 79-82
and figurative language 82
and misinterpretation 94-5
and preservation of meaning 101, 120
see also context, cultural
currency, in translation 126

decision-making 7-9, 11, 14, 19-22, 108, 113-5, 118-9, 190,226

Index 255

decoding, and interpretation 22, 76,212

Deibler, E.W. jr 182
Descriptive Translation Studies 6-8 (see also research programme)
descriptive-classificatory approach 17-20,21, 198
Dickens, Charles 131, 144-6
difference
contextual 48-50, 57, 63-4, 76, 78-9, 82, 95, 120-2, 124, 139-40, 175,
179-81,186,188,194,196,231,238
interlingua195, 146, 177, 179, 183, 195,228-31,238
discourse 72, 79, 93
features 167
markers 151-5
distortion 93, 101
Do1itsky, M. 224
domain, choice 4-7, 17-23
Dooley, R.A. 193-4
Doyle, A.C. 187-8
drama, translation 15, 125-6, 144
Dumagat people 97-8, 102-3
Dye, W. T. 96, 98
dynamics of translation 71-2,80,85,95 (see also equivalence, dynamic)

effect, contextual (communicative) 29-35, 40, 43, 101-3, 106-8, 110, 118,
120, 122, 124, 133-4, 147-8, 157-8, 163-5, 169, 178, 191,206,208,233
increased 118, 148
and inference 157
special145
and speech quotation 133-4
see also adequay, of contextual effects
effort, processing 28,31-5,41,44,81,96,234
increased 33-4, 44,81, 110, 118, 147-8, 192-4, 196,234
and interpretive use 106-11, 118, 120-2, 190
minimal 32, 106-7
reduction 35,82, 103, 134, 149, 157
and unnaturalness 122
emphasis 147
256 Translation and Relevance

entry
encyclopaedic 109-10, 141-4, 155-6, 158-9
lexical 141, 142, 155
logical141-4,155
environment, cognitive 82, 101, 116, 118, 120, 124-5, 128, 139-40, 199,
213,222,226
definition 27
and direct translation 175-6, 181, 186, 188, 190
mutua127, 41, 102, 124,213,222
and stimulus 139-40, 199
equivalence 10-17, 119-20, 137, 144,208,218
dynamic 69-71, 75, 79-80,82-3,97-8,181,185,202,203,209
as evaluative 12-14
formm 82,125,202,232
functional 10, 12, 14, 18,47-54,56-64,208,209
and hierarchical structure 14-17
and over-specification 11-12, 20
error, translation 13,49, 154, 174
Ethio-Semitic languages 7, 135-8
evaluation 8-9, 10, 12-14, 19-20,49,63,67,91, 198
expectations, of translation 11, 56,149,159,168,184-6,192-7,199,200,
206,225,228-9
explication 46,49,91-2, 168, 175-6, 180, 182
and communicative approach 140
and faithfulness 121
in idiomatic approach 91-4,97
as unnecessary 175
see also information, explicit
exp1icatures, 46, 225, 233
as analytic implication 40
and context 46, 76-7
and translation 99-101, 105, 121, 168, 177, 196
and propositional form 77
see also explication
explicitness, of verbal communication 24
expressions, formulaic 42, 155-8, 162

failure, translation 118, 179-89, 196, 199,230,235,237

Index 257

faithfulness,
historical 120-1, 181-2, 233
in interpretation 10,41,59,64, 71, 80, 106-8, 125, 127
to semantics 136-38, 143
of style 132
fidelity (see faithfulness)
footnotes 196
form
change 80, 89, 125, 183-4
and meaning 85-6, 87, 182-4, 89, 194
and translation 16, 69-71, 102
form, propositiona124-5, 76-7,138,149,155
and formulaic expressions 158
and interpretive resemblance 36-8, 40, 41-46, 215
formulaic expression (see expressions)
France, R.T. 73-5, 78, 90, 92
Frawley, W. 101,224
function
abstract 115-7
change 55,65,220
and equivalence 10, 12, 14,47-54,56-61
hierarchy 14-17, 113-5
semantic 113-5

game theory 8
gloss 177-8
Glover, W.W. 185-6
Goethe, J.W. 58, 174, 189
Goscinny, R. and Uderzo, A., Asterix 137, 194-5
Grassegger, H. 137
Greek, of New Testament 84,94, 130, 146-8, 177
greetings 42, 155, 158, 162
Grice, H.P. 35
Guarani language 193-4
Gurung language 185-6
Gutt, E.-A. 20, 31, 43,94,97, 135, 138, 151, 167, 174, 180, 182, 187,
202,223,232,234
258 Translation and Relevance

Headland, T, 97-8, 102

Hebbel, F. 165
Hebel, Peter 151-2
Hermans 6, 179, 184, 190,203,223,236
hierarchy, text-specific 14-17,20, 113-5
history, and faithfulness (see faithfulness, historical)
Hofinann,N. 2,15,16, 125-6
Hohulin, R. 180
Holz-Manttari, J. 19-20,23
Honig, H.G. & Kui3maul, P. 14,54-7,193,220
House, J. 10, 12-14,47-51, 154
Hunt, G. 101
Hymes, D. 224

lfugao language 180-2, 188

implications,
analytic 37-41, 44-6, 116, 134-5
contextual 29, 37-42, 44, 79, 81,97, 109-11, 157
synthetic 37
implicatures 44-6, 77, 88-92,99, 105, 109, 138, 149, 168,222,233
additional 44
as context-dependent 46, 76
definition 40
and faithfulness 121
and translation 99-101
strength 90-2, 101, 157, 164, 166, 175,225
indeterminacy 90, 92, 100, 189
inference 29-31,32,37,76,78-7,80,90,154,157
communication as inferential 24-5, 27, 100, 104, 199,212
see also implications, contextual
information
absent 87, 89, 120, 161
additional 80, 93
associated 109
contextual 27, 83, 94, 166, 174-6, 180
Index 259

explicit 49, 84-6, 86, 90-1, 99-100, 184

implicit 32, 48,84,86,97-3, 100, 184, 194
interlingua1 transfer 64, 71
rate 102-3
secondary 103
shared 84, 86
intelligence, artificial 172
intention, 211-3
communicative 128,212
informative 24, 34, 40, 55, 87, 102-3, 106-9, 189-90, 199,213,215
recoverabi1ity 212
interference, source language 82, 122, 184
interpretation
comparison 233
consecutive 5
as context-dependent 44, 46, 103, 107
and contextual effects 32-5
definition 24
intended 32,40-2,45-6, 76, 92-3, 96, 99, 101, 107, 110, 128, 135,
146,155-6,170-7,181,195-7,225-6,233
literal 41, 72
and propositional form 43
simultaneous 5, 122-3, 199,213
see also translation; use of language, interpretive
intonation 15,40, 135, 145,229
intuition
expert 9, 19
of translator 118, 142
invariance 125, 137
irony 111, 145-6, 173
irrelevance, perceived 96-7

Japanese 78, 139-42, 161

Kabatek, A. 195
260 Translation and Relevance

Kade, O. 10, 69
Kasstihlke, R. 185
Kelly, L.G. 1,62, 128
Kingdom of God, dynamic equivalence translations 185-6
Kloepfer, W. 1, 125
Knight, Max 112,116-8
Koller, W. 1, 10,69
Krings, H. 4-5, 18-9,22
Kui3maul, P. 8

Lakatos,1. 21, 202

language, figurative 15,50,82-3,88-9, 175-7
Lann, Evgeny 131
Larose, R. 13
Larson, M.L. 69, 71-2, 84-9, 93, 95, 175
Lefevere, A. 136
Lehmann, D. 1
Levy, J.
on compensation 195n.3
on decision-making 119, 191n.5
on equivalence 10, 125, 144
on hierarchy offunctions 14-15, 16, 113-5, 117
on lexical units 159
on meaning of original 111-2, 117, 119
on onomatopoeia 150-1, 160
on poetry 163-4
on style 8
on theoretical literature 1
on translation of weights and measures 126
lexicon, mental 159
linguistics, modem 69
literature, theoretical 1-2, 47-8, 125
logic, in relevance theory 36-7
Longfellow, H. W. 69

machine translation 5, 172

Index 261

Ma1mkj<er, K. 224, 226, 235

manifestness, definition 27
manual, instruction 216-21
Massaro, D.W. 19
meanmg
'bonus' 74-5, 90
change 184
and communicative clues 138-9
connotative 161-2
correctness 70
emotive 72, 86
expressed 26, 83,206,225
figurative 82-3, 88-9, 175
in formulaic expression 155
in idiomatic approach 85-6,99-100
intended see interpretation
of original 71-72,83-5, 100-3, 123, 124, 127, 186,225-6
semantic 151, 158-60
surface 74-5, 77, 92
truth-conditional theory 25-6, 36, 154-5
measurement, translation 126
memory 27, 110, 121-2, 157-9, 172
mental states, existence 22
message
definition 72, 79
and form 183
and interpretive use 99, 101, 105
of original 69-72, 79-80, 83-5, 92, 95, 97
metaphor
and implicit information 77,88-90, 121, 158
dead 158
methodology vii, 3, 9, 17,21
changes 21-2
see also research programme
mind
and relevance theory, 21, 25-6, 30, 36, 39, 141, 144-5, 162
and translation 205, 234
minimax strategy 8
miscommunication 179, 181, 197
262 Translation and Relevance

misinterpretation 76-7, 94-5,111,172,181,185,190,222

mistrans1ations 154, 184
Morgenstern, C. 111-3, 116, 118-9, 129, 137
Mossop, B. 132, 134, 197
multi-disciplinary 3-4, 19,22,207
Muschard, J. 224

names, and transcription 150-1

Namy, C. 122-3
naturalness
oflanguage use 71
of translation 47,95-6, 107, 122, 184
N avarro-Errasti, M.P. 224
Newmark, P. 9, 69, 121, 128-9, 150, 179, 199
Nida, E.A. 69,183,184
Nida, E. A. and Taber, C. 10, 65, 69-71, 79-80, 82, 84
notes, explanatory 80
notices, translation 158

o
O'Flaherty, W.D. 191-2
onomatopoeia 150, 159-60
oral translation 122-3
original
differences from 48-50
and interpretive use 99, 105-6, 123
message 69-72, 79-80, 233
and secondary communication 76-98
translator's interpretation 172,233
ostensive 212 (see also communication, inferential)
over-specification 11-12, 19

parallelism 149
paraphrase 5, 128, 152-3, 159,225
Index 263

parody 169
Payne, D. 195
Plummer, A. 147
poetry 89-90,111-8,119,123125,129,131,139-42,145,161,162-6,174,
218
Chinese vs German 188-90
phonemic translation 136-7
Sanskrit 191-2
postscript ix, 202
pragmatics 16, 125,207
precision 131-5,164
principle of relevance (see relevance, principle)
principles, translation 193, 195, 198-9, 224
contradictory 126-7
origins 119-26
process perspective 5, 18-19,21,205
product perspective 5, 10, 18,21
profile, textual 12-15
properties, 16, 18,20-1, 133,233
linguistic 25,133,169-71,177-9,191,194,198
logica136
phonetic 150-1
semantic 136, 138, 140, 143, 156
sound-based poetic 162-6
syntactic 144-50, 157
proverbs 15, 158-9
province oflanguage user 12
pseudoscientific 202
pseudo-translation (see Toury)
pun 113-5, 191
purpose, translation 16-17, 50,208
see also skopos

quality assessment 47, 154

Quine, W.V. 213
quotation
direct 132-6, 168, 169-71, 173-4, 178,232,236
264 Translation and Relevance

indirect 132-3, 236

interlingual see translation, as quotation

reader, interests 11, 69, 130 (see also expectations)

Reboul, A. and Moeschler, I. 207
redundancy 2, 15
ReiJ3, K. 8, 125
ReiJ3, K. and Venneer, H.I. 14, 16-7,22-3, 101, 199,220
relations
semantic 164
syntactic 163-4
relevance
in artificial intelligence 172
as comparative 33-4
as context-dependent 33, 44, 107, 199
definition 31, 31
optimization 22, 32-3, 106,225,227,237
and role of translator 67
ways of achieving 210-11, 221, 223 (see also use oflanguage)
relevance, principle 32-3, 40-2,81,83,90,96-7,102-3,106-7,110,118,
120-2,197-9,200,222,235
relevance theory vii-viii, 21, 24-46, 198,206,210-1
and cognition 21-2, 25, 27 (see also mind)
descriptive/interpretive use 35-46, 58-68, 99-104, 198
and direct translation 170-3
as explanatory 21-2,203,235,237
and message of original 72, 81
see also context, translation practice
reliability 181-2
rendering, free 177, 184, 189,200,202
repetition 147-8
representation and computation 144
representations
descriptive use 36, 39
interpretive use 39
mental 25-6, 36-9
and meta-representations 234
Index 265

semantic 25-6, 42, 134, 136-43, 151, 156, 159, 162-3

research programme (paradigm) for translation 17-21,203-4
communication-oriented 203
competence-oriented (CORT) 205-6, 236
culture-oriented 235
descriptive-classificatory 17, 203-5
Descriptive Translation Studies 203, 213, 223, 225
explanatory power 21-2,206,235,237
input-output 204, 206-8, 217, 224
philological 203
predictive power 206,226,237
prescriptive 203
shift 17, 203
resemblance
extent 191-2
linguistic 178, 191, 194, 198
semantic 136-7, 143, 151
structural 178
resemblance, interpretive 24,36, 105-6, 111, 118, 134,210,214,217,
219,225,231-2
expected/actual 184, 189, 229
intended 189-90, 229-30
between propositional forms 36-8
between thoughts and utterances 39-41
between utterances 41-46
and authenticity 228
and direct quotations 169, 171, 178
and direct translation 170-1, 177, 189-90, 194-6,228
restricted 166
responsibility
of communicator 91
of translator 190,199,231
rhyme 15,70,112-3,115,116-7,136-7,144,163-4,166
rhythm 15, 70, 136, 145, 163, 164, 166
Ruckert, F. 189
rules 119-21, 125-6, 198,224,226-8
contradictory 124-1,227
deductive 141-2
266 Translation and Relevance

s
Sa1kie, R. 224
Sanskrit, translation 191-2
Savory,T. 126-7, 130-1, 137, 146-7, 149, 163, 178,227
Sch1eiermacher, F. 174
Schulte, R. 3
sCIence
translation as 2-5,8-9, 17-20
translation not seen as 9
Scott-Moncrieff, 108, 111, 118
semantics, decompositiona1 view 80
sentence structure 144-6, 149
Shakespeare, William 58, 161, 164-5, 174
Sheridan, R.B. 151
Silt'e language 92, 135, 170
Sim R.I. and Gutt, E.-A. 223
simile 87-9, 91, 157, 175
skopos 16, 17,204,218,220
Snell, B.M. and Crampton, P. 5, 55, 63-4
Snell-Hornby, M. 20, 199
speech, direct/indirect 132-3
Sperber, D. vii
Sperber, D. and Wilson, D.
on communication 21-2, 25, 34-5, 66, 136
on context 26
on contextual effect 31, 147-8
on direct quotation 169
on faithfulness 106
on figurative language 88-9
on intention 190,212-3
on interpretive use 210
on irony 145
on logical and encyclopaedic entries 141-2
on meaning 83
on relevance theory vii-ix, 21, 23-4, 32
on representation 138, 178
on semantic properties 162
on stimulus 133
Index 267

on stress 134
on translation 105
see also Wilson, D. and Sperber, D.
Staiger, E. 165-6
Stark, D. 224
Steiner, G. 1,9, 143, 157
Steinhauer, H. 152
Stendhal 107, 176
stimulus
definition 23-4, 133
and interpretation 169, 179
reproduction 159
Von Strauss, Victor 188-9
stress, focal 134-5, 170
structure
surface/deep 72, 85-6
syntactic 122, 164, 177
style
of original 69-70, 126, 13 0-1, 134-6
in translation 8, 47,126,131,136,161,167
summary 105, 129
Svejcer, A.D. 10
symbols, translation 125, 126
synonyms, differences in meaning 162

technology, translation as 4
testing, performance 63
text
genre (see typology, text)
typology (see typology, text)
as unit of analysis 207
think-aloud-protocol 234
thoulyou distinction 108-9, 118
thought, as mental representation 24-5,36,41,58, 199
time, and relevance 123
Tirkkonen-Condit, S. 224, 231, 235
tourist brochure 48-54, 216-221
268 Translation and Relevance

Toury, G. 6-7,203,213,223
pseudo-translation 215
source-text postulate 215
transcription 150-1, 160
translating, ways of (see translation, approach)
translation
account202-4,211,214,231,235
approach 202, 203
as communication act 207,211,213,228,235-6,237
and complementary materials 187, 196-7, 231, 238
covert 47-68,215-21,236
and culture 47-8
definition 47
cu1tura180
definition 5-7, 119, 135-6,209,236
direct 136, 156, 158-9, 166-7, 168, 169-77, 179-80, 196,200,228,232
idiomatic approach 71,83-98, 100, 182, 193-4,202
indirect 136, 171-3, 177, 179-81, 186, 194,200
interlinear 144, 236
linguistic 79-80, 82-3, 203
limitations 228-9
1iteral131, 178, 182, 189, 191,200,202,232
literary 107-11, 130-2, 144, 188-90, 199,207
by machine 172
as multidisciplinary 3-4, 19,21
and non-translation 209-11,215,219
notions 207, 209, 222-3
oral (see interpretation, simultaneous)
overt 48
phonemic 178, 199,236
popular views 93
process 2, 3, 5, 7, 8,9, 18-9,21,217-8,232-4
purpose 206, 208 (see also skopos)
as quotation (re-expression) 130-5, 171,205-6,209-11,214,215,223,
229,231,235-6
restricted 128-9
asshortcut59,61,64,219
successful (see communication, successful)
unified account 168, 171-99
Index 269

unique nature 171,208,218,236

see also art; equivalence, dynamic; interpretation; science; technology
translation theory/theories vii, 1-3,5-9, 11-12,68, 198-200,206,208,218
critique 143, 235
domain 5-7, 21-2, 204, 220, 235
history 236
linguistic 203
and message oforiginal72-75, 79, 99,101-3,104,105
single-text based 12
see also action theory, relevance theory
translation practice, 5, 143,208,223-228
application of relevance theory 202,224,227,237
translator
interpretation of original 172, 225
responsibility 190-7, 199, 231
role 66-7, 69,186-7,199
transliteration 5
tropes 157-8
Von Tscharner, E.H. 188-9
Turner, N. 147-8
typology
function 206
text 18,20,128,206-7,221
translation 18,207,211

u
unnaturalness 122
usage, frequency 121-2, 148, 157
use of language, descriptive 210
interlingua1 56-68, 127, 190, 192, 198
intra1ingua1 61
and representations 36, 39
use of language, interpretive 210
and direct translation 168, 169-73, 178,232
interlingua158, 64, 105-29, 132-6, 168, 190-2, 198,211
intra1ingua1 61, 132
and message of original 99
and representations 38-9
270 Translation and Relevance

utterance
description 162-3
and interpretive resemblance 24, 39-46, 200, 210

value
communicative 134
conceptual 160
scientific 202 (see also pseudoscientific)
stylistic 160-1
value judgements 13-14,20
values, literary 112-5
Vemay, H. 119
verse (see poetry)
Vinay 125
vousltu distinction 107

w
De Waard, J. and Nida, E.A. 4, 125
Weber, D. 224
Weizman, E. and Blum-Kulka, S. 222
Wendland, E. R. 94, 99
Wilson, D. and Sperber, D.
on direct quotations 133, 169
on faithfulness 41, 106
on relevance theory 23
on resemblance 36-7, 45, 190
see also Sperber, D. and Wilson, D.
Wilson, Deirdre vii, 42, 155, 158,234
Wi1ss, W. 1,3,4,5,8-9,11-12,18,69,174
Wonderly, W.L. 64-5
word order 144, 146-7, 177
word play, in translation 111-12, 137, 191
Wyatt, Sir Thomas 123-4

Yuasa, N. 139-40, 142-3, 161-2, 174-6

Index 271

z
Zukofsky, C. and L. 136, 178, 199

Index compiled by Meg Davies (Society ofIndexers)

Extended and revised for 2 nd edition by E. -A. Gutt