Language Processing in Discourse

What do we choose to begin a sentence? How do we highlight the most
important information in a sentence?

This book argues that language systems determine language use to a

greater extent than is generally assumed. The author demonstrates how the
typological characteristics of a language determine even the most general
aspects of our stylistic preferences.

Through extensive analysis of examples in German and English, the

author demonstrates how analogous options of sentence structure must be
surrendered in order to achieve felicitous translations. Two major aspects
that determine the appropriateness of language use are examined: language
processing and discourse-dependency.

The most important area where the typological characteristics of a

language interact with language processing, Doherty argues, is the area of
information structure, that is, the way in which thoughts are packaged
into sentence structures to express informational progress. Investigating
examples such as the different verb positions of German and English, she
shows how such variances have far-reaching repercussions in terms of word
order, case frame and structural explicitness.

Essential reading for translation scholars and linguists involved in the

comparative study of English and German, this book will also be of interest
to scholars of psycholinguistics and cognitive science, as well as translators
and linguists more generally.

Monika Doherty began her career with the study of English, Russian and
General Linguistics in Berlin (PhD in 1969, Professor in 1981). She
concentrated on semantics, later pragmatics, and translation. In 1990 she
was appointed Professor for Translation Theory. Two major publications
contributed to her success with a wider readership: Übersetzen (2nd edn
1997) and Das grammatische Varieté (4th edn 1998).

Routledge studies in Germanic linguistics
Series editors: Ekkehard König and Johan van der Auwera

1. Negative Contexts
Collocation, polarity and multiple negation
Ton van der Wouden

2. When-clauses and Temporal Structure
Renaat Declerck

3. On the Meaning of Topic and Focus
The 59th Street Bridge accent
Daniel Büring

4. Aspectual Grammar and Past-time Reference
Laura A. Michaelis

5. The Grammar of Irish English
Language in Hibernian-style
Markku Filppula

6. Intensifiers in English and German
Peter Siemund

7. Stretched Verb Constructions in English
David Allerton

8. Negation in Non-standard British English
Gaps, regularizations and asymmetries
Lieselotte Anderwald

9. Language Processing in Discourse
A key to felicitous translation
Monika Doherty

London and New York

Language Processing in
Discourse

A key to felicitous translation

Monika Doherty

First published 2002
by Routledge
11 New Fetter Lane, London EC4P 4EE

Simultaneously published in the USA and Canada
by Routledge
29 West 35th Street, New York, NY 10001

Routledge is an imprint of the Taylor & Francis Group

All rights reserved. No part of this book may be reprinted or reproduced or
utilized 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.

British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library

Library of Congress Cataloging in Publication Data
A catalog record for this book has been requested

ISBN 0-415-28189–X

This edition published in the Taylor & Francis e-Library, 2004.

ISBN 0-203-21693-8 Master e-book ISBN

ISBN 0-203-27300-1 (Adobe eReader Format)

(Print Edition)

Contents

Preface

Acknowledgements

xiii

Abbreviations

xiv

Setting the scene

Ascending a tightrope 2
Discarding dictionary problems 3
Taking sides 6
The style of the original and a Translation Maxim 8
Optimal relevance and the importance of a common basis 9
Control paraphrases 11
Linguistic knowledge 14
Sentence processing 16
Considering selection restrictions and wrapping up 18

Questions of order

Parametrized processing conditions 22
Parametrized beginnings 23
Information structure: the point of departure 24
Reordering information structures 26
The most important element 28
Structural focus 30
Basic verb positions 31
Context sensitivity 32
Hierarchies 33
Focus projection 35
Focusing elements 37

Complex sentences

A double-storeyed focus 40
Scrambling 41

vi Contents

Focus spacing 42
Dummies 44
A difference in initial positions 46
Another type of focus spacing 47
Generalizing the findings 49
Cross-clausal 50
Degrees of accessibility 52
The different ways to perfection 53
Reviewing and resetting the stage 55

In favour of primary relations

Alternative perspectives 58
Placing a modifier 61
The parsing strategy of Minimal Attachment 63
After the subject or before it 65
Competing for attachment 67
Different perspectives 68
Contrasting the agent 70
A covert shift 73
The morphological advantage 75
Lexical repercussions 76
Revisiting the beginning 78

Structural weight

Anaphoric ambiguities 82
The missing gender 83
Tucked away 85
Decomposing the unpredictable 87
Grammatical dummies 88
An extra clause for the second focus 90
Adverbial expansion 92
Too short 95
Textual relevance 97
Making it visible 99
A comment 100

Grammaticalized clues

103

It-clefts 104
Focusing a clausal subject 106
Pseudo-clefts 107
Presupposing and contrasting 108
Dispensable information 109
The great variety 112

Contents

vii

Processing aid or processing burden 115
Ambiguities 116
The domino effect 118

Shifting boundaries

121

Linking 122
Linking by resumption 123
Separating 125
Felicitous contrasts 128
Shifting sentence borders 130
Extrasentential reordering 133

Relativizing optimality

137

A preamble 137
No liberalism 139
Quite normal 140
Statuesque and dense 143
Overspecifying 146
Normalizing the extraordinary 147
Artistic redundancies 150
No choice 151
Leaving the tightrope for the trapeze 153
Pre-programmed losses 155
Flouting the Maxim 158

Reviewing the scene

160

Parametrized focus structures and theoretical implications 160

Glossary: technical terms seen through the Keyhole

165

Notes

180

Bibliography

183

Index

189

Preface

This book addresses students and teachers of English and German, linguists,
translators and translation scholars. Some preliminary remarks may be in
place to help unify expectations. I have written the book in the belief that
language systems determine language use to a greater extent than is generally
assumed. It is not just the way in which the words of a language and its
grammatical properties commit us to categorize the world in language-
specific ways, and not just how each register requires us to follow special
conventions of language use – it is the fact that the typological characteristics
of a language have an impact on the way in which we prefer to encode our
thoughts. That is, the typological characteristics of a language determine the
most general aspects of our stylistic preferences.

In my view, this involves two major aspects of language use: language

processing on the one side and discourse-dependency on the other – hence
the first part of the title, Language Processing in Discourse. The most
important area where the typological characteristics of a language interact
with language processing in discourse is the area of information structure,
that is, the way in which thoughts are packaged into sentence structures to
express informational progress in discourse. Thus, this is a book about
information structure and deals with questions like: What do we choose as
the beginning of a sentence? In which way do we highlight the most impor-
tant information of a sentence? What are the lexical and grammatical
options that can secure optimal transfer of a thought in its discursive con-
text? I consider these questions fundamental to a felicitous use of language,
whatever additional aspects may have to be studied for non-informative
functions of language.

If we want to find answers to these questions we have to study language

use in discourse, assess its appropriateness and compare the results with
similar data from other languages. We could do this statistically with
parallel texts or with translations. But the texts may not be really parallel
and the translations may be distorted by mistakes. Although the individual
mistakes might eventually be filtered out by the sheer quantity of statistical
data, assessing the discursive appropriateness of a sentence requires us to
analyse individual cases anyway – which we can do straightaway without
recourse to statistical data.

Preface

Analysing individual cases, we may, however, end up with highly

unreliable and subjective judgements, losing any chance of predictive
generalizations. But my experience with many generations of students is that
the individual judgement can be stabilized and become more ‘objective’ (that
is, representative of majorities) if sentences are assessed in comparison to
other, systematically varied paraphrases within the same context. For
typological purposes, when comparing original and translated sentences, the
direction of such a systematic variation is predetermined by the differences
between the languages involved.

The preferred paraphrases will occasionally differ widely between target

and source language. But wherever different preferences recur we can expect
systematic reasons and look for the typological characteristics that might
explain the observational data. And this is what the book will do: it will
compare individual examples of original sentences in German and various
English translations, assess their discursive appropriateness and suggest
typological explanations for the language-specific preferences observed.

In view of the different backgrounds of the addressees, the procedure may

be considered problematic for various reasons. Firstly, it presupposes some
knowledge of linguistic concepts and hypotheses. As I know from my
experience with audiences of students and translators, explicit linguistic
knowledge – even of the school-grammatical type – is not very widespread
and does not meet with much interest, either. To relieve the main body of the
book from all the linguistic knowledge its readers should share, I have
therefore briefly described the linguistic terms used in an alphabetical
glossary at the end (some elementary syntactic concepts are revised in the
first chapter).

Although I am acutely aware of the fact that there are many different and

often contradictory interpretations of the linguistic concepts, I have decided
against discussing alternative views in order to simplify the main argument
of the book. I am certain that the main ideas are also of interest to readers
who do not want to specialize in linguistics, but may be willing to follow a
linguistic argument which is not burdened with too many technical details.
Thus, I have written the book in a non-technical, common-sense format,
which the more academically oriented reader is asked to excuse in the
interest of a wider readership. (A few suggestions for further reading will be
given in the notes at the end for the more linguistically interested reader.)

In addressing a wider readership, one runs the risk of simplifying

theoretical issues to such an extent that they may no longer represent the
state-of-the-art within the specific theories they belong to. The danger of
trespassing over the limits of accepted disciplines is enhanced by the fact that
none of the theories accessed has yet developed a sufficiently rich repertoire
to explain all the observations we can make in studying translations. But the
translational perspective can also sharpen understanding of the linguistic
concepts needed and thus help to develop them further. (For example, some
of my assumptions about the information structure of complex sentences, or

Preface

about the greater textual relevance of certain foci, are theoretical extra-
polations which need much more research work to become properly
integrated into accepted frameworks.)

Second, from the point of view of text linguistics and translation studies, I

am aware of another peculiarity of the book. Concentrating on information
structure, the presentation proceeds from sentences as the linguistic basis of
information packaging and considers only those aspects of a discursive
context which are relevant for the information structure of the example
under discussion. This is a highly restrictive approach, using, so to speak, the
microscope as an instrument for questions of discourse analysis and trans-
lations; as it is in direct opposition to the multi-functional perspective of text
linguistics and the culture-dominated approach of modern translation
studies, it will require an extra amount of tolerance on the side of the text-
linguistic or translation-theoretic reader.

Third, for the basic typological claim of the book, a restriction to two

languages may look highly suspicious. However, German and English are
related but typologically alternative languages, and we can observe a
remarkable amount of restructuring in felicitous translations between the
two languages. It remains to be seen in which way other languages promote
similar or different patterns of restructuring. (Even a very superficial look at
closely related languages, as for example English and Norwegian, German
and Dutch, or widely divergent languages, like English and Japanese,
German and Russian shows that the simple connection between alternatively
distributed information and basic grammatical parameters characterizing
German and English has to be enriched by further distinctions to explain the
different preferences we can observe in translations between these other
languages.)

With respect to German and English, the examples discussed in the book

are seen as prototypical cases of a great number of translations with similar
information structures. The research projects preceding and accompanying
the book revealed that the most frequently recurring patterns of restruc-
turing between English and German were of the type illustrated in the book.
The restructuring patterns are related to the linguistic themes of word order,
case frame, explicitness and sentence boundaries, which will be looked at
from the perspective of information structure and language processing in
Chapters 2–7.

The first chapter of the book will set the scene for felicitous translations.

And this may be seen as yet another problematic aspect of the presentation:
felicitous translation or felicitous use of language in general is not exhausted
by the informative function of language. Except for the last chapter, where
the discussion covers examples from a judicial text type and several literary
ones, the book deals with one default Maxim of Translation, one text type,
one register and one pair of languages. This is, as every translator knows, not
even the tip of the iceberg.

Setting all functional variables at a default value, I have tried to

xii

Preface

concentrate on one, basic aspect of language use. There are many transla-
tional purposes and many text types dominated by other aspects of language
use, and with other aspects in the foreground, considerations of information
structure may be of little interest: rhyme and rhythm, metaphor and puns,
stereotyped and stylistic figures, cultural gaps and mismatches involve
translation problems which are only indirectly related to questions of infor-
mation structure. Yet even here, where we may face the greatest limitations
of surface equivalence, the choice of substitutes will to some extent be guided
by the informational relevance of the problematic elements. (One of the most
urgent tasks for future studies in this direction will be to identify the special
conditions for spontaneous oral language use, where the prosodic options
may dispense with the focus expectations typical of written language
altogether.) Nevertheless, although it is also necessary to study the special
conditions of ‘marked’ text types, registers or translational purposes, it
seems reasonable to sort out the basic conditions first.

The main part of this book was written during my ‘sabbatical’ in the winter
of 1998/9. It is the result of several years of research work with colleagues
and students on questions of position and explicitness in translations
between German and English. Most examples in the book were taken from
selected chapters in a popular scientific German book. The original and its
translation form part of an electronic corpus of German/English texts
gathered during three successive research projects (on parametrized
perspectives, adverbial clauses and cleft sentences). I would like to thank
Gideon Toury and Peter Eisenberg for their critical and encouraging
comments on the manuscript; Phyllis Anderson for her painstaking work
with the corpus and the bibliography, tracing the discussions of past and
current research sessions; Michael Davies for his prompt and subtle control
of the English wording of the book, including the ranking of the control
paraphrases of the English translations; Sigrid Venuß and Thomas Schulz for
their precise and patient assistance with the orthography and layout of the
various components of the book; Birgit Ahlemeyer, John Bergeron, Inga
Kohlhof and all my students for their stimulating arguments on a great
variety of aspects directly or indirectly related to the topic of the book.

Acknowledgements

xiii

Abbreviations

The following abbreviations are used in the text to refer to sources:
B = T. Bernhard; E = H. M. Enzensberger; EW = M. Eigen and R. Winkler.

xiv

Setting the scene

As the object of translation includes anything that can be expressed as well as
anything limiting the power of expression, a key to translation could be a key
to an infinite number of worlds, real ones and virtual ones. It would take all
the disciplines of philosophy and science to deal with such a topic. But the
object of translation is not so much what may be expressed and hence
translated, it is the way in which something is expressed in one language as
opposed to another. Or, more precisely, it is the way in which the differences
in the expressive power of two languages may be overcome.

There are basically two perspectives we could choose from to study

translation: there is the historical perspective, focusing on the paths already
taken in the effort to bridge the differences between languages. If we say,
‘Translations are facts of target cultures’ (Toury 1995: 29), which can only
be studied if they are properly contextualized, we opt for the historical
perspective. But we can also view translation as a set of possible correspond-
ences between languages, and each translation as a contextualized instance
of these possibilities. Such a basically generative, but at the same time strictly
context-oriented, line will be followed by the Key to Felicitous Translation
(henceforward referred to as ‘the Key’), which bases theoretical claims on
detailed empirical evidence as seen through the critical eyes of a competent
translator.

However, even if we restrict the concept of language to natural language

and ignore all non-linguistic means of expression, the set of possible
correspondences between languages is endless, since the set of linguistic
structures that can be formed in each language is endless. Thus the Key will
focus on a subset of possible correspondences leading to translations which
can be considered felicitous or ‘optimal’ in terms of language processing in
discourse. What optimal conditions for language processing in discourse are
and how their clarification can provide a deeper insight into translation
relations – possible or contextually instantiated ones – will be demonstrated
in this book.

Setting the scene

Ascending a tightrope

The ability to translate from one language into another is one of the most
miraculous abilities we can develop. Everybody who has not acquired this
ability thinks that it is enough to know two languages to do it. This is
definitely a good basis, but it is not sufficient.

Knowing a language can, of course, mean different things. If we take it to

mean what we know in our own language, we could say that we are, ideally,
able to understand anybody who speaks or writes English, and that we can
express anything we want to express – as long as it can be expressed in
English at all. Although this is a euphemistic understanding of our common
abilities, in which not everybody participates in the same way, the basic
capacity we all share involves so many highly sophisticated aspects that we
can only consider it as ‘one of the wonders of the natural world’. Steven
Pinker, from whose famous book The Language Instinct this is a quotation,
has shown all of us what an enormous intellectual achievement knowing a
language is.

But even if someone has a perfect command of two languages, as good as

any bilingual speaker, this does not equate to translational competence itself.
Translators have developed an additional capacity, over and above the
‘normal’ capacities of language use: in the languages they work in, trans-
lators are, ideally, not only able to identify and produce any expression of
these languages, but also able to identify and produce any pair of corres-
ponding expressions that could be formed between these languages.

The feat they perform when applying this capacity is not unlike walking

on a tightrope – as everybody knows who works as a translator or interpre-
ter. The difficulties often begin before the acrobatic part of the translation
act sets in. There is, first of all, the original you have to understand, which is,
more often than not, quite a challenge – not so much on the linguistic side
itself, but due to its context-dependency. You can never be sure about the
meaning of linguistic elements unless you know the context they have been
used in. That this is by no means a trivial matter will become clear when we
analyse our examples in regard to their ‘textual relevance’ (Chapter 5).

It is the built-in vagueness and ambiguity of linguistic elements that makes

language adaptable to the infinite number of situations we use it in, and the
unlimited number of elements we use it for. Part of the enormous variability
of individual linguistic elements is their potential for being combined, but
this, too, can be subsumed under the heading of contextual dependency. Seen
thus, ‘context’ has many different aspects: the immediate context of an
expression is the phrase it is used in, the clause it belongs to, and the sentence
containing the clause; and then, less immediately, any of the preceding or
subsequent sentences that are relevant for the interpretation of the linguistic
element we started from. This linguistically encoded context of an element is
sometimes referred to as ‘co-text’, to distinguish it from the ‘extralinguistic’
context relevant for the interpretation of a linguistic element.

The extralinguistic context of a linguistic expression comprises our mental

Setting the scene

model of the situation, as well as all the aspects of world knowledge the
linguistic expression is expected to activate in our minds. The more we know
about the things talked about, the better we can cope with the task of com-
prehending the original. Yet there are so many limitations to our knowledge
of the world that we have to content ourselves in many cases with the
co-textually presented information relevant for the interpretation of an
expression, enriching it by whatever thematically relevant information we
can obtain.

However formidable a challenge it may be to comprehend an original, it

does not require any other faculties than those we need for the monolingual
use of language. The walk on the tightrope begins only when we want to
transfer into another language what we have read off from the original text.
One of the most important statements characterizing these problems was
made by Roman Jakobson, who said that languages do not so much differ in
what they can express but in what they must express. To this we might add:
and in what they ‘prefer’ to express. We can be sure that any original and any
translation contain many linguistic expressions which are not needed for the
message itself, but for the special requirements of the languages involved.

If we know something about these requirements, we can identify the

linguistic elements of the original that serve this particular aspect, and may
not have to be transferred into the other language. Certain aspects of the
original are at our disposal – unless the translation is to reveal the peculiarity
of the language of the original, which is not what translations normally aim
for. On the other hand, we are free to add linguistic elements to the transla-
tion for which there are no corresponding elements in the original with the
sole aim of meeting some special requirements of the target language.

The differences between what languages prefer to express will go un-

noticed as long as the linguistic expressions of two languages are not related
to each other by the type of relations characteristic of translations. Actually,
there are many differences in the use of languages which are of a less obvious
nature, revealing themselves only to the eyes of the competent translator.
Translation relations teach us more about the different use of two languages
than we can learn from their systematic comparison.

The linguistic discipline that looks at differences between language

systems is contrastive linguistics, and to some extent, translation relations
have always been included in its empirical basis. One cannot describe a
linguistic element exhaustively unless one looks at its contextual properties.
But the discipline that must look into the contextual properties systematic-
ally due to the very nature of its purpose is translation studies: translation
relations are always context dependent.

Discarding dictionary problems

Most of what we read has been translated, and as most people’s ideas
proceed from other people’s ideas, including translated ones, the effect of

Setting the scene

translation on our view of the world can hardly be overestimated. Against
this background, it is quite amazing how little attention the topic of
translation itself attracts. Beyond the realm of translators and the few
studying and teaching translation, there seems hardly anybody who is aware
of or cares about any of the problems involved in translating.

Sometimes, however, we cannot help stumbling over some of the short-

comings of incompetent translations, as when we find ‘cancer soup’ on a
menu of a Prague restaurant or this passage from a brochure of a car rental
firm in Tokyo:

When passenger of foot heave in sight, tootle the horn. Trumpet him
melodiously at first, but if he still obstacles your passage then tootle him
with vigor.

(Davies 1989: 41)

But even in these cases, everybody would just think of the problems
somebody must have had with English and not of the translation problems
arising from mismatches between Czech or Japanese and English.

True, not everything rendered in another language needs to be a trans-

lation. Inferences of our own language can occur in the other language even
if there is no original serving as an inferential basis. What matters is that our
thoughts can hardly be thought without the medium of a language, whether
our own language or another one.

In any case, whether misuses like the above originate in translations or

not, they do not belong to the core of translation problems. The real
problems of translation will in most cases remain invisible to the average
reader. They are only visible to those readers who can compare the original
and the translation and are thus able to discover differences between their
content and style. There are relatively few who can do this. Most readers
have to accept what they get for what it claims to be: Leibniz, Goethe,
Nietzsche, Kafka, Brecht, Freud, Einstein or whatever literary, political,
scientific or technical writer, etc., the translation may be attributed to. What
problems the translators faced, the ways in which they solved them, what
congenial ideas they created, what blunders they made – all this remains
invisible as long as the translation is not held against the original. But if you
could read and understand the original, you would normally have little
reason to read a translation of it, unless, for example, you specialize in trans-
lation or translation studies. No wonder translators and their problems meet
with so little interest despite their dominant role for the cultures of this world.

As everybody knows who has tried it seriously, it is one thing to express

oneself in a language and another thing to express what has already been
said in a different language. While thinking a thought and expressing it are
very much like two sides of a coin, taking shape simultaneously, translating
equates to exchanging existing coins of different currencies. But while
exchange rates specify transfer rules clearly, the unlimited combinatorial

Setting the scene

possibilities of languages open up an infinite number of potential corres-
pondences. Thus, even the most gifted translator is bound to encounter
problems that cannot be solved automatically.

And there is an additional difficulty. It is by no means guaranteed that a

translation really says the same as the original. In fact, if we take a very close
look at the original, we may even find that nothing is fully translatable if
only because the differences between the languages involved can never be
fully overcome. This book will provide ample opportunity to study this
question from various sides.

But the side from which the question of translatability has been

approached most often will be excluded right away as it relates to individual,
‘idiosyncratic’ differences, which do not lend themselves to generalizations.
Yet we cannot ignore the fact that it is the unique nature of such problems
that determines the main battlefield of translation. The differences which
everybody thinks of first are those between the words of languages.
Although dictionaries seem to suggest that there is always a word in the
other language for every word of our own language, the range of the mean-
ings of these words differs, so that one word corresponds to many different
ones in the other language, and some of the meanings may not have any
correspondences at all. However, differences like these are not insurmount-
able. Missing meanings can be supplied through syntactic composition. This
is the normal, the ‘default’ case, which we will rely on in the Key.

Missing meanings, however, can also be supplied by the creation of new

words, or the addition of new meanings to existing words – that is, by an
extension of the vocabulary, the lexical repertoire of a language. This is what
happens within our language anyway. As we all know, new words crop up
continuously as new situations arise and new objects come into being. In this
respect, too, languages constitute very powerful systems with a potential to
cope with any new aspect of our lives we might want to express. As different
languages absorb different aspects of lives other than our own, translation
involves the creative potential of anyone who translates an original referring
to such new aspects. Famous translators have therefore often been credited
for the impact they had on their language and its expressive potential.

As we know from the impressive compendium Translators Through

History, edited by Jean Delisle and Judith Woodsworth, famous translators
were always people with a religious, political, poetic or scientific mission,
who made use of translation as a means to increase their influence. They
were definitely not the anonymous type of translators struggling to make a
living from translations. But it is also the anonymous translator, or, more
precisely, legions of translators, who by reducing the differences between the
languages they work with contribute to changes in these languages. This is a
fascinating topic for historical study, which is, however, altogether outside
the scope of this book. Instead, the Key will concentrate on the combina-
torial power of languages as they are and all the translational problems and
chances associated with it.

Setting the scene

Taking sides

Making use of new words or new uses of existing words, translators can fill
in missing words, so-called lexical gaps, in the language they translate into.
But in many cases they would rather content themselves with exploiting the
combinatorial possibilities of their target language and use syntactic means,
that is paraphrases, to compensate for missing lexical elements or gram-
matical differences.

To a certain extent, one can always express the same idea in various ways.

Although variations involve differences, the differences could be ignored
under certain conditions (as for example within a specific context). In fact,
paraphrases are the normal way to compensate for differences between
languages, whether lexical or grammatical, and the crucial question in
translation is, which paraphrases were chosen and for what reason, or which
paraphrases should be chosen to reach one’s translational goal. Unless we
have a special reason not to do so, we would normally want to overcome the
differences between the languages in an optimal way. But who can tell us
what the optimal way is? Can there be anything but a subjective answer to
this question?

Theoretically, there are two basic options the translator can choose from.

The differences in the languages involved in a translation force the translator
to take side either with the language of the original – the source language – or
with the language of the translation – the target language. Even if it is both
one and the other way in most translations, there will be an explicit or
implicit dominance of one side.

Taking side with the source language means retaining as many of the

linguistic aspects of the original as possible. In the extreme, this means a
literal translation, where nothing but the form of the words is exchanged,
never mind how many grammatical or stylistic rules of the target language
are violated.

This is an accepted procedure for the demonstration of the formal

properties of another language. Linguistic work on less-known languages
could not be conceived of without such literal translations. But even the
fiercest proponent of literal translations of religious or literary origin would
accept some changes in the interest of comprehensibility. Unless presented as
interlinear versions commented upon in terms of real translations, literal
translations are more often than not incomprehensible. For example:

There small laugh there big stop keep on.

tells us hardly any more than the original version in Dyirbal, an Australian
language from North Queensland:

anydya ban midi miyandaygu banggun bulgandu dyabilganinyu.

Setting the scene

It is only the translation getting closer to the target language which will give
us an idea of the meaning:

small (woman) wanted to laugh (but) the big one stopped (her).

(Cooreman 1988: 730f)

Obviously, if we want to understand what the original says, we must not
take side with the source language – at least, not exclusively.

To get at the meaning of the original, some of its formal aspects will have

to be given up. But if I give up some forms of the source language, which ones
should I give up and which ones retain? If I were to take comprehensibility as
the only guide, I might still end up with grammatically ill-formed sentences.
Thus, we could easily understand the close translation of:

Gestalt ist in unserem Denkorgan als Ganzes reflektiert.

(EW 1983a: 88)

*Form is in our brains as a whole reflected.

despite its grammatical defect. (The asterisk signals grammatical unaccept-
ability.) But it seems that, unless there was a special political, literary or
academic reason for ungrammatical expressions already given in the
original, grammatical acceptability is what everybody would expect of a
translation.

Whatever else we might want to alter in the translation of the German

original, the grammatically correct position of the verb is a requirement
nobody would dispute:

Form is reflected in our brains as a whole.

It goes without saying that grammatical acceptability means more than
choosing the correct position of a word. Just think of the many grammatical
categories for the expression of time and aspect, negation and emphasis,
modality and quantity, et cetera that are used differently in different
languages.

So far, we have taken side with the meaning of the original and con-

strained the corresponding expressions in the target language by their
grammaticality. But grammatical acceptability alone is not sufficient. After
all, there are mostly several paraphrases available to overcome the
grammatical differences between the source and target languages. Thus, for
example, we could also reorder the two extensions of the verb:

Form is reflected as a whole in our brains.

Setting the scene

or even reformulate the entire version as:

Our brains reflect form as a whole.

So here we are back to the crucial question from above: which paraphrase
would be chosen and for what reason? Which paraphrase would be optimal?
As the Key is meant to focus on felicitous translation, we will from now on
concentrate on the second question.

The style of the original and a Translation Maxim

As the differences between the above paraphrases are relatively subtle, we
will postpone their discussion somewhat and look at a more drastic example
to get a better idea of what else we have to take into account. In their
discussion of the perception of form, shape or Gestalt, from which the first
example was also taken, Eigen and Winkler say that:

Mittelwertbildung ist gleichbedeutend mit Informationsverlust.

(EW 1983a: 167)

This means roughly:

Forming an average means losing information.

But the sentence has been translated as:

Whenever we represent a system by an average we necessarily lose some
information about the total system.

(EW 1983b: 144)

We could translate these two clauses back into German as

Sobald wir ein System durch einen Mittelwert darstellen, verlieren wir
notwendigerweise etwas Information über das Gesamtsystem.

which seems to be equally far away from the original statement as the
English translation is from the first rough paraphrase. All these versions are
grammatically acceptable, however, and express the same idea. Yet they do
it in different ways and the differences between the original and the trans-
lation are obviously greater than what we would have expected. It seems that
the original says what it says in a way that is not reflected in the existing
translation.

Let me refer to how the original says something as the ‘style’ of the

original. If we want to retain the style of the original in addition to its
meaning, we must find a way to express the meaning of the original and,

Setting the scene

maybe, retain those forms of the original that constitute its style. But while
the style of the original is somehow constituted by the way in which the
forms of the source language have been used, retaining the original forms –
where this is possible – will not automatically reproduce the style of the
original. Similar forms may well have different values in the target language.
What we must look for in the target language are those forms that can
reconstruct the style of the original. We shall see whether this is possible at
all and, if so, to what extent.

Wanting to retain the style of the original is, however, just another aspect

of the overall decision to take side with the original, that is, it is part of one of
the two alternative translational norms we may opt for. We are free to
choose. We may opt for a free translation and adapt the original to whatever
social, political, ideological, artistic or scientific commitments we have or
consider adequate for the readers of the translation. As long as we do not
lose sight of the meaning of the original altogether, we would probably still
speak of a translation in many of these cases. But we could also opt for a
stricter norm and stay as close to the meaning and the style of the original as
is possible within the constraints of the target language.

It is this second way the Key will choose in its search for an optimal way to

overcome the differences between the languages, their systems and their use.
It will be considered the default case of a translation norm, against which
any of the more specific cases can be measured.

Optimal relevance and the importance of a common basis

The first part of the Translation Maxim is what the Key proposes as a
yardstick for equivalence: the closer a translation is to the meaning and
style of the original, the more equivalent original and translation are.

Equivalence is a highly complex concept, for meaning is made of many
different ingredients (a rough survey will be given shortly), just as style is
assembled from myriad different aspects.

The following chapters will successively demonstrate several of the more

general factors involved in the equivalence between original and translation.
The Maxim considers ‘licensed’ only those deviations from equivalence
which can be shown to pay tribute to the constraints of the target language.
This is a highly restrictive norm and presupposes a great degree of awareness
about the subtleties of such constraints.

Being aware of the differences between the constraints of target and

source language is a major precondition for measuring the quality of
translations relative to the Maxim. As the Key will show, this means being
aware of many highly intricate aspects of language (systems and use), but it
presumes that quality assessment is possible.

Staying close to the meaning and style of the original requires us to access

our knowledge about the ways in which source and target language are used.
To a certain extent, the way in which languages are used is universal.

Setting the scene

Needless to say, translations, representing a specific type of communication,
share some of the properties characteristic of communication in general.
Whenever we participate in communication, we intuitively adhere to some
basic principles or maxims of cooperation, which we expect anyone partici-
pating in the communication to follow, too. Now, of course, we are free to
follow maxims or to deviate from them, but deviations can only be registered
as such against the background of the norm.

Relevance theory says that the most general principle followed is the

Principle of Optimal Relevance. Relevance is measured against cognitive
gains and processing effort. Everyone participating in communication can be
expected to follow the goal of maximizing relevance, of securing ‘the
greatest possible cognitive gains for the least expenditure of effort’ (Carston
1988: 59). Cooperation between the participants in communication is
automatic in the sense that:

every ostensive act of communication ‘communicates the presumption
of its own optimal relevance’.

(Sperber and Wilson 1986: 158)

That is, with every act of communication, whether speaking or writing, you
tell the others, or you are told, that it is worth while listening or reading.

The fact that there are so many things communicated which are far from

relevant for a listener or reader does not restrict the general validity of the
principle since even failure to meet the principle is registered automatically
against the background of the principle.

What are cognitive gains – or, more general, cognitive effects – and what is

processing effort? Relevance theorists distinguish three types of cognitive
effects. All three of them relate to our knowledge or our beliefs, our
assumptions about the world. The assumptions we have could be confirmed,
extended or rejected by what we are told. There would be no gains if none of
the three effects came about. If there were no assumptions that the
information could confirm, extend or contradict, the act of communication
would be considered irrelevant.

The information would be considered as not optimally relevant if the

processing effort needed to arrive at the information communicated was
disproportionate to the cognitive gains made. But what is it that determines
processing effort? For one thing, it is clear that novelty of information will be
taken as a yardstick of processing effort. If what has been said is familiar
to you, you will require little effort to understand it. If you are not familiar
with most of what has been said, you will have considerable difficulty
understanding it. If you do not know anything at all of what has been said,
communication will fail altogether. No processing effort whatsoever could
compensate for that.

With a sentence like the following, the fact that we are familiar with some

of the words is of little help:

Setting the scene

Imprint a polymer with right-handed isomers, and left-handed isomers
won’t fit into the cavities.

(New Scientist, 13 September 1997: 36)

But in context we have been familiarized with most of the new concepts and
are able to process the sentence without difficulties. ‘Right- and left-handed’,
we are told, are mirror-image pairs of molecules which differ only as a right
hand differs from a left hand. And ‘imprinting’ is a technique used by
chemists to make polymers ‘remember’ the shape of a molecule by moulding
polymers around a target molecule and then washing it away, thus creating a
cavity only the target will fit. Knowing all this, we can identify the form of
the sentence as an imperative with the meaning of a conditional – ‘If you
imprint . . . won’t fit . . .’ – which tells us that the ‘memory’ of the polymer,
the cavity, is very precise when it ‘remembers’ the shape of a particular
molecule.

Even so, there might be many aspects of a message which we fail to

understand despite the effort we invest when processing its linguistic form.
This may be either because we do not have the special knowledge required
over and above the context (what about ‘polymers’ and ‘isomers’, for
example?) or simply because the message communicated is poorly formu-
lated and fails to meet the Principle of Optimal Relevance.

What it is to be poorly formulated, or well-formulated at that, is a topic in

its own right – in fact, it is the major topic of this book – and processing
effort is a key concept for this topic. But before we can spell this out, we have
to agree on a common empirical basis for the future line of reasoning.

Control paraphrases

As every processing effort is relativized to what we know when we start
processing a linguistically encoded piece of information, there cannot but be
great differences between the background knowledge each of us can make
use of. The differences will be partly compensated by the context, especially
co-text, of the information, and by all the relevant assumptions we happen
to share as members of the same language community. If we ignore the fact
that there are an unlimited number of subgroups in a language community
and concentrate on the ‘standard’ speaker of English who would read
the type of text exemplified by the Key and its examples of translations
between German and English, we may assume a great deal of common
knowledge regarding the content and form of a contextualized piece of
information.

Regarding linguistic form, shared knowledge comprises not only the

grammatical competence necessary to understand, for example, the follow-
ing set of paraphrases, but also the stylistic competence which tells us that,
under certain conditions, (f) is the paraphrase which meets the Principle of
Optimal Relevance best:

Setting the scene

(a) At the beginning of creation stands the emergence of form.
(b) At the beginning of creation there is/was/we have/see . . . the emergence

of form.

(c) The emergence of form stands at the beginning of creation.
(d) The emergence of form began at the beginning of creation.
(e) The emergence of form began with creation.
(f) Creation began with the emergence of form.

Comparing the paraphrases in context, we shall see that it is (f) that

secures the ‘greatest possible gains for the least expenditure of effort’. For a
first assessment, all we need to know is that this is the beginning of a popular
scientific text on form (Gestalt), and that our paraphrases are translation
versions of a German original. Let us ignore the role of the German original
for a moment and ask ourselves which of the six English versions we would
prefer at the beginning of a popular scientific text on form. Provided there
are no special requirements, I claim that the variants (e) and (f) are the
paraphrases most of us would prefer. (The minimized contextual conditions
do not allow us to distinguish between these two versions yet.) It will take
some time to spell out all the details contributing to this intuitive ranking.

If we take a look at the German original:

Am Anfang der Schöpfung steht die Gestaltbildung.

we can see that the set of the English paraphrases starts with the version:

(a) At the beginning of creation stands the emergence of form.

which is formally closest to the German original. Thus, we can suspect that
the differences between the various paraphrases reflect differences in the use
of English and German, which provide different conditions for optimal
processing. The detailed analysis of this and other examples will show that
this is indeed the case.

If we want to reach a better understanding of language-specific processing

conditions, we have to know something about the ways in which languages
can differ from each other. As a whole, this is an endless affair, but there are
some major aspects associated with word order, syntactic functions,
structural explicitness and discourse incrementality (sentence boundaries)
which can cover a wide variety of frequently recurring phenomena. We will
look at them one by one in the following chapters, preparing the ground in
the remaining part of this chapter by surveying some of the more general
linguistic and psycholinguistic aspects involved in the discussion.

Assuming that we take the final two paraphrases as the best possible

versions among those we have compared, does this really mean that they
are the optimal versions? What is the relationship between our intuitive

Setting the scene

preferences and the Principle of Optimal Relevance? What about cognitive
gains and processing effort in the individual paraphrase? As all paraphrases
express basically the same idea, they should yield the same cognitive gains.
The degree of novelty, too, seems to be the same in all paraphrases, in that we
are more or less familiar with all of the concepts used.

What differs from paraphrase to paraphrase, however, is the linguistic

structure encoding the message. And this means that we face different pro-
cessing conditions in each paraphrase. With a short, isolated sentence like
this, the idea of processing effort may seem somewhat far-fetched. Yet if we
look more closely at the version(s) we prefer, as opposed to those we do not
prefer, we will find a processing advantage in each preferred version which
eliminates a processing disadvantage of the other version.

Take the first paraphrase with its initial temporal adverbial:

(a) At the beginning of creation stands the emergence of form.

In combination with stands, the initial adverbial admits the subject of the
sentence in a position after the verb, that is, in a position which is not the
normal position for English subjects. If we assume that normal structures are
easier to process than those which deviate from our normal expectations, the
first paraphrase has a processing difficulty that does not exist in the other
paraphrases, which all place the subject before the verb. (The second para-
phrase with there is a border case as there is only a structural dummy filling
the position of the real subject, which is yet to follow.)

If this line of reasoning is correct, we can answer the above question right

away. We can expect the preferred version to be the paraphrase with the
fewest processing difficulties, the version on which we have to spend the least
processing effort. All else being equal, this paraphrase will carry the highest
degree of optimal relevance.

If speakers or authors opt for the version with the highest degree of

optimal relevance, they do this in the interest of the reader, whom they want
to understand their message quickly – as this will in the end serve their own
goals. Optimal relevance is determined by processing effort of language
perception. The processing effort of language production, that is, the effort
that goes into encoding a message, is something else. Yet, to secure optimal
relevance we have to control what we produce, which we can only do by
making use of language perception. Thus, even in writing or speaking, the
criterion of optimal conditions for language perception is of prime
importance.

But why should expectations about the position of subjects play a role in

language processing? After all, quite a number of people do not even know
what a subject is, let alone what its normal position is or the special con-
ditions under which we can deviate from the norm.

Setting the scene

Linguistic knowledge

Knowledge about the properties of a language exists in two forms, explicit
and implicit. We all share the implicit knowledge which enables us to use our
own language adequately and so, for example, place the subject in its
grammatically required position. But when we want to talk about this
knowledge, we have to make it explicit. As we have no direct access to our
implicit knowledge, we can talk about it only in the form of hypotheses –
hypotheses about the elements constituting it and about the way in which
these elements are used.

Some of the theoretical concepts used when we talk about language have a

long tradition. They belong to the things we hear about at school and may
not remember afterwards. But if we want to name the conditions impeding
or promoting optimal processing, we have to make the implicit knowledge
explicit and identify the differences between the paraphrases.

Most of the differences are tied to the grammatical properties of the

linguistic structure, that is, to those properties that bind words together into
phrases and relate them to classes of sentences constituted by different words
but similar relations. Let us recall some of the basic concepts used in the
description of sentence structures and their constituents. The major
categories of word classes are verbs (like stand and begin) and substantives
or nouns (like creation and emergence) and there are some minor categories
like articles (the) and prepositions (of, at). Together they form ever-larger
word groups, ‘phrases’, like the beginning, at the beginning, emergence,
emergence of form, the emergence of form, with the emergence of form,
began with the emergence of form, up to whole sentences: creation began
with the emergence of form.

Word groups have different functions in their sentence, depending on the

relation in which they stand to one another. The verb with its extensions
functions as the predicate. The predicate is asserted about the nominal
phrase that functions as the subject, which is, in other words, the phrase the
predicate is attributed to. In (f), for example,

(f) Creation began with the emergence of form.

began with the emergence of form is the predicate attributed to the subject
creation. Subject and predicate are primary relations, which can be modified
by secondary relations, as for example adverbials, like at the beginning of
creation, which modifies the entire sentence.

There are many English sentence structures like (f).

(e) The emergence of form began with creation.

is just one of them. But there are also many German sentence structures
similar to (f), although the original is not among them.

Setting the scene

Verbs, nouns, subjects, adverbials – these and similar categories and

relations are the inventory from which the variety of paraphrases are
formed. The variation could concern the positions of constituents, as in:

(a) At the beginning of creation stands the emergence of form.

and:

where the temporal adverbial and the subject swap places. It could also
concern the grammatical relation of the constituent as in:

(e) The emergence of form began with creation.

and:

(f) Creation began with the emergence of form.

where the things talked about are alternatively encoded as subject or as
temporal adverbial; it could even concern word formation, as in Gestalt-
bildung vs. the emergence of form, etc.

The difference in the grammatical relations could be associated with a

difference in the word class, as in (c) and (f ), where the temporal relation is
expressed by the noun beginning in the adverbial phrase or the verb began in
the predicate, and so on and so forth.

Linguistic knowledge does not end with the repertoire of syntactic

elements and relations, however. It comprises all the information of the
lexicon of a language and this does not only determine the way in which
individual elements can be combined into phrases and clauses, it determines
also what the lexical elements mean.

Meaning can be subdivided into semantic meaning, carried by the lin-

guistic forms themselves (including phonetic and syntactic-morphological
aspects), and contextualized meaning, enriched by the knowledge associated
with a particular contextual usage of a linguistic expression. The categorial
repertoire of semantic elements can be generalized into individuals and
predicates forming propositions, that is, meanings that can be subjected to
modal modifications and assessed in regard to their truth values. Let me
illustrate this.

The examples:

(e) The emergence of form began with creation.

and:

(f) Creation began with the emergence of form.

Setting the scene

express the same propositional meaning, constituted by the individuals
creation, the emergence of form and the predicate began (with).

The contextually ‘instantiated’ meaning (that is, the meaning that is used

in a certain context) is used in an interactive situation, where it acquires a
certain communicative sense,

in our example that of an assertion (by which

the speaker claims that what (s)he says is true). Seen within its discourse
context, the meaning of a sentence constitutes a simple or complex
information unit, structured into various layers according to their functions
within the progress of discourse as topic and comment, background and
focus, which are basic concepts to be taken up in more detail later on. For
now, it may suffice to note that (e) and (f) structure their propositional
meanings alternatively.

Although the line is difficult to draw, there is a linguistic and an extra-

linguistic part of meaning. The former is carried by our (mostly implicit)
linguistic knowledge, the latter by our knowledge about the relevant world
and general principles of inference. Looking at translation under the aspect
of contextual appropriateness, we have to access both types of meaning. A
great deal of the extralinguistic meaning is provided by the linguistically
encoded co-text. Although the contextualized meaning is built up from
linguistic knowledge, it is enriched by all the extralinguistic aspects associ-
ated with it, which can be highly specialized knowledge (as in the example
about the polymers) including all the culture-specific aspects which may play
a role in the discourse.

Linguistic meaning itself can be expressed explicitly and implicitly, with

different degrees of affirmativeness in case of presuppositions and implica-
tions. The latter can again be distinguished into conventionalized impli-
cations, that is purely linguistic meaning, and conversational implicatures
(Grice 1975), which draw upon extralinguistic meaning. It is the conver-
sational implicatures associated with the original and (f) that compensate for
the difference between the propositional meanings of the German and the
English version. (See the further discussion of this example in Chapter 2.)

Sentence processing

Identifying the linguistic properties that distinguish various paraphrases is
simply a precondition for the crucial question to which we will now return:
why would one prefer a paraphrase with property X to other paraphrases
within a certain context? And the general answer to this question suggested
was: property X is easier to process than alternative properties. But how can
we prove this? Why for example should subjects positioned before verbs be
easier to process than subjects positioned after verbs?

The answer indicated above was that the preverbal position was the

normal position for English subjects, and that this is something that belongs
to the implicit knowledge possessed by everybody who speaks English. And
everybody who speaks English knows under which conditions s/he can

Setting the scene

expect exceptions to this: in yes–no questions, like ‘Was the boy there
again?’; after negative adverbials at the beginning of the sentence, as in
Hardly had he begun . . .; and under some odds-and-ends conditions,
including those in paraphrase (a). But in contrast to the first two classes of
exceptions, which are easy to identify by their specific conditions – finite
verb or negative adverbial in initial position – paraphrase (a) has no indi-
cators which could ‘license’ the inverse order, subject after verb.

Now we can see that it is not the inverted order of subject after verb itself

that hampers processing. The canonical cases of subject–verb inversion will
certainly not present any processing difficulties. It is the unexpectedness of
the inversion, the fact that we have not been prepared for it by the initial
elements, that turns the order of subject after verb in (a) into a processing
obstacle.

‘Obstacle’ certainly sounds like a big word for something that may take no

more than milliseconds. Still, we process a sentence from its beginning to its
end, in a slow motion picture, we could say, step by step, or ‘incrementally’
as the technical term has it. (Just look at the eyes of a reader moving from left
to right, top to bottom or whatever the direction may be that the sentences
are written in.)

If the sentence is long, we will break it up into smaller chunks, the way we

do when reading out aloud. Although theorists specializing in language
processing (psycholinguists) still disagree on the details of the process, there
is a great deal of evidence for a certain autonomy of these individual chunks,
or ‘information units’. That is, our brains seem to become quite absorbed in
the process of analysing such an information unit, starting from its syntactic
structure and proceeding to its explicit and implicit meaning. Inside the
chunk, expectations are built up on the basis of what we have already
processed in the chunk and identified in accordance with our implicit
knowledge of linguistic structures. This enables us to predict some aspects of
the next elements. Afterwards, at the end of the process, when we ‘wrap up’
our processing results and integrate them into those of the preceding
information units, we will know whether our current analyses and predic-
tions are right or not.

If we end up on the wrong track, and the results do not make sense, we will

have to break off and correct our analysis – something that may happen
almost anywhere in a sentence. If sentences are not well formulated, we may
spend much time and effort in finding out what the intended message was.
This end may often be achieved only by integrating the results of our
processing into what we already know from the context, throwing out all the
interpretations that do not match the context. Sometimes, even this last step
will fail. But in all cases which are not optimally formulated it will take an
extra effort to overcome processing obstacles, sort out ambiguities and
correct processing errors.

As all these processes take up almost no time and seem to happen more or

less behind our backs, we are hardly ever aware of any individual problem.

Setting the scene

What we do notice instead is a vague feeling of irritation and, perhaps,
momentary disorientation. In any case, we have the definite feeling that
whatever else the message may be, it is not well formulated.

A translation that is not well formulated fails the Translation Maxim in

two aspects: it does not meet the constraints of the target language which
follow from the Principle of Optimal Relevance; and it misses the style of the
original. Any reader of the translation will assume that the style of the
original did meet the corresponding constraints of the source language
(unless there is evidence to the contrary, relativizing optimality – a topic we
will turn to in the last chapter of the Key.)

Considering selection restrictions and wrapping up

The requirement ‘subject before verb’ is met by five of the paraphrases,
including one, (b), which allows us to retain the original topic. If we could
stay closer to the original this way, why do we not opt for (b)? The answer is
that things begin to get better only when we use a verb that matches the type
of elements it relates to each other. Creation and emergence are both con-
ceived of as events or processes, but predicates like stand and be are more
static than processual. Stand even involves the idea of an upright position,
which can be metaphorically taken to signal stability as in the agreement
stands, but it does not match processes.

Although the beginning of events could be viewed as a point in time (that

is, a time interval without duration), the concept of stand does not really go
well with the concepts of creation and emergence. Be, if it is used on its own,
is no better in this respect. It refers to the existence of things, and processes
like the emergence of form are not normally conceived of as things. The
paraphrases with stand or be are stylistically poor because they produce a
mismatch between the meaning of the verb and the meaning of its ‘argu-
ments’, the constituents related to each other by the verb.

The fact that the violation of selection restrictions does not meet the

Principle of Optimal Relevance is obvious. However, there is the German
original, where the combination of stehen with Gestaltbildung and Anfang
der Schöpfung does not violate any semantic restrictions. What this means is
that semantic restrictions are language-specific. They are properties of the
lexical elements of a specific language and cannot be predicted from our
general knowledge about states and processes or the like.

There may be something regular in the way selection restrictions are

eliminated with the transfer of meaning from concrete to abstract areas,
something that could be related to the parametrized properties of a language.
We will return to this question at a later point, discussing the phenomenon of
alternative perspectives. For now, we will turn our backs on such lexical
problems, however predominant a challenge they may be in the real world
of translations, and return to the domain of language-specific processing
conditions which relate to any sentence.

Setting the scene

Let me recall the basic assumptions I have made in setting the scene.

Choosing a non-historical perspective of translation, which views trans-
lations as contextualized instances of possible correspondences between
languages, I have suggested focusing on a subset of translations following a
translational norm that is linked to the Principle of Optimal Relevance.
Felicitous translations will in this sense be translations that have found an
optimal way to stay as close to the meaning and the style of the original as
possible within the constraints of the target language.

Using sets of control paraphrases (minimally varied paraphrases of the

original structure), the method suggested appeals to our implicit knowledge
about the appropriateness of paraphrases relative to each other within a
certain context. Although this means abstracting away from the different
knowledge bases of individual readers, the tight contextual and co-textual
constraints placed on the paraphrases should secure a relatively high percent-
age of agreement among native speakers of a similar language competence.

As the comparison is restricted to a certain set of paraphrases, we can

never exclude the possibility that there are other paraphrases which may
be even more optimal than those compared. Yet the decision about
appropriateness would subject them to the same procedure of comparative
assessment in context.

Comparing the contextual appropriateness of a paraphrase relative to

others in a certain context means first of all comparing the processing
conditions of the paraphrase relative to others. Processing conditions are
context dependent and language specific. Both aspects interact in the way in
which the information is distributed onto the linguistic structures of original
and translation.

The next six chapters, focusing on word order, case frame and structural

explicitness, will elaborate in detail the main assumptions of the book on the
interaction between language-specific properties and discourse. Let me add
to the scene a bird’s-eye view of the ideas involved. It is obvious that
languages do not only differ in their vocabularies and grammars but also in
the way in which those systematic options are used. There are a great
number of cases where even similar means are used differently. Now, this
book claims that the primary reason for the different uses lies in a small set of
basic grammatical properties, which form something like a mould through
which similar material may be shaped differently.

Despite their many similarities in terms of word classes, syntactic

functions or grammatical categories, English and German are quite
impressive cases of such basically differing moulds. Irrespective of their
different aspectual and numerical categorizations of events and things, they
could form many similar sentence structures, using the same word order,
parts of speech, active or passive perspectives, syntactic coordinations or
subordinations, and the like. But there are a great deal of cases where they
use a different word order, different parts of speech, the opposite voice, or a
different linkage of clauses.

Setting the scene

If you ask native speakers why they do not make use of the structural

options that would be closer to the other language, they can only tell you that
it does not sound right. But the reason behind this vague feeling is the native
speakers’ tacit knowledge about the way in which the basic features of their
language interact with the general conditions of the discourse to which a
sentence belongs.

Spelling out the details of this interaction is a long and highly complex

affair – as the discussion of the examples in this book will amply demonstrate
– but the main idea is relatively simple. There are some basic parameters
determining the special grammatical ‘mould’ of a language. The most
important aspects distinguishing German and English are related to, firstly,
the direction in which words, first of all verbs, are extended into groups of
words and, secondly, the freedom with which we may deviate from this basic
word order and ‘move’ parts of the sentence to other positions, as for
example from a position after the verb to one before it.

Borrowing a concept from molecular biology, we could say that English is

a language with a rightward ‘spin’ – the direction of verbal extensions is to
the right – and German is a language with a leftward spin – the direction of
verbal extensions is to the left. On the other hand, English word order is
relatively rigid and does not permit too many deviations from the basic
order, while German may deviate from its basic word order in many ways.

There are a few grammatical reasons for deviating from basic word order.

For example, if we want to form a yes-no question, we have to exchange
subject and auxiliary. But in most cases, deviations from basic word order
are related to the integration of a sentence into its discourse. And this is
where the interaction between grammatical features and discourse con-
ditions comes in.

The integration of a sentence into its discourse follows some universal rules

of information structuring, which language-users adhere to in the interest of
felicitious communication. Of the many ways in which we can formulate a
thought, we will intuitively prefer those which transport our message most
efficiently, which means that we will not only choose the appropriate words
but also the appropriate sentence structure to allow addressees to link the
message to their thoughts as required.

However, the different moulds of English and German do not allow us to

follow the same strategies in linking our sentences to the discourse, that is,
we cannot but adhere to different discourse-linking strategies if we formu-
late a sentence in a language with a rightward spin and a relatively rigid
word order or in a language with a leftward spin and a ‘free’ word order.
The overriding difference is that the greater word-order freedom in
German makes the impact of discourse linking much stronger in German
than in English. It is to the most important area of this difference that
we shall now turn.

Questions of order

There can be no doubt that questions of order have top priority. In
processing messages, we have to submit to the White Queen’s remark: ‘It’s a
poor sort of memory that only works backwards’ (Carroll 1968: 254). As
Alice says, ‘We can’t remember things before they happen’. We can only
interpret sentences against the background of what we know, or, more
precisely, what we can think of at the time of processing a message. Some of
the things we can think of at this very moment will be the things we were told
before in the immediately preceding message, including all the thoughts that
the previous sentence(s) had activated in our brains. Elements which have
already been activated are, certainly, easier to identify than elements we
have not yet thought of. Thus, it seems only natural to place the familiar
before the new. But even if we ignore all those particular situations where we
might want to deviate from this natural order of presentation, the natural,
easy-to-process order of elements is a highly complex affair, subject to
linguistic and extralinguistic constraints. It is clear that grammar plays a
decisive role here and as languages differ in their grammars, we can expect
grammatically based differences in the order of elements. But as we shall see
soon, grammar has its repercussions at the level of style related to contextual
appropriateness and conditions of optimal processing.

The key concept for questions of order is that of information structure,

which links word-order rules to discourse-structuring rules, that is, to the
rules that control progress in discourse. As the following discussion of
initially simple, then increasingly complex, cases will show, translations
have to be reordered in many cases if they are to achieve contextual
appropriateness and meet optimal processing conditions in the target
language.

As the grammatically determined order and its interaction with con-

textual aspects is no simple matter – its complexity being doubled by the fact
that we are dealing with two languages – it will take us quite a while to get
through the various aspects of order. But as questions of order are also dealt
with in the subsequent topics of primary relations, structural weight, and all
the others, the extra effort needed for the following sections will serve us well
in more then one way.

Questions of order

Parametrized processing conditions

General conditions for easy processing ought to be the same in all languages.
The Principle of Optimal Relevance makes us prefer linguistic structures that
enable us to quickly analyse the grammatical relations between the elements
of a sentence which determine its meaning. But individual languages present
different structural properties for these general processing conditions. The
expectation of the subject-before-verb order and the special licensing
conditions for its inversion relate to a feature of English not shared by
German.

When I say that processing conditions are parametrized, I refer to the fact

that the grammatical parameters which determine the profile of a language
determine the conditions for processing linguistic expressions in that
language, too. Let me give you a first rough survey of what this means. As
has already been pointed out, one of the most important parameters to
distinguish is that of directionality, the rightward or leftward ‘spin’ of a
language. It determines the direction of the structural extension which a
categorial head, like a verb or a noun, can take. There are languages where
this is a consistent cross-categorial property, as for example Japanese, which
extends verbs, nouns and postpositions (the parametrized version of
prepositions) to the left. English and German are more of a mixed nature. In
English the dominant direction of verbs, nouns and prepositions is to the
right. At first sight, it seems to be to the right in German, too, but if one takes
translation as empirical evidence and is thus forced to include a wider variety
of possible extensions, one will encounter very many examples where the
different order of the elements points to an alternative directionality of the
English and the German verb phrase. Thus, one can assume that, basically,
the English verb phrase extends to the right and the German verb phrase
extends to the left. There are other factors superimposed upon this basic
difference, especially in German, which deviates from the basic word order
for discourse reasons. English is much more constrained in this respect,
because of the second important parameter distinguishing the two lang-
uages. This is the parameter of configurationality, which has its effect on
word order, too, as it determines grammatical relations like subject or object
by structural configurations (see below).

However, word order is not only determined grammatically, it is also

determined pragmatically by the context/discourse, in which the individual
syntactic structure participates. In contrast to the grammaticalized proper-
ties, which are obligatory, the pragmatic conditions are binding only in the
sense of optimal relevance. Yet if we ignore the pragmatic conditions and do
not choose the structures which are in line with optimal relevance, we cannot
help producing certain effects, too. These are, again, parametrized, that
is, they differ according to the grammatical parametrization underlying
them. (Control paraphrases, especially analogous translations, illustrate this
extensively.)

Questions of order

The interface between grammatical and discourse conditions is ‘infor-

mation structure’. It is not only highly relevant for word order, but also for
questions of structural explicitness and sentence boundaries, that is, for all
the topics we are going to look at in the following chapters. It is one of the
major claims of the Key that information structure is grammatically
parametrized and that our understanding of felicitous translations can be
enhanced greatly by making this type of knowledge explicit. We will now
take a close look at the grammatical and pragmatic details of a pair of simple
and, in the next chapter, a pair of complex sentences. We will use the method
of control paraphrases and a magnifying glass to extract from the examples
all the linguistic and psycholinguistic knowledge needed to describe,
generalize and explain felicitous translation and word order.

Parametrized beginnings

In German, the position of the finite verb, which is always in the second place
in the main clause, can freely be preceded by any other constituent. This
means also that the subject can occur in different positions in the sentence.
Grammatically seen, the order:

Am Anfang der Schöpfung steht die Gestaltbildung.

is just as normal as:

Die Gestaltbildung steht am Anfang der Schöpfung.

There is, however, a difference in the way in which these sentences link up
with their surrounding context. The difference is very much like the differ-
ence between the paraphrases:

(e) The emergence of form began with creation.

and:

(f) Creation began with the emergence of form.

The contextual link belongs to one of the major criteria of optimal encoding
and we will return to it shortly.

What we should notice now is the grammatical difference between the

German options (adverbial or subject) and the English option (only subject).
This is no coincidence, but prototypical for the parametrized beginnings of
German and English sentences. Although many sentences begin alike – for
example with the subject – the examples above and most of those following
show that there are different conditions in English and German controlling
the initial position in a sentence.

Questions of order

The difference between the English and the German conditions relate first

of all to what is traditionally called free word order in German. While
English has relatively tight grammatical restrictions on word order, which
can be lifted only under very special conditions – as we saw with the subject–
verb order – German is, grammatically seen, much more liberal. Although
we can assume something like a canonical order of things in German, too, a
basic word order, this basic order is given up much more often in German
than in English. In contrast to the two or three special conditions requiring a
deviation from the basic word order in English, German word order is
regulated by a whole bundle of conditions, most of which are related to the
integration of the sentence into its context.

The difference between German and English initial positions belongs to

one of the two basic grammatical differences between the two languages. In
fact, each language belongs to one or another of an alternative pair of
options, free or fixed word order, or a mixture of both. Languages with fixed
word orders are mostly configurational languages, which means, as you will
recall from above, that grammatical relations are expressed by certain
structural configurations. An English subject, for example, is characterized
by its preverbal position; nominal phrases in postverbal positions are
complements to the verb, as for example, objects. German, on the other
hand, is more of a non-configurational language. Subjects are not bound to a
special position; they are not part of a special structural configuration. The
difference may be connected to the difference between German and English
word forms. Subjects are characterized by a different case form in German
(nominative) as opposed to objects (dative or accusative) – a difference
English can make only in the area of pronouns like ‘he’ and ‘him’, ‘she’ and
‘her’, but not with fully lexicalized noun phrases.

The difference between German and English is enhanced by the fact that

German has a grammatical gender, it has das Symbol (it), der Verlust (he) and
die Tatsache (she), while English has ‘the symbol’, ‘the loss’ and ‘the fact’.
The pronominal distinction helps to identify grammatical relations in German
much more often than in English, where it neutralizes all these differences.
We will look at such an example under another heading later on (p. 83).

Whatever the internal relations between the parameter of configur-

ationality and the different case forms may be, there can be no doubt about
the importance of grammatical parameters for language-specific processing
conditions, especially in regard to questions of order. We will have ample
opportunity to study the repercussions of the grammatical difference, the
most important one being its effect on the pragmatic, discourse-related role
of the initial position.

Information structure: the point of departure

If we have two elements to distribute within a sentence, we are, theoretically,
free to order them either way. Grammatically seen, paraphrases (e) and (f)
are equally acceptable.

Questions of order

The emergence of form began with creation.

or:

Creation began with the emergence of form.

Yet the message is clearly different. Although both sentences assert the
simultaneity of the emergence of form and creation, the point of departure of
the assertion is specified in the subject, which means that it is the emergence
of form we talk about in the first sentence but creation in the second
sentence. In one case we characterize the emergence of form by linking it to
the beginning of creation, in the other we characterize creation by linking it
to the emergence of form.

The point of departure is traditionally called the ‘topic’ of a sentence, or

‘theme’, while the remaining part is called the ‘comment’ made about the
topic. The grammatical relation between subject and predicate is a
prototypical way of expressing this. The German original does not take the
subject but the temporal adverbial as its point of departure:

Am Anfang der Schöpfung steht die Gestaltbildung.

The subject is part of the predicate carrying the comment. The reverse:

Die Gestaltbildung steht am Anfang der Schöpfung.

turns the subject into the point of departure, with the temporal adverbial as
its comment. In both sentences we seem to identify the initial part of the
sentence with the topic and the part following it with the comment.

The distribution of the elements into topic and comment is one way of

structuring the information conveyed by a sentence. There is at least one
more aspect constituting the information structure of a sentence. We will
come to it shortly. As for the topic/comment distinction, however, we are
faced with a difference between the original and the translational versions (e)
and (f):

(e) The emergence of form began with creation.

(f) Creation began with the emergence of form.

The concept of beginning is part of the German topic and part of the English
comment in both versions. Does this mean that the messages of the
translation and the original differ?

If we look at the way in which the message presents itself at the surface of

the linguistic structures, the original and the translation do differ. That is,
they are not fully equivalent. Let us call this surface-based equivalence
‘surface equivalence’. With all the differences between languages, we can

Questions of order

predict that surface equivalence is hardly ever possible. But if we stand back
and look at the message from a greater distance, taking into consideration all
the aspects we could expect the reader to associate with the surface-based
interpretation, it will be enriched by many associations and implications,
which could compensate for differences between the surface interpretations
of the original and its translation.

Thus, the concept of creation itself implies a beginning. It refers to some-

thing that was not there before. To say:

Creation began with the emergence of form.

localizes the implied beginning of creation and the emergence of form at the
same time. Taking explicit and implicit elements together, the English and
the German sentences have the same meaning – though it is distributed
differently among the explicit and implicit components of the message.

Yet even this global perspective cannot eliminate the difference in the

information structures of the original and its translation. This is obviously
part of the surface equivalence, which will mostly have to be sacrificed in the
interest of grammatical acceptability and optimal processing conditions.
And, in this case, the ‘loss’ boils down to almost nothing when we weigh it
against the gains we have made with (e) and (f) as opposed to the other
paraphrases with their processing disadvantages.

Still, as (e) and (f) choose different topics, they cannot really be equally

optimal translations. If we compare them with the German original, we
could say that (f) is more optimal (in terms of the Maxim) than (e), since the
topic of (f) is at least to some extent identical to the topic of the original.

Reordering information structures

Let us now turn to the other aspect of information structure: the distinction
between focus and background. While the topic – that is, the starting point,
the element the sentence is about – is generally located at the beginning of a
sentence, the focus – that is, the most important part of the information – is
located at a later point, frequently at the end of the sentence.

Thus, unless we have some contextual evidence to the contrary, we would

consider the German original:

Am Anfang der Schöpfung steht die Gestaltbildung.

and its translation:

Creation began with the emergence of form.

as examples of such canonical distribution of information. Irrespective of the
difference in the topic, Am Anfang der Schöpfung/at the beginning of

Questions of order

creation as opposed to creation, the distribution of topic and focus is the
same here in German and English. The topic is located at the beginning of the
sentences, and the focus, Gestaltbildung/the emergence of form, is at the end
of the sentences.

But this is not always the case. In fact, there are many examples where the

position of the most important element of a sentence differs in German and
English as well as many examples where the English focus occurs in topic
position.

Let us look at one of these examples. The text speaks about the attempt of

physicists and chemists in the first half of the nineteenth century to apply
Newton’s mechanics to the atomic structure of matter, introducing the
problems they were bound to face with the sentence:

Allerdings stand diesem Vorhaben ein schier unlösbares Problem im
Wege.

(EW 1983a: 165)

This has been translated into English as:

An apparently insurmountable problem, however, stood in the way of
this effort.

(EW 1983a: 143)

If we ignore the sentence adverbial allerdings/however, we can say that the
original order of topic and focus has been reversed in the translation. If we
consider the definite noun phrase, diesem Vorhaben/this effort, to be the
topic, and the indefinite noun phrase, ein schier unlösbares Problem/an
apparently insurmountable problem, to be the focus of the sentence, then we
can say that the topic precedes the focus in German and follows the focus in
English.

The grammatical reasons for the change are, at least to a certain extent,

obvious. The syntactic properties of im Wege stehen/to stand in the way of
are different in German and English. While German requires an indirect
object, a dative object, English can only add this argument to way: in the way
of this effort.

The different syntactic integration of effort/Vorhaben into the sentence

structure results in different conditions for the use of this element as topic or,
more precisely, for its location at the beginning of the sentence. While
German can say:

Diesem Vorhaben stand ein unlösbares Problem im Wege.

that is, prepose the dative object and leave the nominal extension of the
verbal complex im Wege stehen at the end of the sentence, English can only
prepose this effort together with its structural head, in the way of:

Questions of order

*This effort stood an apparently insurmountable problem in the way of.
In the way of this effort stood an apparently insurmountable problem.

However, the grammatically acceptable version not only has a different
topic than the original, it is clearly not as good as the existing translation:

An apparently insurmountable problem stood in the way of this effort.

The reason for the preference of the latter is the condition of the preverbal
position of the English subject, with which we are already familiar. But this
time, the difference in the optimal distribution of information amounts to a
difference in the overall distribution of information, and if we were to take
position as the defining criterion for topic and focus, we would have to say
that the original and its translation are different messages altogether. But
intuitively we would like to say that the English and the German sentences
are just different formal versions of the same message. If this intuition is
correct what do we have to assume in order to account for it?

The most important element

In its spoken form, the most important element is generally associated with
the main stress in the sentence. If we were to read the German and English
sentences aloud, we would place the main stress on unlösbares Problem/
insurmountable problem, which is the most important element of the
sentence in either version.

The technical term for the element carrying the main stress is ‘focus

exponent’. It is our implicit knowledge about focus assignment that makes us
stress a particular element in the sentence, that is, choose a certain element as
focus exponent. But why should it be insurmountable problem/unlösbares
Problem?

If we consider something to be the most important element, it will be more

important than all the others it is compared to, but only in relation to the
specific aspect it is important for. For the most important element of a
message, obviously, this aspect will be the purpose of the message, in its most
neutral form the purpose of informing. In this case, the criterion determining
the importance of a particular element will be the degree of novelty it has for
the recipient. New elements will be considered more important than ‘old’
ones, the elements with which we are already familiar.

The criterion of novelty versus familiarity/givenness requires us to take

account of the context, whether immediate or wider, text or situation. It is a
contextual criterion which helps us to determine focus by comparing the
informational elements of a sentence with those of its context.

In the end, the criterion of novelty does not suffice as there are, for

example, sentences which have only new or only old information. But it will

Questions of order

be enough to settle the question for the last example. The predicate stand in
the way of relates two arguments, only one of which, an insurmountable
problem, contains new information. The other one, this effort, belongs to the
knowledge we have gathered from the context: the attempt of physicists and
chemists to apply Newton’s mechanics to the atomic structure of matter in
an effort to arrive at a unified description of all the characteristics of matter.
That is, this effort is old, contextually given information, while everything
else is new, the insurmountable problem as well as its relation to the effort:
stood in the way of.

The difference between given and new information is, to a certain extent,

reflected in the difference between definite and indefinite phrases. In
particular, if the predicate relates a definite phrase (like the beginning/this
effort) and an indefinite phrase (like a problem/problems) to each other, we
can expect the definite phrase to contain given information and the indefinite
phrase new information.

If we identify everything new with focus, everything but this effort would

be part of the focus. This is a concept of focus common to many focus
theories.

Information structure is partitioned into focus and background, that is, in

most cases new and given information. Seen this way, the focus of a sentence
may extend beyond its focus exponent, the element that carries the main
stress of a sentence. Focus theories speak of a ‘projecting’ focus whenever the
focus is not restricted to a focus exponent.

Now, if we compare the distribution of the focus exponents, we can say

that the English version starts with the focus exponent:

An apparently insurmountable problem, however, stood in the way of
this effort.

while German places the focus closer to the end of the sentence:

(Allerdings stand) diesem Vorhaben ein schier unlösbares Problem im
Wege.

The difference between the two languages looks quite arbitrary, but it is the
result of parametrical differences in either case. It is the subject-before-verb
parameter which places the most important element at the beginning of this
English sentence, and it is the directionality parameter which accounts for
the final position of the verbal extension: im Wege. I will now address this,
though not without summing up what we have seen so far. While the
beginning of the English sentence is determined by grammatical principles,
which can override the canonical distribution of information in the sentence,
the beginning of the German sentence follows information structural
aspects. It links up to the local context, resuming a given element.

Questions of order

Structural focus

It seems that the position of the most important element is variable. If we
take our first two examples, we find that there are three different positions
for the focus exponent: at the end of the German and English sentences in the
first example, and in the middle of the German sentence and at the beginning
of the English sentence in the second example.

There is a structural regularity, however, underlying this distribution and

all the other examples we have yet to look at. In a way, the distribution
reflects the basic architecture of a message, which in its most elementary
form links a subject and a predicate. The element which will always be
present in the predicate is the verb, which as a rule relates one complement
(for example, an object or a predicate) to the subject. It is this complement,
namely the closest argument of the verb, which we can normally expect to
contain the focus exponent of a sentence.

If there is no focusable argument,

it could also be the verb that serves as the focus exponent.

Now, this concept of a focus exponent bound to a certain position in the

structure of a sentence is purely formal. However, the idea that the verb
attracts the focus seems to be quite ‘logical’. After all, it is the predicate
which normally determines the comment we make about the topic of a
sentence.

The closest extension of the verb, the verb-adjacent phrase, is a structural

criterion which may or may not coincide with the contextual criterion for
focus. Normally, that is, in well-formed sentences, the contextual and the
structural focus will identify the same element as focus, and, for the time
being, we will simply presuppose the identity of structural and contextual
focus.

Thus, in:

Creation began with the emergence of form.

it is the emergence of form completing the verbal expression began with
which we can identify as the structural focus exponent of the sentence.

But with stand/stehen the situation is different. We can take stand in the

way/im Wege stehen as a fixed verbal expression, where the local adverbial
specifies part of the lexical meaning of the expression. Consequently, the
subject, an apparently insurmountable problem/ein schier unlösbares
Problem, is the closest argument to this complex verb and will contain the
structural focus exponent.

For reasons that will be given shortly, we can assume something similar

for am Anfang stehen, and therefore expect the subject, die Gestaltbildung,
to contain the focus exponent in the German original of the first example,
too. Thus, English and German follow the same rule in selecting a structural
focus exponent – each time it is the closest argument of the verb that will
contain the focus exponent. But the difference between the English and

Questions of order

German grammatical parameters causes the focus exponent to occur in
different structural positions. While the subject is bound to the position
before the verb in English, it may follow the verb in German – even at some
distance, as the German original of the second example shows:

Allerdings stand diesem Vorhaben ein schier unlösbares Problem im
Wege.

While the position of the English subject is determined configurationally –
that is, purely grammatically – the position of the German subject varies
according to its role in the information structure. If it is the focus exponent of
the sentence, the subject, too, will be placed at the end of the German sentence.

The prediction seems to be borne out in the first example:

Am Anfang der Schöpfung steht die Gestaltbildung.

but not in the second, where the end of the sentence is occupied by part of the
complex verb: im Wege. What is worse, we seem to have lost sight of the
generalization that the focus exponent is the element closest to the verb.
However, both objections can be dispersed by what may be considered the
most important grammatical difference between the German and the English
sentence structure, namely the basic position of the finite verb and the
associated directionality of its structural extensions.

Basic verb positions

While the finite verb precedes its complements in English, it appears in two
alternative positions in German – it precedes its complements in the main
clause:

Am Anfang steht die Gestaltbildung

but follows them in the subclause:

daß die Gestaltbildung am Anfang steht.

The verb thus appears at the left or the right periphery of the verb phrase. But
if we have a complex verb phrase consisting of an auxiliary, like ‘be’, ‘have’,
‘will’, ‘could’. . ., which helps to express various grammatical aspects of the
verb, and a main verb, the main verb will occur in the right peripheral
position in main clauses, too. Compare the position of the finite verb, würde,
at the beginning of the verb phrase, and the position of the main verb, stehen,
at the end of the verb phrase:

Am Anfang würde die Gestaltbildung stehen.

Questions of order

Now, it is the main verb, the lexically meaningful verb, which takes
complements, determining their number and form: begin with, stand in the
way of . . .. It determines also the neutral order of these arguments if there are
more than one. Although linguists still argue about the basic order of
arguments for each individual verb, the assumption that there is a basic order
is generally accepted. Today the idea of a basic order is even extended onto
those extensions of the verb or verb phrase which are not predetermined by
its lexical meaning, but added freely, that is, the various classes of adverbials.

The interesting thing about the position of the focus exponent and the

verb in German is that the basic order of the extensions of the verb is exactly
the same in main and subclauses, irrespective of the position of the finite
verb. And if we take the position of the closest argument of the verb as an
indication of the basic position of the verb, we have to conclude that the final
position, that is, the position of the verb in the subclause, is the basic posi-
tion. The closest argument of the verb will always appear as the right-most
extension of the German verb phrase whether it is followed by a verb or not.
Thus, the basic order of arguments would be the same in both types of clauses:

jemand gibt

jemandem ein Buch

/someone gives someone a book;

daß

jemand

jemandem ein Buch gibt.

Judging from the basic order of its complements, the German verb has its
basic position at the end of the verb phrase, which is at the same time the end
of the sentence.

Comparing verb positions relative to their complements, we can say that

English has a left-peripheral verb phrase, and German a right-peripheral
verb phrase, with the complements of the verbs extending in alternative
directions. With the left-peripheral position of the English verb and the right-
peripheral position of the German verb, verb-adjacency of focus can mean
different focus positions in English and German. In the case of several verb-
phrase internal constituents, the difference will become more noticeable and
can amount to something we could, in a simplified way, call mid-focus in
English and end-focus in German.

Context sensitivity

A number of other factors, especially information structural ones, are
superimposed on the basic order. Thus, for example, if we have an indefinite
and a definite argument, which amounts to a new and a given element, the
definite argument will normally precede the indefinite:

Der Mann gibt das Buch einem Freund./The man gives the book to a
friend.

This is what we have in our second German example:

Questions of order

Diesem Vorhaben stand ein unlösbares Problem im Wege.

If im Wege is part of the verb, we can say that the indefinite noun phrase ein
unlösbares Problem is the focus exponent as it occurs in the right-most
position, followed only by a part of the verb.

English is less context-sensitive. Following the subject-before-verb rule,

the translator has placed the indefinite subject, an insurmountable problem,
at the beginning of the English sentence, producing a focused topic:

An apparently insurmountable problem stood in the way of this effort.

Still, we can identify the subject as the closest argument of the verb, which is
a verbal expression, stand in the way of, that links its complement (here, this
effort) to the subject.

The lexical difference between jemandem im Wege stehen/stand in the

way of someone amounts to a difference in the syntactic hierarchy of the
structural extensions which allows us to identify the English subject, too, as
the argument closest to the verb – closer than the postverbal extension, this
effort, as this is only an extension of in the way of. Due to the different
syntactic hierarchy, we can identify the subject as the focus exponent of the
English sentence, too.

This type of focused, ‘prominent’ subject occurs quite often in English,

while the corresponding structure in German will be reversed whenever there
is another, less ‘prominent’ extension of the verb.

Concluding the discussion of our second example, then, we can now

confirm our intuition that the English and the German sentences are just
different formal versions of the same message. And if we assume that the
structural focus indicates the contextual focus, we can say that the English
and German versions secure optimal processing conditions for equivalent
focus interpretations and, in this respect, can thus also be considered
equivalent messages.

Hierarchies

Why does the Key use all these highly abstract grammatical concepts to talk
about the focused elements of a sentence? Can we not compare an original
and its translations on the basis of our intuitions about what is more and
what is less important in the message? After all, anybody could have
identified the foci of the last two examples much more quickly without the
criterion of verb-adjacency, which presupposes an entire theory of grammar.
Certainly. But how would we generalize our observations?

Meaning itself is nothing we can hold on to. It changes with each sentence.

Position does not allow any simple generalization, either. You cannot, for
example, say that the focus is at the end of a German sentence if you mean
the orthographical end of the sentence. There are many sentences like the one

Questions of order

with im Wege stehen where the focus is not localized at the end. Both in
German and in English, it would be impossible to find a generalization
related to punctuation. As the examples showed, the focus may be at the end,
at the beginning, or somewhere in between.

The grammatical concept of verb-adjacency allows us to generalize about

all German and all English examples. The concept may not always be easy to
apply (particularly in German, where the verb leaves its basic position when
used in a main clause, but also in English, where we have so many focused
subjects), but what should have become clear by now is that adjacency, or
closeness, is not a linear, but a hierarchical concept. Let me demonstrate this
with a few more examples. Take the following sentence where German and
English use the same structure except for the difference in the predicate:

Gestalt beruht auf Ordnung in Raum und Zeit . . .

(EW 1983a: 88)

Form is a product of order in time and space . . .

(EW 1983b: 69)

The closest extension of the German verb (beruhen/is based on) is the pre-
positional phrase, auf Ordnung in Raum und Zeit. It should contain the
focus exponent. But why should we consider the prepositional phrase the
closest extension of the verb? The verb is a predicate relating two arguments
to each other, the subject and the prepositional object. Could we not
consider the subject as closer to the verb than the object? After all, the two
preceding examples had their closest argument in the subject.

If we simplify things a great deal, we could explain this as follows –

structural extensions of verbs can be either lexically predetermined ‘argu-
ments’ or free modifiers (free adverbials). As a rule, arguments are ‘closer’ to
verbs than adverbials, and within arguments objects are closer to verbs than
subjects.

The distance between the verb and the adverbial reflects the ‘scope’ of the

adverbial, that is, the size of the phrase which is modified by the adverbial.
This differs with different adverbials: modal adverbials specifying the way in
which something happens belong to the inner layers; local adverbials are
more distant, their scope including the modal adverbials; and temporal
adverbials form part of the outer layers, their scope including those of the
others. The details of the many different types of adverbials – some of which
may even occur in several classes – are highly intricate and, of course,
language specific. But the principle of hierarchical ‘layers’ surrounding the
verb and its arguments according to their semantic-syntactic scope is the
same. The hierarchy manifests itself in the linear order of adverbials, which is
left- or right-centred according to the language-specific directionality.

The hierarchy of arguments, on the other hand, is lexically based, that is,

each verb determines the number, type and basic order of the phrases it

Questions of order

accepts as its arguments.

Some patterns occur very often and are seen as

prototypical. For example, in most cases the subject is the highest argument
of a verb with several arguments. While objects and predicates extend the
verb into a verb phrase, subjects extend the verb phrase into a sentence. With
certain verbs, however, as for example stehen, the subject is considered the
closest argument of the verb.

If we take all of this together, we can classify the prepositional object auf

Ordnung in Raum und Zeit in:

Gestalt beruht auf Ordnung in Raum und Zeit . . .

as closer to the verb than the subject Gestalt, but the subject Gestaltbildung
in the German original:

Am Anfang steht die Gestaltbildung.

as closer to the verb than the adverbial.

Identifying the structural focus is not yet the end of the story. We still have

to integrate the result of our formal analysis into the discourse the sentence
belongs to. This may require an extension of the focus onto more elements or
– in case of a mismatch – reanalysis. Let us take a look at these cases of
contextual ‘adaptation’.

Focus projection

Gestaltbildung is a word, though a complex one; the emergence of form is a
phrase. Die Gestaltbildung is also a phrase. If a phrase contains the focus
exponent, which of its elements is it? As function words, like articles, are
normally unstressed, there is not much to choose from in the case of die
Gestaltbildung. The focus exponent is the lexical element of the phrase, the
noun: die G

ESTALTBILDUNG

. But what about the phrase: the emergence of

form?

Following the idea that the focus exponent is mostly the element that is

given the main stress in the sentence, linguists suggest going by stress rule. It
is the last stressable lexical element in a phrase that receives the main stress in
the phrase. Thus we get the emergence of

FORM

. But this is definitely not the

focus we think of when we look at the entire example in its context. Even if,
for the moment, we ignore the apparent paradox that the focus exponent is a
given element, we would want the focus to cover, at least, the whole phrase,
the emergence of form.

This is where the concept of focus projection comes in. The focus feature is

projected onto a larger part of the sentence; the extension of the projection is
guided by structural constituency limited by contextual givenness.

To a certain extent, givenness can be read off from definite articles, as we

said. But unique concepts, like die Schöpfung, are used with a definite article

Questions of order

also if they are not contextually given. Thus, the criterion is not very reliable
and we do have to look at the context itself. The context of our first example
is nothing but the title of the section introduced by this sentence. While it
contains Gestalt (which is rendered as form), it does not contain any of the
other elements of the sentence. We therefore have to conclude that the focus
projection extends over the whole sentence:

[Am Anfang der Schöpfung steht die G

ESTALTBILDUNG

We could also assume that Schöpfung/creation is a concept we are all
familiar with, which may well have been the reason for the authors to choose
it as their starting point. But the concept is certainly not present in our minds
when we read the sentence. I will therefore consider it part of the focus
projection, to be distinguished from the rest by its status as topic:

[[

Topic

Creation] began with the emergence of

FORM

]

The German original would be quite similar:

[[

Topic

Am Anfang der Schöpfung] steht die G

ESTALTBILDUNG

If we analyse our last example, the result is different. The last stressable
element is the noun: Zeit in German or space in English. But coordinations
relate phrases with similar properties, so I will assume the focus exponent to
be identical with both conjuncts: R

AUM

und Z

EIT

;

TIME

and

SPACE

. (The

different order of the conjuncts is an intricate problem, which I will ignore
for the sake of simplicity.)

What about focus projection? A look at the context tells us that Gestalt/

form is given, but so is Raum und Zeit, time and space. The preceding
sentence is:

From this formlessness emerge the structures of time and space: light and
darkness, day and night, heaven and earth, land and sea.

(EW 1983b: 69)

Thus, we have to assume that the focus projection excludes the structural
focus exponent, shifting the focus onto the next focusable element:

Gestalt [beruht auf O

RDNUNG

] in Raum und Zeit . . .

Form [is a product of

ORDER

] in time and space . . .

That is, the context tells us to ‘defocus’ the structural focus (whether this has
any effect on the prosodic focus structure is an open question); structural and
contextual focus do not coincide. We have to reanalyse the sentence when

Questions of order

integrating its information structure into the preceding context. This is a
processing disadvantage which can hardly be avoided. But as we have very
much the same situation in both languages, this need not bother us too much.

Focusing elements

There is one more formal aspect of the last example we should make a note
of. The focus exponent, unlösbares Problem/insurmountable problem, is not
the last stressable element of its phrase, ein schier unlösbares Problem/an
apparently insurmountable problem. This would be only problem. But even
if we do not know anything about the context, our native intuition tells us
clearly that the focus falls on insurmountable, too.

This time, our intuition is guided by our implicit grammatical knowledge

not only about argument positions, but also about the focusing effect of
certain lexical elements, as for example the adverb schier – a slightly old-
fashioned expression, oscillating between degree and truth, something
between ‘almost’ and ‘apparently’.

Other focusing elements are ‘only’, ‘even’, ‘also’, ‘almost’, ‘already’ and

many more. They all mark an element in their scope as focused. In contrast
to structural focus, which can even project over the whole sentence, lexical
focus does not project over more than its immediate constituent and it is
always contrastive in its interpretation, that is, it excludes other possibilities.
The contrastive nature of negated elements like unlösbar/insurmountable
will thus automatically attract the focus which the focus element assigns to
an element in its scope.

In most cases, focusing elements are used to mark an extra focus, that is,

there are sentences which have more than one focus and focusing elements
will mostly indicate such cases. Take for example the German original:

Eben diese M

ÖGLICHKEIT

muß in den frühen Perioden der Evolution . . .

von großer B

EDEUTUNG

gewesen sein.

(EW 1983a: 309)

where we have at least two foci: the lexical focus after eben and the
structural focus at the end. In English, the focusing element itself is focused,
but its meaning ‘passes’ the focus effect on:

In the early periods of evolution, this

VERY

possibility . . . must have been

of great

IMPORTANCE

(EW 1983b: 276)

As in this example, contrasted elements are in very many cases given
elements, which means that focused and given need not exclude one another.

Occasionally, lexical and structural focus seem to coincide, as is the

case in our example with unlösbar/insurmountable. Structurally, the focus

Questions of order

exponent is assigned to problem, but the lexically marked unlösbar/
insurmountable is focused, too. And as the adjective carries a contrastive
focus, it seems to be even more important than the nominal head of the
subject:

Allerdings stand diesem Vorhaben ein schier

UNLÖSBARES

ROBLEM

Wege.

An apparently

INSURMOUNTABLE

PROBLEM

stood in the way of this effort.

We will see many more examples of multiple foci and also non-lexical means
of marking them. But before we move on, let us once more recall the major
lines of reasoning.

With the necessary contextual constraints on our common basis of

(implicit) knowledge, we share intuitions on the contextual appropriateness
of systematically varied paraphrases. This depends upon the grammar of a
language but also upon associated language-specific conditions for optimal
processing. Major conditions for optimal processing are set by grammatical
parameters like the directionality parameter (left-/right-directedness of
verbal extension) or the configurationality parameter (for example, free vs.
preverbal position of subject). Optimal conditions reduce undecidedness and
ambiguities during incremental processing. This also applies to focus inter-
pretation. As for the latter, we can distinguish contextually determined,
pragmatic focus and linguistically determined focus, either lexically bound,
contrastive focus or grammatically determined, ‘normal’ focus. The latter is
bound to structural positions and projects from a focus exponent over other
elements according to the contextually determined focus interpretation. In a
paraphrase that meets optimal processing conditions, structural and
contextual focus interpretations match each other. And this is one of the
challenges in translation: if contextual and structural focus interpretations
diverge in analogous translations between languages with alternatively set
parameters, translations will have to be restructured so as to meet optimal
processing conditions in the target language.

Complex sentences

So far, we have looked at very few and very simple examples, but have
introduced a number of basic concepts which ought to apply to any example
we might encounter, including more complex sentences. Focus theories,
however, which we were able to exploit for translation problems of simple
sentences, do not yet reach beyond clause boundaries.

Yet if we study more complex examples, we can recognize the same

patterns behind the structural changes of the translations. Although each of
the examples has its individual properties, there are clearly parallels between
the translation of simple and complex sentences in terms of information
structure.

Sentences can be complex in various ways: a verb can have one argument,

or two or three arguments, a verb phrase can be modified by adverbials, one
or several, a noun phrase can consist of nothing but a pronoun, or a nominal
head with pre- and postnominal extensions: arguments and attributes, one
or several, and so on and so forth – all variations in phrases. But sentences
can also be complex in that they contain two or more clauses, including
clause-like structures with infinitives, participles, gerunds, that is, non-finite
phrases.

It is the clausal aspect of complexity in which we are now interested, as we

want to know whether language-specific conditions for focus interpretations
can be extended onto sentences with more than one clause. What can we
expect? Syntactically seen, clauses are sentences inside sentences, and share
most of the properties possessed by autonomous sentences. But as a clause is
part of a larger unit, it has a syntactic function within this unit. It can
be a subject clause, an object clause, an adverbial clause and so on, when it
has the function of subject, object, adverbial and so on in the clause it
belongs to.

Analogously, we can expect a clause to have its own information

structure, with one or even several foci, and at the same time participate in
the information structure of the sentence or clause it belongs to. If there is a
structurally determined focus exponent in simple sentences, can we expect
something similar in complex sentences? If this is the case, can we then
expect something similar to the different focus positions in German and

Complex sentences

English in complex sentences, too? Something which might even lead to a
different order of clauses in translations?

Let us, again, proceed empirically, and look at the translation of complex

sentences from German into English. Whereas, as a rule, most sentences
retain the order of clauses in the translation, the internal make-up of clauses
is changed. And the changes reveal very much the same pattern of language-
specific focus structures as the ones we know already from simple sentences.

A double-storeyed focus

In talking about entropy, Eigen and Winkler argue against the uncritical use
of average values:

In jedem Einzelfall wird man sorgfältig zu prüfen haben, ob bei der
Mittelwertbildung nicht gerade das Wesentliche verlorengeht.

(EW 1983a: 172)

This would be analogously (except for the insertion of process and elements):

In each individual case one will have to determine carefully whether in
the process of averaging the crucial elements are not lost.

However, the existing translation has extended the structure further and
reordered constituents in both clauses:

It is important to determine carefully in each individual case whether the
crucial elements in a problem are not lost in the process of averaging.

(EW 1983b: 149)

In a first comparison, we can, indeed, make out some similarities with
preceding examples. If we look at the German original, we recognize some of
the characteristics of the simple sentence in either clause and in the overall
structure of the complex sentence.

Let us first look at the contextual conditions. Talking about entropy as an

average value (to be arrived at by a certain probabilistic formula), Eigen and
Winkler, or rather, the translator, conclude their presentation of this
expression for entropy with the sentence:

We hope to have shown that its use is appropriate only in cases where
average values convey meaningful information.

and mention some legitimate cases in the following:

This is clearly the case in speaking of the distribution of energy among
the different quantum levels of a molecular system. Here the individual

Complex sentences

distribution is of little interest. The chemist wants to know, for instance,
how much heat is produced in a reaction and what temperature is
optimal for a technological process. A knowledge of entropy is equally
useful in various aspects of communication science and technology.

The sentence under discussion opens the next paragraph, which turns the
discourse towards cases like information aesthetics, where the mere average
does not suffice and much more detailed statements about distribution are
necessary.

Against this background, it is clear that the more important informational

elements are localized at the end of either clause: sorgfältig prüfen, das
Wesentliche verloren. The initial elements in jedem Einzelfalle, bei der
Mittelwertbildung are less important or even given. But as we have a
complex sentence, the information structures of the individual clauses are
integrated into the information structure of the complex sentence as a whole.
Are there any clues in the context as to which clause is to be interpreted as
more important?

As the preceding sentences elaborate the positive cases for the use of

average values, the negative possibility of the interrogative clause establishes
a contrast with the preceding passage, extending the discourse from the
appropriate cases onto the inappropriate ones. The idea of losing the crucial
elements introduces the subsequent topic of discourse and can thus be con-
sidered as most important in the sense of discourse progress. Contextually
seen, we could say that the focus of the complement clause serves simul-
taneously as focus of the whole sentence. As the matrix clause and the
complement clause each have their foci at the right periphery, the focus
structure of either clause and of the whole sentence correspond to our struc-
tural expectations of end-focus in German. But let us look at the structural
focus conditions more closely.

Scrambling

The contextual analysis confirms what we can read off from the linguistic
form of the sentence. The complement clause indicates the focus exponent,
das Wesentliche, structurally, by its local adjacency to the verb, as well as
lexically, by the focusing adverb, gerade:

ob bei der Mittelwertbildung nicht gerade das W

ESENTLICHE

ver-

lorengeht.

The use of an additional, lexical focus marker is especially welcome as the
definite form of the closest argument to the verb may otherwise tempt us into
restricting the focus to the verb:

ob bei der Mittelwertbildung nicht das Wesentliche [

VERLORENGEHT

Complex sentences

What you may not have noticed is that in the original the adverbial has been
moved (technically, ‘scrambled’) out of its basic position:

ob bei der Mittelwertbildung nicht das Wesentliche _ verlorengeht.

This leaves the verb-adjacent focus position to the subject. Thus, in addition
to the focusing adverb, it is the extra order which contributes to a more
efficient processing. We could also recognize the focus on the subject after
the focusing adverb in a sentence with basic word order:

ob nicht gerade das W

ESENTLICHE

bei der Mittelwertbildung ver-

lorengeht.

But we might take everything else, including verlorengeht, as given, and
interpret das Wesentliche as narrow focus:

ob nicht gerade das [W

ESENTLICHE

] bei der Mittelwertbildung ver-

lorengeht.

We could exclude this interpretation afterwards, by accessing our contextual
knowledge, thereby investing additional processing effort. But the order of
the original:

ob bei der Mittelwertbildung [nicht gerade das W

ESENTLICHE

ver-

lorengeht].

with the focus exponent in its neutral, projecting position at the end of the
sentence, secures the correct interpretation right away.

The distribution of information in the German complement clause is

optimal. But as this is a complex sentence, we have to look at the other
clause, too, before we are really able to assess the translation. So what can
we say about focus interpretation in the German main clause?

Focus spacing

In the main clause:

In jedem Einzelfalle wird man sorgfältig zu prüfen haben . . .

there are two parts which are focused: in jedem E

INZELFALLE

, sorgfältig

PRÜFEN

. Let me begin with the predicate. There is no argument of the verb,

thus it is the main verb itself in its basic position, at the end of the main
clause, which will be identified as focus exponent of the main clause:

wird man sorgfältig zu

PRÜFEN

haben . . .

Complex sentences

As sorgfältig is not given, focus projection will include it (but not man, which
is generally given by the very nature of its meaning):

wird man [sorgfältig zu

PRÜFEN

haben] . . .

But if man is given and is the subject of the main clause, why does the original
not begin with man? After all, the basic word order would localize the
subject before the adverbial:

Man wird in jedem Einzelfalle sorgfältig zu prüfen haben . . .

The answer is that in jedem Einzelfalle extends the cases mentioned before
onto all possible cases, that is, it expresses a topic shift which extends the
discourse topic onto all elements that could be subjected to the process of
averaging. In this sense, Einzelfalle introduces partly new information which
means it is also focused, contextually. But is it also focused structurally?

If the extended predicate is immediately preceded by in jedem Einzelfalle,

that is, the prepositional adverbial is used in its basic position, things are
unclear. In this position, after the finite verb and with no other nominal
phrase following, the focus on the prepositional phrase and thus the topic
shift may be missed altogether:

Man wird in jedem Einzelfalle [sorgfältig zu

PRÜFEN

haben] . . .

Processing would clearly be hampered, even if we made the right decision
afterwards, recognizing that there are two foci in the sentence, namely on the
prepositional phrase and on the predicate.

In general, we can say that if there is more than one focus in a sentence,

easy processing requires us to make each of them visible. Two foci following
each other may be difficult to delimit as we could always content ourselves
with one focus and defocus the other one.

Obviously, the structure of the original helps to avoid the undecidedness.

We can see that the main clause contains two foci, which are distributed
evenly around less relevant information:

[In jedem E

INZELFALLE

] wird man [sorgfältig zu

PRÜFEN

haben] . . .

Preposing the prepositional adverbial to the beginning of the sentence marks
the adverbial as focused and, by separating the two foci of the main clause,
helps us to identify them both directly. We will call this stylistic device ‘focus
spacing’.

Focus spacing is also welcome in simple sentences, but particularly helpful

in complex ones, where reanalysis may be more costly or not possible at all.
Thus, even the overall structure of our complex example profits from the
balance in the main clause. If we were to miss the focus on the final element

Complex sentences

in the main clause, we would process the complement clause with the faint
feeling that something had gone wrong, perhaps without ever getting a
second chance to correct ourselves.

Altogether, the German original appears to be quite optimal in terms of

processing conditions:

[In jedem E

INZELFALLE

] wird man [sorgfältig zu

PRÜFEN

haben], ob bei der

Mittelwertbildung [nicht gerade das W

ESENTLICHE

verlorengeht].

What can we say about the English translation?

Dummies

Let us look at the analogous version first:

In each individual case one has to determine carefully whether in the
process of averaging the crucial elements are not lost.

The adverbial in the second clause, in the process of averaging, is squeezed in
between the connector whether and the subject:

whether in the process of averaging the crucial elements are not lost.

We would not greatly appreciate the adverbial before the English subject if it
initiated an independent sentence:

In the process of averaging, the crucial elements are not lost.

but with whether preceding, it is even worse.

Everything is fine, however, if we follow the existing translation and ‘un-

scramble’ the adverbial, that is, place it at the right periphery of the sentence,
where it has its basic position in the English sentence structure anyway:

whether the crucial elements are not lost in the process of averaging.

As the focus is locally associated with the left-peripheral verb, we are
prepared to find less important elements at the end of English sentences. The
definite adverbial itself suggests givenness, which is, as we know, in line with
what the context says. As the subject is definite, too, we seem to be left with
the negated predicate as focus exponent and focus projection:

whether the crucial elements are [not

LOST

] in the process of averaging.

But now the interpretations of the German and English complement clauses
differ. In German, the focusing adverb suggested das Wesentliche as focus
exponent of its clause with the focus projection including verlorengeht:

Complex sentences

ob bei der Mittelwertbildung [nicht gerade das W

ESENTLICHE

ver-

lorengeht].

As this interpretation was in line with our contextual analysis, something
must be wrong with an interpretation where the focus is restricted to lost.

Das Wesentliche has become part of the English subject, where it is tucked

away as prenominal adjective. If we take a second look at the existing
translation, we notice that the structure of the English subject is quite
complex, with pre- and postnominal modifiers: crucial elements in a
problem. In fact, the English subject is much more explicit than the German
subject, das Wesentliche. Some of the structure has been added for
grammatical reasons as English adjectives cannot be nominalized as freely as
German adjectives. If it is to function as a nominal phrase, the crucial needs
some dummy noun, as for example element or problem, after it.

But the crucial elements in a problem adds more dummies than are needed

for grammatical reasons. The second extension does not add much to the
message, either – in a context of scientific reasoning problems are no less
given than elements. In one respect, however, the extension opens a new
possibility. The attribute of the subject is indefinite. Thus it localizes the
crucial elements in a new information element which can be focused. But
as crucial is the only really meaningful element of the subject, all the
dummies will be defocused and the focus eventually shifted onto the
adjective: the

CRUCIAL

elements in a problem.

The processor can, as it were, identify the focused element only back-

wards when it attaches the indefinite prepositional phrase, in a problem, to
its head the crucial elements. But the mechanism is basically the same as the
one needed for the predicate where we have to subtract the definite
adverbial, in the process of averaging, from the focus projection. Both cases
are instances of defocusing strategies. While defocusing at the end of the
sentence may here be triggered by the definite article, defocusing at the
beginning could be triggered by the dummy nouns: problem, elements.

If this analysis is correct, the structural properties of the existing

translation can help us to identify two foci in the English complement clause,
crucial and lost. The corresponding two elements of the German original are
adjacent to each other, which means that they can be covered by one focus
projection:

ob bei der Mittelwertbildung [nicht gerade das W

ESENTLICHE

ver-

lorengeht].

But with the subject-before-verb condition in English, the subject and
predicate become separated and need an extra focus each, which amounts to
a ‘discontinuous’ focus:

whether the

CRUCIAL

elements in a problem are not

LOST

in the process of

averaging.

Complex sentences

Summarizing the discussion of the complement clause, we could say that:

insertion of a focusing element (gerade) and scrambling secure easy focus
interpretation in German, while insertion of an indefinite dummy noun and a
return to basic word order serves to improve processing conditions for focus
interpretation in English.

A difference in initial positions

The English main clause, too, postposes the adverbial and rephrases the
subject, together with the modal aspect of the predicate, in a more explicit
structure. If we identify the last stressable element, case, as focus exponent,
focus projection will include the entire clause except the pronominal
dummy, the expletive it:

it [is important to determine carefully in each individual

CASE

] . . .

But could we not stay closer to the structure of the original? After all, the
analogous version of the main clause would not be as bad as that of the
complement clause. The initial adverbial is not squeezed in between a
connector and a fully lexicalized subject. It just precedes a pronoun:

In each individual case one will have to determine carefully . . .

Basically, the analogous version does not present any processing difficulties
and at first sight its focus interpretation seems to be the same as that of the
original:

In each individual

CASE

, one will have to determine

CAREFULLY

. . .

The first focus is due to the fact that the adverbial has been moved out of its
basic position to the front of the sentence, which is, in English, outside the
sentence in its stricter sense – it forms a constituent of its own, separate from
the rest of the sentence. Marking an element as focused by moving it out of
its basic position is a possibility in both English and German, but the details
of this process differ as the structural characteristics of the elements involved
differ. (We will look at one of these differences directly.) The second focus is
the structural focus on the verb-adjacent extension. Except for the final focus
exponent sorgfältig prüfen (determine carefully), this is exactly the focus
structure of the original:

[In jedem E

INZELFALLE

] wird man [sorgfältig zu

PRÜFEN

haben] . . .

So why did the translator not choose this analogous version?

Neither the syntactic nor the information structures of the original and the

translation are really equivalent despite their similar surface. We have, by

Complex sentences

now, seen enough empirical evidence for a grammatically based difference
in the initial position of English and German sentences. While the initial
adverbial in German is unmistakably part of the clause whose finite verb it
precedes, the initial adverbial in English belongs to the complex sentence as
a whole.

Thus:

In jedem Einzelfalle wird man sorgfältig zu prüfen haben, ob bei der
Mittelwertbildung nicht gerade das Wesentliche verloren geht.

localizes the initial adverbial in the first clause, while:

In each individual case, one will have to determine carefully whether the
crucial elements in a problem are not lost in the process of averaging.

adjoins the adverbial to the sentence as a whole. That is, we have to process
the whole sentence in English before we know what the initial adverbial
refers to.

Moreover, preposing the adverbial has a focusing effect in German and

English, though the English adverbial is marked more strongly than the
German adverbial. This can be seen as a result of the alternative direction-
ality of English and German verb phrases, which orders adverbials and
arguments differently. The basic position of this type of adverbial is before
the object in German and after the object in English (unless the object is a
clause and therefore extraposed, as is the case here). Fronting of the English
adverbial (not of the German adverbial) changes the basic order of the
internal constituents relative to each other. As the topic position is further
away from the basic position of the English adverbial, processing of the
topicalized adverbial requires a greater effort simply because we have to
store the result for a longer period before we can interpret it in its proper
hierarchical position.

The greater effort would be legitimate if we wanted to emphasize the

adverbial more strongly, for example, to contrast it. But although in each
individual case is contextually focused, it is not contrasted. Thus, preposing
of the English adverbial does not match the discourse conditions and we
have to find a version which is less marked.

Another type of focus spacing

By placing the adverbial after the first clause, we enable the processor to
analyse the adverbial relative to the first clause:

One will have to determine carefully in each individual case whether
the crucial elements in a problem are not lost in the process of aver-
aging.

Complex sentences

This is, obviously, a processing advantage. But it sacrifices processing ease in
terms of focus identification as it lumps the two foci (carefully, in each
individual case) together.

Now that we have come quite close to the existing translation, improving

parsing and focus identification in accordance with the specific English
conditions, we are ready to admit that the translator’s idea of rephrasing the
weak ‘onset’: one has to as it is important to merits some extra applause. The
extension by another predicate introduces the option of a second focus in the
first clause, making it in a way parallel to the second clause with its two foci.
Without the extension, optimality will make us expect only one focus at the
end of the main clause and thus content ourselves with a focused

CAREFULLY

or a focused

CASE

. But with the extension, we get a complex clause with an

embedded information structure, interpreting

CAREFULLY

as focus exponent

of determine and

CASE

as focus exponent of the main clause predicate, it is

important . . .:

It [is important to [determine

CAREFULLY

] in each individual

CASE

]. . .

which is structurally just as nicely balanced as the original, though less
concise:

In jedem E

INZELFALLE

wird man sorgfältig zu

PRÜFEN

haben . . .

The structural extension is just one of many similar (sometimes even
grammaticalized) cases. We will take them up more systematically later on.

The fact that the focus exponents are not exactly identical need not bother

us. Equivalent information structures – that is, information structures that
present the same elements as focused and the same elements as ‘unfocused’
(background) – will be reached via focus projection, which includes the
element identified as focus exponent in the other language: sorgfältig zu

PRÜFEN

/determine

CAREFULLY

; W

ESENTLICHE

verloren/

LOST

If we compare the original and its translation as a whole, we can confirm

the initial impression that the differences between English and German
information structures in simple sentences recur in the clauses and in the
overall structure of complex sentences. German clauses and sentences end
on focused constituents; other constituents, also focused ones, are moved
to the left, including the left periphery. Thus, the complex sentence of the
original starts with a focused topic and ends on the major focus of the
sentence:

[In jedem E

INZELFALLE

] wird man [sorgfältig zu

PRÜFEN

haben], ob bei der

Mittelwertbildung [nicht gerade das W

ESENTLICHE

verlorengeht].

In English, we get the verb-adjacent focus early in the clause, and other foci
may be indicated by an additional degree of structural explicitness:

Complex sentences

It [is important to [determine

CAREFULLY

] in each individual

CASE

whether [the

CRUCIAL

elements in a problem] are [not

LOST

] in the process

of averaging.

In both languages optimal processing promotes focus spacing, that is, the use
of extra means to avoid focus ‘clusters’. In line with the language-specific
conditions, these means are different, though they are taken from a universal
repertoire, from which reordering and structural extensions are major
candidates.

Generalizing the findings

As the empirical approach presented individual trees, the overall view of the
wood may have remained unclear. Before we face the next, even more
complicated case let me summarize the findings in a more systematic way.
There are two major lines of reasoning we follow in studying felicitous
translations from the perspective of language processing in discourse. The
first line considers information structure, in particular focus assignment
relative to the discourse the sentence belongs to. The second line concen-
trates on the linguistic, and thus language-specific means, of expressing
focus. Let us look at the basic assumptions about discursive and linguistic
focus successively.

For the time being, we identify focused elements with contextually new

elements but also admit partly new elements – as in our last example, where
in each individual case merely extended the number of cases talked about.
The discussion of the last example shows also that we can decide about the
status ‘new’ or ‘given’ only if we understand the relations between the
sentence and its preceding context in all relevant aspects. This will as a rule
involve more than just the immediately preceding sentence, and it will
depend heavily upon the inferences we can draw from any of the things said
explicitly. In our last example we had to go back five sentences to determine
the discourse relations between the sentence under discussion and its
context. It was the first of these sentences which set the discourse theme to be
elaborated by the following passages, including our example. And it was the
relation to the preceding four sentences that told us that we were turning
from some cases where average values do convey meaningful information to
the possibility of cases where average values do not convey meaningful
information. But we needed a lot of inferring to get from the coarse-grained
discourse structure to the detailed focus structure of our sentence. Conveying
meaningful information implies that crucial elements are not lost only if we
know that, first of all, averaging means losing information, and, secondly,
averaging is meaningful only if it does not cause crucial information to be
lost. Luckily, we have a strong intuition about contextual focus assignment
in general, however difficult the inferencing process may be.

To a certain extent our intuition is also guided by the information

Complex sentences

structuring, focusing devices of the linguistic forms. But the linguistic forms
do not fully match the contextually determined focus structure, besides being
vague or even ambiguous. Nevertheless, although English and German
present different conditions for the linguistic encoding of focus, the
mechanisms of assigning structural focus positionally, and of marking focus
by changing positions, are to some extent similar. And the final strategy of
adapting the linguistic focus structure to the discursively determined focus
by defocusing and shifting the focus onto the proper element is similar, too.
Yet the resulting structures can be rather different in their overall
appearance due to lexical properties (including selection restrictions as in
our first example) and syntactic properties (concerning subcategorization
features as jemandem im Wege stehen/to stand in the way of someone),
especially the major grammatical parameters distinguishing German and
English word order and its variability. With the greater variability of the
German word order reordering makes focus structures more visible, while
the greater rigidity of the English word order allows for greater vagueness of
focus structures (to be disambiguated contextually), and promotes a greater
degree of structural explicitness to help identify additional foci.

Most of the differences between the original and the target version were

clause-internal, that is, they could have occurred in a simple sentence, too,
but there are cases where informational elements are also distributed
differently across clauses. It is to such an example that we will now turn.

Cross-clausal

With all the different conditions for optimal processing of German and
English sentences, a target version may deviate considerably from the
original. However, if we add yet another observation to the preceding
assumptions, we can even make out regularities in the following example:

Erst mit der Vervollkommnung des Übersetzungsmechanismus konnte
es mehr und mehr zu einer klaren Rollenaufteilung zwischen Geno- und
Phänotypus kommen.

(EW 1983a: 309f)

While the analogous version would be something like:

Only with the perfection of the translation mechanism could it come –
more and more – to a clear role division between genotype and phenotype.

the existing translation says:

The gradual clarification of role division between genotype and pheno-
type could come about only as the translation mechanism was perfected.

(EW 1983b: 276)

Complex sentences

What we can recognize straight away is the initial position of the English

subject, which has obviously swapped position with the adverbial. But the
adverbial has not only been placed at the end of the English sentence, it has
also been rephrased as a clause. And the subject, which was taken from what
was a prepositional object in the original, has undergone some restructuring
of its internal hierarchy as well as having the original adverbial of degree
incorporated as an adjective. Though the analogous version is, no doubt, far
from optimal, the existing translation seems to be unnecessarily distant from
the original sentence structure. Judging by the Maxim, has the translator not
taken too many liberties?

Let us see if we could make do with less. Starting with the analogous

version, there are some lexically based processing problems we will first try
to get rid of. The adverbial more and more is inserted as a parenthesis, which
requires an extra amount of processing effort. We could put its idea as
eventual, gradual into a more natural position:

Only with the perfection of the translation mechanism could it gradually
come to a clear role division between genotype and phenotype.

Yet there is still something odd in the combination of it comes to a clear
division and the modifier gradually. It could come to a clear role division or
to a gradual role division, that is we could have one modifier or the other, but
both modifiers cause confusion. The oddity disappears if I make use of the
paraphrase of the existing translation:

It comes to a gradual clarification of role division between . . .

It may be interesting to note that the German original does make use of the

two modifiers in just this configuration without any difficulties. Since both
aspects have to be expressed in an optimal translation as neither can be
recovered from the context, the categorial change from adjective to noun
(clear/clarification) and consequently from adverb to adjective (gradually/
gradual) offers a way out. We will register the difference as one of the many
idiosyncratic combinatorial differences between the lexical items of English
and German, and return to the parametrized grammatical differences we are
already familiar with.

We have definitely made some progress now:

Only with the perfection of the translation mechanism could it come to a
gradual clarification of role division between genotype and phenotype.

Yet the existing translation reverses the order of the original. Thus, the
question is again: which paraphrase is more optimal? As questions of order
are likely to relate to aspects of information structure, we should, again, turn
to the context to see what the author must have meant, and, consequently,

Complex sentences

what focus structure needs to be expressed by the translation in line with the
Maxim.

Degrees of accessibility

Now, if we look for the two key words translation mechanism and role
division, we find a reference to translation eight sentences back, and a
reference to role division twelve sentences back. We have to read through
almost the entire article to understand the concept of translation in genetics,
but the sentence about role division provides much of the background for
our example. It says:

The division of roles between phenotype and genotype was evidently not
so clearly fixed in the early stages of life’s development as it is now in the
present-day products of evolution.

(EW 1983b: 276)

The paragraph separating this sentence from our example considers the
possibility of a reverse translation, which is said to be highly unlikely. But
while the transfer of information normally moves from DNA to RNA to
protein, experiments have shown ‘that limited amounts of information can
also be made available by proteins. In the early periods of evolution this
possibility . . . must have been of great importance’. In the early periods of
evolution, it seems, then, a limited amount of reverse translation was
possible, genotype and phenotype not yet clearly divided. But now that the
translation mechanism was perfected by evolution it has become irreversible,
genotype and phenotype clearly fixed.

If we apply our criterion of givenness to the sentence, we can say that role

division as well as translation mechanism are given. From our general
knowledge about evolution we can also conclude that the translation
mechanism was perfected, and the role division eventually fixed. That is,
directly and indirectly, we have access to all informational elements of the
sentence.

If we take an even closer look at the context, we can make out some

difference between the given elements. While all these elements are given,
their contextual antecedents are not equally accessible. If we go by local
distance in the text, the previous reference to role division lies further back
than the previous reference to translation. While the latter remains a
discourse topic throughout the preceding paragraph, which considers the
possibility of reversed translation, the difference between genotype and
phenotype is returned to only in our example. If we measure the novelty of
an element by the cognitive effort needed to retrieve it from memory, or by
the inferential steps needed to derive it from previous knowledge, then
translation mechanism is easier to access than any other information in
this sentence.

Complex sentences

But then Übersetzungsmechanismus/translation mechanism does not

appear alone. It is part of an adverbial phrase whose head implies a contrast-
ive relation: Vervollkommnung des Übersetzungsmechanismus/perfection
of translation mechanism refers to the normal direction of informational
transfer as opposed to that of a reverse translation, which was possible
before the translation mechanism was perfected.

Thus, Vervollkommnung carries a contrastive focus and is anchored to

the immediately preceding context by its argument: translation mechanism.
It provides a canonical starting point for a German sentence.

The different ways to perfection

What about the appropriateness of the linguistic encoding? While the
prepositional object zu einer klaren Rollenaufteilung zwischen Geno- und
Phänotypus is the closest extension of the verb kommen and will therefore
automatically contain the focus exponent of the sentence, the definiteness of
the initial phrase mit der Vervollkommnung/with the perfection does not
allow us to recognize the initial focus at once. We would have to figure it all
out contextually – if it were not for the focusing element, erst/only.

In this way, the German original has two foci, one of which is indicated by

its verb-adjacent position, supported by the intensifying adverbial, mehr und
mehr, while the other is marked by the focusing particle erst:

Erst mit der V

ERVOLLKOMMNUNG

des Übersetzungsmechanismus konnte

es mehr und mehr zu einer klaren R

OLLENAUFTEILUNG

zwischen Geno-

und Phänotypus kommen.

In English the way to perfection is different. That is, an analogous

translation means different processing conditions. Introduced by only, the
adverbial at the beginning of the sentence belongs to that class of elements
that allow or rather require subject–verb inversion. In other words, it is one
of those rare cases where we do not expect the English subject to precede the
finite verb. Yet the inversion of the basic order affects the focus inter-
pretation of the sentence. It indicates that the focus associated with only –
the focus exponent of the adverbial phrase associated with only – is the focus
exponent of the sentence. Everything else is given information.

[Only with the perfection of the

TRANSLATION

mechanism] could it come

to a gradual clarification of role division between genotype and
phenotype.

But this is contextually inappropriate and not equivalent to the original.
Embedding the processing result into the context, we may be able to correct
ourselves afterwards, but considering the complexity of the text, we would
certainly welcome a contextually appropriate encoding right away. (This

Complex sentences

argument would also apply if I were to assume that the backgrounding effect
is only one of two or more possible focus interpretations – that is, that the
information structure of the sentence is ambiguous and I would have to access
the context to decide which of the possible readings is the appropriate one.)

The fact that the initial position of the adverbial is subject to different

processing conditions in English and German is a recurrent theme in our
discussion. But the preceding examples were only bound to the subject-
before-verb constraint in English, which pushes the topicalized adverbial
into a position before the sentence proper, requiring additional processing
effort. This time, the initial adverbial is not only marked by its topical-
ization, but also by the focusing adverb only and the following subject–
auxiliary inversion. The marking is much stronger than in German, where
there is nothing extraordinary about the topicalization of the adverbial.

If we reverse the order, we get:

It could come to a gradual clarification of role division between
genotype and phenotype only with the perfection of the translation
mechanism.

Or without the placeholder it, with come about instead of come to:

A gradual clarification of role division between genotype and phenotype
could come about only with the perfection of the translation mechanism.

The second version is better. With the alternative predicate, focus need

not any longer be restricted to the focused elements after only. It is now the
subject which is the closest argument of the verb and thus the structural focus
of the sentence. The focus exponent is in its grammatically neutral position,
where it can also project onto other elements. While the context restricts
focus projection to the head of the subject and the prenominal adjective, the
focusing element only marks the second focus of the sentence:

GRADUAL

CLARIFICATION

] of role division between genotype and pheno-

type could come about only [with the perfection of the

TRANSLATION

mechanism].

But the existing translation differs from this version in two more aspects.

First, there is the extension of the nominal form of the adverbial, the
perfection of the translation mechanism, into a clause:

only as the translation mechanism was perfected.

The clause has a processing advantage over the phrase. Instead of having to
decide which of the two definite noun phrases is the focus exponent of only,
we can now identify the focus automatically, and correctly, with the verb.

Complex sentences

Second, the existing translation uses a definite subject instead of the

indefinite original. And it does so with good reason. Within the context of
the modal verb could, the indefinite subject could be interpreted as a future
possibility. This is, as we all know, contrary to facts, at least as long as we
talk about the real world, in which phenotypes and genotypes are clearly
distinguished. Thus, we can appreciate the translator’s move to the definite
subject:

The gradual clarification of role division between genotype and
phenotype . . .

German can afford the indefinite reference as the modal verb konnte refers
unambiguously to the past.

However, by becoming a definite noun phrase, the initial element loses its

focus status:

The gradual clarification of role division between genotype and
phenotype [could come about only as the translation mechanism was

PERFECTED

This looks like an unnecessary difference between original and translation.
But as the idea of a clearly fixed role division was mentioned as much as a
dozen sentences earlier, the difference is negligible. The context allows us to
present the subject as given or as new, as we like, or rather, as is required by
the pertinent conditions for optimal processing.

In the end, we can say that the differences between the original and its

existing translation were basically of the type we could observe in all the
earlier examples. The initial adverbial is partly given and thus chosen as a
focused topic in German. It is replaced in this position by the English subject.
Yet this time, the English subject goes back to a prepositional object in
German. This is a type of functional change we have not yet commented
upon, although it may occur quite frequently in translations between English
and German. And it is linked to the typological parameters determining the
language-specific constraints on the position of adverbials and subjects in
English. Let us now summarize what we know about questions of order so
far, and extend the topic onto those functional aspects afterwards.

Reviewing and resetting the stage

Except for the first example, where almost the same elements appeared in the
initial and final positions of the German and English sentences, suggesting a
similar distribution of topic and focus in both languages, we have looked at
ever-more complex cases of a reversed distribution of initial and final
elements. The inversion was accompanied by a number of other differences,
including structural extensions from phrases into clauses.

Complex sentences

We have examined focus interpretation from two sides: using an ever

finer-grained conception of contextual givenness, and using the theoretical
concept of basic word order, which determines the structural position of the
focus exponent. This position differs in languages with differing direction-
alities.

We had found that the choice of initial elements was guided by the criteria

of givenness or focus spacing in German, but by the grammatically
determined order of subject-before-verb in English. Inversion of initial and
final elements was accompanied by other changes, due to further differences
between the two languages, concerning, for example, the use of definite and
indefinite phrases in various positions.

In all examples, the existing translation could be shown to be the preferred

paraphrase of a set of systematically varied paraphrases. It was always the
target version which succeeded in securing optimal processing conditions in
English while staying as close to the original as possible. (This was a
welcome feature of the data selected. Not all existing translations can be said
to serve this purpose, for whatever reason.) All the changes we noted
between the German original and its English translations were licensed by
grammatical or grammatically based differences between the two languages.
The most important differences were related to the grammatical parameters
of directionality (left-/right-peripheral direction of verbal extensions in
German and English respectively) and configurationality (especially,
subject-before-verb in English).

The two complex sentences we looked at in the end demonstrated these

parametrized differences within and across clauses. The German clauses/
sentences began with adverbials (in jedem Einzelfalle; bei der Mittel-
wertbildung; mit der Vervollkommnung) which had been topicalized (that
is, moved out of their basic positions within the verb phrase into the initial
position of the sentence) in the interest of a clear focus structure, separating
the focus in the comment (sorgfältig zu prüfen; das Wesentliche verloren; zu
einer klaren Rollenaufteilung) from the focus in the topic (in jedem
Einzelfalle; mit der Vervollkommnung). In line with its constraints on
topicalization, English retained the corresponding adverbials (in each
individual case; in the process of averaging; as the translation mechanism
was perfected) in their basic, right-peripheral position at the end of the verb
phrase, but added more structural weight to the focus that would otherwise
be missed (it is important . . . in each individual case; the crucial elements in a
problem; as the translation mechanism was perfected).

The surface of sentences is linear but the blueprint behind it is hierarchical

and multi-layered, so that a difference in the linear order of the original and
its translation may have many facets. Although we have looked through a
handful of examples at great length, the analyses were by no means
exhaustive. This may be a discouraging statement for the exhausted readers,
who have fought their way through long and complicated arguments on
information structure and parametrized processing conditions, structural

Complex sentences

and contextual focus, basic verb positions, hierarchy and directionality of
verb extensions, narrow and wide focus interpretations, marked word order
and focusing particles, focus spacing and dummies, graded accessibility and
idiosyncratic selection restrictions. But all these theoretical concepts suffer
from two shortcomings: they are still much too fuzzy and they are not even
sufficient to cover all aspects of the few examples we have been discussing.

Walking the tightrope as a translator or interpreter, you may think, is bad

enough, but reflecting upon the conditions of optimal translation is walking
the tightrope as a centipede and thinking about the optimal use of each leg.
Yet in this respect, thinking about translating and interpreting is no different
from thinking about anything else we can perform optimally or not. The
curiosity that has led you so far might help you through the next scene
as well.

We may refer to things in this order or in that without changing any of the

other aspects of the referents. But this has to be considered the exception.
None of the examples we have analysed at some depth were restricted to
word order changes, but all of them involved changes in their grammatical
relations, that is, in the syntactic functions which are carried by the con-
stituents of a sentence.

In some cases the changes were very subtle, like that between dative object

and prepositional complement – diesem Vorhaben im Wege stehen/stand in
the way of this effort – or they were structurally particularly complex, as
with the clausal extension of the modal expression man wird zu prüfen
haben/ it is important to determine. But a good deal of the reframing patterns
can be summarized under one heading: perspective. This is closely related to
the preceding topic of word order and presents another classical example of
general transfer patterns based on parametrized processing conditions,
which we will now address.

In favour of primary relations

In several of our examples, the English subject was not only moved into its
proper preverbal position, but was recruited from some structural part of the
German original that had a different grammatical function. The last case of
such ‘reframing’ just occurred: the German placeholder es was dropped, or
rather replaced by a subject that was an obligatory adverbial in the original:
(kommt es . . .) zu einer klaren Rollenaufteilung . . . /the . . . clarification of
role division came about . . .

What is the reason for the adverbial-to-subject switch? What is its relation

to the parametrized processing conditions of German/English we have noted
so far?

Alternative perspectives

Das Wesen des Gleichgewichts wird von jedermann intuitiv begriffen.

(EW 1983a: 176)

says the original, and the translator turns it into:

Everyone intuitively understands the nature of equivalence.

(EW 1983b: 152)

The translation not only arranges the phrases of the original into a different
order, it projects the informational elements of the statement onto different
grammatical relations. The subject of the German original has become the
object of the English translation, while the subject of the English translation
goes back to an adverbial phrase with von in German. The shift in
perspective is due to the verb form: passive in the original, active in the
translation.

We could have had a passive form in English, too:

The nature of equivalence is intuitively understood by everyone.

or an active form in German:

In favour of primary relations

Jeder begreift intuitiv das Wesen des Gleichgewichts.

Yet either version would be considered less good than the ones we started
with.

To the extent that the change in perspective means a change in order, the

difference may well involve aspects of information structure. However, there
are also examples where only the grammatical relations change, while the
order of the original remains.

Concluding his description of a portrait of Cardinal Jules Mazerin, Horst

Bredekamp begins his last passage about ‘the collector as Prometheus’ with
the sentence:

Damit ergibt sich ein Zusammenhang.

(Bredekamp 1993: 33)

which is translated into English as:

This forms a framework.

(Bredekamp 1995: 279)

Here, too, we could make use of the other perspective:

This way, a framework emerges.

or in German:

Dies ergibt einen Zusammenhang.

But, again, the alternative perspective would be less good in either language.

Still, there are enough translations like the following which retain the

perspectives of the original, whatever else they may change:

Für den Molekularbiologen unserer Zeit ist die identische Reproduktion
einer Gestalt zweifellos das größere Wunder als ihre gelegentliche
Metamorphose

(EW 1983a: 96)

For the working molecular biologist today, the identical reproduction
of an organism is no doubt a greater miracle than the occasional
metamorphosis of one.

(EW 1983b: 77)

So, the question is, when and why do we prefer alternative perspectives?

As the examples illustrated, languages have different means for the

expression of perspective. The most obvious device is the grammatical

In favour of primary relations

alternative between active and passive sentences, but there is also the
different perspective of transitive verbs vs. intransitive or reflexive verbs. By
choosing the one or the other perspective, we change the way in which the
participants in an event are translated into syntactic functions.

Seen through the medium of language, we can classify the concrete or

abstract entities of our mental models of the world according to the roles
they may play in the states of affairs talked about. Someone or something
could have the role of the agent in an event, or that of a patient, of a goal, a
place, an instrument, etc. The inventory of roles is mainly determined by the
syntactic functions we can distinguish according to the case frames of
predicates, whether these are determined morphologically – that is, by word
forms – or structurally, by certain hierarchical configurations. Altogether,
there are not too many cases to be distinguished since the formal repertoire
of languages is limited.

Although the number and type of roles distinguished varies slightly with

the theoretical approach taken, the need to distinguish between roles and
functions is not disputed. How else should we describe the shifts in
perspective illustrated above?

The idea of a shift in perspective presupposes something like an unshifted,

canonical perspective in relation to which another perspective can be viewed
as shifted. The assumption is that certain configurations are prototypical
and more basic than others. Take for example, the link between agents and
subjects as opposed to that between patients and subjects: the concept of an
active individual causing something to happen is more likely to be gram-
matically encoded as subject of a sentence than the element that is subjected
to the act. The two roles are sometimes referred to as ‘logical subject’ and
‘logical object’ as opposed to syntactic subject and syntactic object.

According to their positions in the syntactic structure of a sentence,

syntactic functions form a hierarchy in which the subject has the highest
position. Objects have a lower position, and free adverbials still lower.
Similarly, semantic roles are assumed to form a hierarchy in which agents
have priority over patients, and patients are still higher than, for example,
instruments.

The normal perspective would thus be a case in which the hierarchical

position of a semantic role corresponds to the hierarchical position of the
syntactic function it is linked to. A shift in perspective is achieved by
projecting a lower semantic role into a higher syntactic function. This
happens visibly when we passivize the sentence, or, in a covert and less
noticeable way, when we use certain classes of verbs that require a passive-
like link between the semantic roles of their arguments and their syntactic
realizations. ‘Emerge’, for example, assigns the role of a patient to the
subject. There are many verbs that can be used transitively and intran-
sitively, that means with both perspectives, as in: ‘I can open the door’, or
‘the door opens’.

Languages may differ in their means of perspective. First of all, their

In favour of primary relations

vocabularies differ: there may not be the same verb, or the same verb may
have a different range of options. Second, they may not have the same
grammatical device, or the same device may be used differently. Thus
German has a type of reflexive verbs, as for example sich ergeben aus/to
result from, that does not exist in English. Or there are some uses of the
passive in English which are not possible in German, and vice versa.

The translational problem in these cases is to find an equivalent

expression – provided we want to retain the perspective of the original. But
as the examples show, this may sometimes not be the case even if the same
means of expressing perspective are available. There are, again, different
preferences we can observe in the language-specific use of perspectives, and
they, too, can be shown to result from the grammatically parametrized
conditions for easy processing.

Here, the conditions for easy focus identification will again play a decisive

role for the choice of a contextually appropriate perspective. But there is an
even more elementary processing problem, which does not relate to
(sentence-external) context. We will turn to this aspect first.

Placing a modifier

Describing the way in which our brains interpret a form as an integrated
whole, Eigen and Winkler write:

Die Konturen einer räumlichen Struktur werden von den Augenlinsen
eingefangen und maßstabsgetreu auf der Netzhaut abgebildet.

(EW 1983a: 88)

The existing translation reframes the passive original as an active sentence.
(It drops the first of the coordinated predicates, eingefangen/caught, which –
for the sake of simplicity – we will assume to be implied.)

The lenses of our eyes transmit the contours of a spatial structure in their
original proportions to the retina.

(EW 1983b: 70)

If we used a passive version instead, we could start the sentence with the
original subject:

The contours of a spatial structure are transmitted . . .

But how should we continue the sentence? In the order of the original, where
the modal modifier precedes the arguments of the verb?

The contours of a spatial structure are transmitted in their original
proportions to the retina by the lenses of our eyes.

In favour of primary relations

Clumsy, you would say. But with the modifier between the arguments, the
translation seems even worse:

The contours of a spatial structure are transmitted to the retina in their
original proportions by the lenses of our eyes.

And with the modifier at the end, we get:

The contours of a spatial structure are transmitted to the retina by the
lenses of our eyes in their original proportions.

which, no doubt, is the worst case as it attributes the original proportions to
the lenses of the eyes. Each version is problematic in some way since it does
not clearly show what belongs where.

There are yet two more grammatical positions for the original propor-

tions. First, the modifier could follow the subject directly:

The contours of a spatial structure in their original proportions are
transmitted to the retina by the lenses of our eyes.

This, too, seems good for some processing confusion. Second, the adverbial
precedes the subject in initial position:

In their original proportions, the contours of a spatial structure are
transmitted to the retina by the lenses of our eyes.

This version is easier to analyse, but it seems to say something else than the
original.

All versions have one deficiency in common – their form is difficult to

analyse or misleads us into a structural analysis that no longer conveys the
message of the original, even if not altogether nonsensical.

The existing translation avoids all these processing problems by shifting

the perspective from passive to active:

The lenses of our eyes transmit the contours of a spatial structure in their
original proportions to the retina.

There is now no structure intervening between their original proportions and
the antecedent of the possessive pronoun their, namely the contours of a
spatial structure and everything else is in its proper, easy-to-process position.
As the correct analysis of the grammatical relations between the constituents
of a sentence is a precondition for the interpretation of the sentence, all the
passive versions are disqualified.

Still, the German original is passive and has none of the English problems.

Everything can be identified directly. Here it is once more (without the first
predicate):

In favour of primary relations

Die Konturen einer räumlichen Struktur werden von den Augenlinsen
maßstabsgetreu auf der Netzhaut abgebildet.

Why should the syntactic analysis of active and passive sentences in English
and German be subject to different conditions? After all, the shift from active
to passive looks quite similar. The object of the active sentence is used as the
subject of the passive sentence, and the original active subject may be added
as an adverbial with by/von to the structural extensions of the verb.

There are two sides to the coin: linguistic and psycholinguistic. Let us

begin with the latter.

The parsing strategy of Minimal Attachment

With each constituent we process we have to decide its place in the syntactic
hierarchy of the sentence – that is, we have to attach the new element to the
constituent we think it belongs to. Psycholinguists have a special term for the
syntactic processing of structures, namely ‘parsing’. The processor is said
to follow certain strategies in parsing. One is the strategy of Minimal
Attachment (introduced into the literature by Lyn Frazier) that makes us
choose the most economical possibility when we add a new element to
preceding ones.

For example, we could attach the modal modifier in:

The lenses of our eyes transmit the contours of a spatial structure in their
original proportions

to the immediately preceding noun phrase:

[[ a spatial structure][ in their original proportions ]].

But the clash in number – singular of structure and plural of their – requires
us to correct the analysis and attach the modifier to the next higher noun
phrase:

[[ the contours of a spatial structure] [in their original proportions ]].

(‘Higher’ refers to the position in the syntactic hierarchy of the sentence
structure; the more constituents a phrase comprises, the higher is its position
as compared to the phrases it consists of.)

Yet we could also parse the structure in another way, taking the verb and

its object together:

[transmit the contours of a spatial structure]

and attach the modifier to this verb phrase:

In favour of primary relations

[[transmit the contours of a spatial structure] [in their original
proportions ]].

The higher the place of attachment is, the more effort we need to access it;
the lower it is, the less energy we need to identify it. The Principle of Minimal
Attachment makes us opt for the least effort possible (even if we have to
correct ourselves afterwards). Thus, we would attach the modal modifier to
the contours of a spatial structure in the active sentence, but to retina or the
eyes in the passive versions:

The lenses of our eyes transmit [the contours of a spatial structure in
their original proportions] to the retina.

The contours of a spatial structure are transmitted [to the retina in their
original proportions] by the lenses of our eyes.

The contours of a spatial structure are transmitted to the retina by the
lenses [of our eyes in their original proportions].

It is only afterwards, when we try to interpret the structure, that we discover
our processing error.

To distinguish between a syntactic analysis and a subsequent

interpretation of its result may seem an artificial distinction that is not very
plausible when we think of the temporal conditions of language processing.
But there is the striking evidence of so-called ‘garden paths’ – that is, the
wrong tracks we cannot help getting onto even if we can afterwards identify
the mistake we have made. Moving from left to right (in English or German),
the parser will very often encounter syntactic ambiguities and, in opting for
one possibility, perhaps, continue on the wrong track, from which it may or
may not be able to get out after a while. For example, we will always fall for
the instrumental reading of:

He smashed the window with his sister.

although we can almost simultaneously see that we need a non-instrumental
reading of the adverbial.

In psycholinguistics, there are two alternative models about the way in

which the various components contributing to the understanding of
linguistic structures work together.

There is the ‘modular’ approach, which,

if taken to its extreme, says that each component works for itself without
ever noticing what the other components do. The only ‘interface’ between,
say, the syntactic and the semantic component is their input and output: the
output of the syntactic analysis is the input for the semantic interpretation.
The alternative view is the ‘interactive’ approach, which says that the
various components are continuously interacting.

In favour of primary relations

There is empirical evidence for both approaches: in particular, immediate

and anticipatory understanding of linguistic structures, on the one side, and
garden path sentences, on the other. Both sides can be reconciled if we
assume that the formal (syntactic) analysis precedes the semantic analysis
only by fractions of seconds.

In:

The woman asked the way laughed.

we mistake asked for the finite verb of the sentence before we are able to
identify it as the passive participle of an attributive phrase and laughed as the
finite verb of the sentence.

If all the information needed for the interpretation of the sentence were

always accessible, we would not get onto the garden path. But as the
occurrence of past participle phrases after subjects is less likely than that of
finite verbs, we go by probability and parse the structure according to the
‘normal’ subject–verb order of an English sentence.

Garden path sentences have been studied for many years in sophisticated

experiments to unravel our processing strategies in detail.

The result is an

elaborate garden path theory, highly relevant for the questions we are
interested in. But theoretical progress on the basis of experiments is slow,
especially if results are falsified in subsequent experiments, and language-
specific conditions for optimal translations involve so many divergent
factors that for the time being we have to skip experiments and trust our
intuition when we try to explain translational data along the lines projected
by promising linguistic and psycholinguistic assumptions. And this is what
we can see in the different pre- and postverbal paraphrases of the example
under discussion.

After the subject or before it

In their original proportions is a free modifier, which can only be attached
tentatively, before the ensuing semantic interpretation of the sentence tells us
where the modifier really belongs. Easy processing of free modifiers is bound
to the close vicinity of the modifier and its head.

Closeness in itself does not yet, however, guarantee easy processing. If we

place the modifier right after the subject, it seems to get attached to the
indefinite noun phrase preceding it:

The contours of a spatial structure in their original proportions are
transmitted to the retina by the lenses of our eyes.

We feel the clash between the singular structure and the plural possessive
strongly, but if we were to reattach the modifier to contours, we would not
be happy, either, as the modifier is an adverbial and should be attached to the

In favour of primary relations

verbal part of the sentence. It is only the close semantic relation between the
contours of a spatial structure and the proportions (of these contours) which
suggests an attributive relation between the modifier and its head.

But while the attribute is part of the grammatical relation of its head, the

adverbial is syntactically autonomous, which means, indeed, that there is a
syntactic function intervening between the subject and the verb. Thus, the
preverbal position of the adverbial violates the subject–verb parameter of
English and, hence, the condition of optimal processing.

There are, as always, exceptions. Some of the so-called sentence

adverbials, like ‘certainly’, which modify the sentence as a whole, can be
used in between subject and verb. And there are the local adverbials that
could be considered a syntactic extension of the preceding noun phrase:

Recent experiments in Ken Robinson’s lab show that . . .

But the modal modifier in our example does not belong to those exceptions
and has to be placed somewhere else.

If we place the modifier at the beginning of the sentence, we produce no

processing difficulty but a different sentence.

In their original proportions, the contours of a spatial structure are
transmitted to the retina by the lenses of our eyes.

The position before the English subject is restricted to adverbials that can be
attached to the whole sentence and do not depend on any detail of the
internal structure of the sentence. These are for example elements expressing
the speaker’s attitudes about the likelihood, desirability etc. of the state of
affairs referred to by the sentence. Other elements like ‘and’, ‘but’, ‘thus’,
‘however’, ‘on the contrary’, etc link the sentence to the preceding context.
The only other adverbials that are quite natural in the initial position are
temporal adverbials.

If adverbials from classes that do not modify the whole sentence, but only

part of it, are used initially, they take on something like a contrastive
meaning. Thus, in a passage on symmetry, the local adverbial:

In modern art, almost all directly perceptible proportion has dis-
appeared.

(EW 1983b: 120)

is used in contrast to the subject of the subsequent sentence:

Architecture . . . represents an exception in still using the cube as a basic
unit.

But the German original of our critical example:

In favour of primary relations

Die Konturen einer räumlichen Struktur werden von den Augenlinsen
eingefangen und maßstabsgetreu auf der Netzhaut abgebildet.

does not show the slightest trace of a contrast on maßstabsgetreu and when
we look at the context, it does not contain anything the original proportions
could contrast with. Thus, the initial position of the modifier is no solution,
either.

Competing for attachment

The attachment problems of the original proportions after the verb in the
passive paraphrases are to some extent due to the fact that the adverb
maßstabsgetreu has been rephrased as a prepositional phrase with a
possessive pronoun, in their original proportions, which the strategy of
Minimal Attachment mistakes for an attribute modifying the preceding
phrase. This does not yield a semantically equivalent paraphrase if the
preceding phrase is not the one the modifier belongs to, which is the case
after the retina, respectively the lenses of the eyes.

But if we place the proportions before the other extensions, directly after

the passive verb, it seems to make attachment difficult for the phrases
following it. We would have no difficulty in parsing the sentence if it had
only one of the three extensions, as for example:

The contours of a spatial structure are transmitted in their original
proportions.

or:

The contours of a spatial structure are transmitted to the retina.

or:

The contours of a spatial structure are transmitted by the lenses of the
eyes.

The moment we have two or even three phrases competing for attachment

to the verb, we have to look deeper into the properties of the phrases that
determine their grammatical relations to each other and to the rest of the
sentence. This is in itself a processing disadvantage, but the passive
perspective seems to make it an insoluble task. While we can perhaps still
cope with the second extension:

The contours of a spatial structure are transmitted in their original
proportions to the retina.

In favour of primary relations

The contours of a spatial structure are transmitted in their original
proportions by the lenses of the eyes.

the third extension is too much:

The contours of a spatial structure are transmitted in their original
proportions to the retina by the lenses of the eyes.

The shift in perspective from passive to active eliminates the parsing
problem; the modifier can be attached to its head right away and without
any disadvantages for the processing of subsequent phrases:

The lenses of our eyes transmit the contours of a spatial structure in their
original proportions to the retina.

But this is the case only in English, not in German, where the passive

version does not present any attachment problems:

Die Konturen einer räumlichen Struktur werden [von den Augenlinsen
[maßstabsgetreu [auf der Netzhaut abgebildet]]].

In German, the right-peripheral verb phrase makes us expect the head of the
modifier after maßstabsgetreu, which is precisely what the order of the origi-
nal offers. But as this could also be the order of things in an active version:

Die Augenlinsen bilden die Konturen einer räumlichen Struktur
maßstabsgetreu auf der Netzhaut ab.

the attachment problem alone does not exhaust the phenomenon of different
perspectives. After all, the authors of the original have opted for a passive
version. Why?

Different perspectives

To focus on the question of perspective, we will now turn to a case without
attachment problems. There was this remarkable sentence about the nature
of equivalence:

Das Wesen des Gleichgewichts wird von jedermann intuitiv begriffen.

and its existing translation:

Everyone intuitively understands the nature of equivalence.

If we compare the active version with the passive version:

In favour of primary relations

The nature of equivalence is intuitively understood by everyone.

we can see that there are no attachment problems in either version. Yet we
would say that the translator was right in choosing the active version in spite
of the passive original. What is it that has been improved by the change in
perspective?

We have not yet looked at the effects alternative perspectives have on

information structure, though we are aware that they can be expected as
long as the shift in perspective means a difference in order, too. The definite
article of the nature of equivalence seems to signal its givenness. A look at the
context confirms the assumption: equivalence has been the discourse topic
for at least one paragraph. Thus we ought to identify the focus exponent
with everybody in either version, which means that we get a focused subject
in the active version and a focused adverbial in the passive version.

But if we look at the German original:

Das Wesen des Gleichgewichts [wird von jedermann intuitiv

BEGRIFFEN

the structural focus falls on the verb begreifen/understand, from where it can
project onto both extensions of the verb, intuitiv, von jedermann, according
to their contextual novelty. Das Wesen des Gleichgewichts could be seen as a
background.

Subtracting the nature of equivalence as given information we would end

up with the same focus projection in the active sentence in English:

[Everyone intuitively

UNDERSTANDS

] the nature of equivalence.

But could the translator not have used a passive version in English, too? And
why did the authors of the German original not use an active version?

Traditionally, one says that the passive allows us to choose another role as

topic of the sentence. In active sentences, the subject, especially that of
transitive verbs, normally has the role of an agent, of someone/something
that makes something happen. In the active sentence we take the agent as our
starting point. In the passive sentence we take the element that is affected by
the event as starting point.

A look at German shows, however, that there must be more to it as the

difference in topics can also be achieved by simply reordering an active
sentence. Thus, the original could have been an active sentence with the
object in topic position:

Das Wesen des Gleichgewichts begreift jedermann intuitiv.

But despite their identical topics, the active version and the passive original:

Das Wesen des Gleichgewichts wird von jedermann intuitiv begriffen.

In favour of primary relations

are different and the difference lies at the end of the sentences. While the
passive version has the main verb in the end position and hence in the
canonical focus position of German sentences, the reordered active version
ends on the modifier. This seems to have the same effect as an active sentence
where the object and the modifier have been scrambled:

Jedermann begreift das Wesen des Gleichgewichts intuitiv.

as opposed to:

Jedermann begreift intuitiv das Wesen des Gleichgewichts.

The comparison clearly shows that if we deviate from the basic order, we get
a marked focus interpretation:

Das Wesen des Gleichgewichts begreift jedermann [

INTUITIV

If we want a non-agentive topic, but do not want a marked focus, we have to
use the passive in German. Things are different in English.

Contrasting the agent

When German students are taught the use of the English passive, they are
told not to use the ‘by-’ phrase unless it is really relevant. Unless we want to
stress ‘someone’, we would rather say:

The letter has been opened.

and not:

The letter has been opened by someone.

Thus, if I say:

The letter has been opened by Cleo.

Cleo is the focus exponent of the sentence. But it is only a narrow focus,
which does not project onto the other elements of the sentence.

The letter has been opened by C

LEO

presents the letter has been opened as background knowledge and asserts
that it was Cleo who has done it as new information.

This is not the only reading possible. A specific context could make us

reinterpret the sentence and focus for example the letter (not the parcel) or

In favour of primary relations

opened (but not read). All of these foci are contrastive, as is the focus on the
verb-adjacent extension, by Cleo. This will be the default reading that comes
to mind first if we do not have any specific contexts. Yet the default reading,
too, has to match its context.

Thus, the analogous translation of our example is at first interpreted as:

The nature of equivalence is intuitively understood by [

EVERYONE

and the question is: does the context of the sentence match a narrow focus on
everyone? This is not the case. Though equivalence has been given as a
discourse topic for some time, the idea that we can understand it intuitively is
new. So what we need is a projecting focus that extends from everyone onto
intuitively understand. This is what we get directly in the active version
where focus projection is only restricted by the definite noun phrase in
the end:

[Everyone intuitively

UNDERSTANDS

] the nature of equivalence

What made us look at the passive paraphrase in the first place was the

German original.

Das Wesen des Gleichgewichts wird von jedermann intuitiv begriffen.

In contrast to the English passive, the German passive has a normal
projecting focus. The difference is in line with the hierarchical distance
between the agents of active/passive sentences in English and German. While
the German agent does not change its basic position relative to the
constituents that remain in the verb phrase, the English agent has to move to
the other side of the verb phrase:

_ . . . wird von jedermann intuitiv begriffen

_ . . . is intuitively understood by everyone

Again, as in the case of topicalization, it is the alternative directionality
which marks the derived structure more strongly in English.

However, let us stay with German for another moment, where not only

the passive but also the active has a projecting focus. Could we not use an
active version in German as in English?

Jedermann begreift intuitiv das Wesen des Gleichgewichts.

The active version presents the given information at the end of the sentence
and not where we would expect it in German, namely at the beginning of the
sentence. We could front the object:

In favour of primary relations

Das Wesen des Gleichgewichts begreift jedermann intuitiv.

but the resulting information structure would now be marked, as we know,
with intuitiv as a narrow focus. Moreover, as the object happens to have the
same form as the subject, we would run into processing difficulties right at
the beginning of our parsing attempts.

Using a passive sentence instead

helps us to place the given information in topic position and avoid all
processing obstacles in terms of parsing and focus identification.

If we look back at the first example:

Die Konturen einer räumlichen Struktur [werden von den Augenlinsen
maßstabsgetreu auf der Netzhaut

ABGEBILDET

we can make out exactly the same motivation for the German passive. It
secures topic position for die Konturen einer räumlichen Struktur, which is
the only contextually implied element. The corresponding active sentence
would be misleading in this respect. The basic order uses a new element as
topic and is thus not contextually appropriate:

Die Augenlinsen bilden die Konturen einer räumlichen Struktur
maßstabsgetreu auf der Netzhaut ab.

But if we topicalize the given element, it does not show that it is the object of
the sentence and misleads us onto a garden path we can hardly correct:

Die Konturen einer räumlichen Struktur bilden die Augenlinsen
maßstabsgetreu auf der Netzhaut ab.

All this is avoided by the passive version of the original.

On the other hand, the existing translation into English as an active

sentence:

The lenses of our eyes transmit the contours of a spatial structure in their
original proportions to the retina.

avoids all the attachment problems of the various passive versions, but
also their confusing and misleading conditions for a narrow focus interpre-
tation:

The contours of a spatial structure are transmitted in their original
proportions to the retina [by the lenses of our

EYES

Thus, the alternative perspectives secure optimal processing under the
language-specific conditions for parsing and focus interpretation in English
and German.

In favour of primary relations

A covert shift

The shift in perspective is not restricted to the difference between active and
passive sentences and in very many cases it is not even associated with a
difference in order. Thus, it may easily escape our attention in a pair like:

Damit ergibt sich ein Zusammenhang.

This forms a framework.

It is the pseudo-reflexive verb in German which produces this undercover
shift in perspective with the subject after the verb. In the English paraphrase,
the verb is followed by the object. So it is in the German:

Dies ergibt einen Zusammenhang

But the order of the informational elements in the alternative versions is the
same: it is the given element that starts the sentence, and the new element
that ends it. So what is the shift in perspective good for?

Now, in English the answer is again related to restrictions on elements in

topic position. A version like:

This way, a framework forms

is more difficult to process due to the cluster of two noun phrases and no
formal indicator which could give us a clue as to their syntactic relation to
each other and to the rest of the sentence. Longer sentences do begin like that
sometimes and can also have an indefinite focus subject before the verb, as
we have already seen in some of our examples. However, with a short
sentence like this one, the extra processing cost is not motivated, especially
as there are other options securing easy processing without much ado.
Instead of projecting the circumstances of an intransitive form into an
adverbial, we can project it into the subject of a transitive verb form:

This forms a framework.

Except for some formally clear-cut cases of sentence adverbials or

adverbs, adverbial phrases before the subject produce a processing obstacle
in English as we have to tease the structure of subject and adverbial apart
when we parse the beginning of the sentence. Although this is semantically
possible, it means some extra effort. It is much easier to automatically
identify the first phrase as the subject of the English sentence, all the more so
since the subject phrase itself may already be very complex.

German, on the other hand, has a free topic position admitting anything in

the initial place before the finite verb. If we compare the two German

In favour of primary relations

versions, though, starting with the subject or the adverbial, it is the adverbial
version which secures optimal processing.

For an example starting with the demonstrative pronoun dies or the

weaker form das, there is, indeed, a processing disadvantage we can make
out easily. Dies could be the subject or the object of a sentence. Compare for
example:

Dies ist bekannt.

Dies kennt jeder.

Undecidedness, even if temporary, is a processing disadvantage. As dies
ergibt vs. dies ergibt sich shows, there is also temporary undecidedness about
which role the subject dies plays (cause or result). Dies by itself does not tell
us. In the first case it would be the logical subject of the sentence, in the
second something like a logical object. When the logical object is promoted
to the syntactic subject of the sentence, it leaves the reflexive pronoun, like a
trace of its original function, in the structure after the verb.

In a language like German, in particular, where the final confirmation

of parsing results may have to wait till the main verb at the end of the
sentence, it is of advantage to know right away what role and what syntactic
function we are dealing with. Shifting the perspective helps. In contrast
to dies:

Dies ergibt einen Zusammenhang.

damit can be attributed its circumstantial function and role right away:

Damit ergibt sich ein Zusammenhang.

The shift between active and passive or passive-like perspectives occurs

both ways between German and English, but the dominating direction tends
to be German passive or passive-like sentences to active English sentences.

German adverbials become English subjects, rising in their syntactic status
from modifiers, secondary grammatical relations, to primary relations. The
asymmetry is to some extent due to the different preverbal options in English
and German. The subject-before-verb condition of English constrains
further extensions to the left. Thus, the reasons for the shift we can make out
at the end and at the beginning of sentences suggest that it is again the
combination of grammatical parameters and optimal processing that
promote the different preference. Interpreting information structures,
identifying the semantic and syntactic role of initial elements, parsing
syntactic structures quickly and correctly are universal goals of optimal
processing that meet with different conditions in active and passive (-like)
sentences in English and German.

In favour of primary relations

The morphological advantage

In the preceding examples, the adverbials of the passive and passive-like
versions impeded processing at the beginning of the analogous English
translations due to the subject-before-verb condition, and at the end of the
sentence due to the marked condition for focus interpretation. But
sometimes adverbials that occur in internal positions do not fare any better.
Thus the original:

Gestalt wird in unserem Denkorgan als Ganzes reflektiert.

which helped us to sharpen the Maxim in the first chapter, was not
translated as:

A form is reflected in our brains as a whole.

in the existing translation, but reframed as:

Our brains always interpret a form as an integrated whole.

(EW 1983b:70)

There is the shift in perspective as in the previous examples; as a consequence
Gestalt has been individualized as a form and thus adapted to the
individualizing context of brains and their individual reflections. Moreover,
the structure is filled in by two additional modifiers. Integrated clearly
reinforces the focus exponent. The function of always is less obvious, but it
seems to enhance the relevance of the entire statement by emphasizing its
general nature. Its position between the initial subject and the predicate
might indicate the left peripheral border of the focus projection:

Our brains [always interpret a form as an integrated

WHOLE

But why should we not use the more analogous version:

A form is always reflected in our brains as a whole?

If we ignore what we know about the meaning of the sentence from the
existing translation, we can see that the analogous version contains a parsing
problem concerning the attachment of the last phrase (which means that we
need not even bother with the question of focus). Syntactically, as a whole
could modify brains. As this would be the most economical way to build the
structure of the sentence, we could expect this reading to come to our minds
first, which would be a nuisance even if we correct ourselves afterwards. The
parsing error is definitely excluded if we use the active version:

Our brains reflect a form as a whole.

In favour of primary relations

In this version, the modifier is next to what it modifies, so that we get the
correct interpretation right away.

The attachment problem was imported from the German original, or

rather by analogous translation of the German original. But a back-
translated version:

Unser Denkorgan reflektiert Gestalt als Ganzes.

would, among other things, not place the given information at the beginning
of the sentence. However, if we move Gestalt to the beginning:

Gestalt reflektiert unser Denkorgan als Ganzes.

it would be interpreted as the subject of the sentence as there is no formal
indicator of its object status. In contrast, the passive version of the original
presents Gestalt in topic position and as syntactic subject of the sentence:

Gestalt wird in unserem Denkorgan als Ganzes reflektiert.

But it was the passive version which caused the attachment problems in
English, placing the modifier after the ‘wrong’ noun. Should we not have the
same attachment problems in German? The answer is that the passive
version does not leave any doubt about where the German modifier belongs
as it is clearly and unambiguously related to the subject by its morphological
form Gestalt als Ganzes. If it were a modifier of the adverbial, it would take
on the form of a dative, in correspondence with its nominal predecessor, in
unserem Denkorgan als Ganzem.

It seems, then, that parsing problems are more likely to occur in English

where the absence of different word forms forces us to rely more on abstract
parsing strategies in processing sentence structures. While German can make
use of a passive perspective to secure a close contextual link between a given
element and its antecedent in preceding sentences in order to avoid pro-
cessing disadvantages of active sentences, English cannot afford to worry so
much about a close contextual link of the initial element. It has to make sure
that the more elementary conditions of parsing are not violated. This seems
to lead to a preference of primary relations which includes the use of an
active perspective, in which various types of adverbials (by everybody, in our
brains) are reframed as subjects.

Lexical repercussions

The difference in perspective has its impact even on the combinatorial
possibility of lexical elements. There is the well-known difference in the
selection restrictions of German and English verbs which do or do not admit
inanimate subjects. Instead of:

In favour of primary relations

The seams of the sack split.

English can say:

The sack split its seams.

which is a shift in perspective not available to German. In German, we can
only say:

Die Nähte des Sackes zerrissen.

The sentence:

*Der Sack zerriß seine Nähte.

is ungrammatical. It has an odd personifying effect, which suggests that
German zerreißen requires an animate subject for its transitive meaning, that
is, it is restricted to what we could call the literal meaning of the verb. In
English, the meaning of the verb is extended onto non-literal combinations
with inanimate subjects.

In many cases the role of the English subject is circumstantial and would

be expressed by adverbials in German. A sentence like:

Clinical experiments have compared . . .

would for example be rendered by:

In klinischen Experimenten wurde . . . verglichen.

Although there are often no clear-cut selection restrictions, it seems that
many of the examples with a shift in perspective get a personifying touch if
back-translated into German.

Borders between what may and what may not be used with an inanimate

subject are, however, fuzzy. The brain, as a place of thinking, would in
German, too, allow verbs like reflektieren/reflect, interpretieren/ interpret.
Still, the active version is in German less neutral than in English or, as it were,
in our passive original, which projects the brains onto an adverbial, in
unserem Denkorgan.

In English, the preference of primary relations seems to promote the

elimination of selection restrictions in lexical transfer, that is, ‘historically’
in changing the combinatorial properties of English words. It may be
interesting to note that psycholinguistic experiments point to a direct
contextual influence on the interpretation of semantic roles at the beginning
of English sentences, which may compensate, to some extent, the levelling
effect of subject beginnings.

In favour of primary relations

In translation, the way from the German adverbial to the English subject

can lead to quite impressive shifts in perspective. To get rid of the conditional
adverbial at the beginning of the German original:

Unter Selektionszwang wird jedoch die bestangepaßte Sequenz aus-
gewählt.

(EW 1983a: 308)

the translation reframes the adverbial as subject and recategorizes part of the
compound as a verb, down-toning the original meaning (-zwang/force):

But a selection process sees to it that only the best-adapted sequence is
chosen.

(EW 1983b: 275)

A back-translated version:

Ein Selektionsprozeß sorgt jedoch dafür, daß nur die bestangepaßte
Sequenz ausgewählt wird.

is highly personifying, besides being unnecessarily explicit. Explicitness will
be our topic in the next part of the book, but the recategorization option can
also be used in an economical way, which we will turn to right now.

Revisiting the beginning

Having made out the reasons for the preference of primary relations in
English, we can now classify the very first example we looked at in some
detail as yet another of such reframing cases. The head of the adverbial am
Anfang was reframed and recategorized as a verb, transforming the original:

Am Anfang der Schöpfung steht die Gestaltbildung.

into:

Creation began with the emergence of form.

The undercover shift reminds us of the example with damit ergibt sich/this
forms. But as neither of the verbs stehen/begin is transitive, the shift in
perspective is not of the canonical type that characterized the previous
examples. Yet it may be no coincidence that English has a stricter selection
restriction with static, or less ‘active’, verbs, and is more at home with
dynamic verbs. What we need for a simple shift is a verb to match the frame:

The beginning of creation _ the emergence of form.

In favour of primary relations

But there is no element that could come close enough to the merely local
meaning of the original – at least not as close as the existing translation,
which restructures the original frame more radically.

Let me summarize what we have seen. Parsing ease and easy focus

identification were the reasons for a passive or passive-like perspective in the
German original and for an active perspective in the English translation. The
conditions for the alternative preferences included structural and morpho-
logical differences causing parsing problems in English or in German,
topicalization constraints in English, and, in both languages, structural
differences resulting in marked (narrow) focus interpretations.

In the cases compared, German uses a passive version to achieve discourse-

appropriate distribution of information, localizing given elements at the
beginning of a sentence with a normally projecting focus. As processing of an
analogous structure in English would be hampered by parsing problems or
misled into a marked focus interpretation, the passive sentences are reframed
as active sentences in the English translations. (The fact that by-phrases are
more strongly focused than the corresponding von-phrases was seen as
resulting from the greater distance between the original position of agents as
subjects and their ‘landing’ sites in passive sentences. The effect is
comparable to the stronger focusing of most topicalized elements in English.)

In some passive (-like) cases where German topicalizes material to avoid

syntactic (subject/object) ambiguities at the beginning of sentences, the
English translation reframes topicalized elements as subjects of active
sentences to avoid parsing difficulties for clusters of phrases in preverbal
position (as well as inappropriate focus interpretations).

In spite of all these differences and their complex interactions, optimal

processing is occasionally secured by analogous sentence structures. Thus,
the analogous translation:

For the working molecular biologist today, the identical reproduction of
an organism is no doubt a greater miracle than the occasional meta-
morphosis of one.

of:

Für den Molekularbiologen unserer Zeit ist die identische Reproduktion
einer Gestalt zweifellos das größere Wunder als ihre gelegentliche
Metamorphose.

illustrates what the sentence says about identical reproduction. In spite of the
restrictions for the beginning of English sentences, the translation preposes a
contrastively focused prepositional object just like the German original, and
– except for the subject-verb condition – arranges everything else in the same
way. Obviously, the topic position is an appropriate position for the
prepositional object with for.

In favour of primary relations

That is, we can expect some verb patterns in English and German to

permit similar word order variations. Unlike a version with basic word
order, the English topicalization here does not present any greater processing
obstacle for parsing and focus identification. Nevertheless, in translation, it
is the ‘identical reproduction’ that will occur occasionally, while the
‘metamorphosis’ – reordering, reframing, recategorizing, etc. – is what we
can generally expect. And optimal processing is in each case our guideline for
natural selection.

Structural weight

One of the most intriguing differences between original sentences and their
translations concerns the degree of explicitness preferred under source
language and target language conditions. Stylistic registers may be charac-
terized by a higher or lower degree of economy, and the Maxim makes us
expect a similar profile for the same register in the translation. A legal
document will, certainly, be more explicit than a warning in both English
and in German. Yet if we study translations more closely, we will find that a
similar degree of explicitness is by no means the rule.

Take the type of pragmatic, stylistically neutral register of popular

scientific texts we have been looking at so far. There are minor differences,
for example where pro-forms are replaced by fully lexicalized elements,
compounds decomposed into phrases, or individual words filled in at various
points of the original structure. But there are also quite extensive differences
where phrases are extended into clauses, or clauses are inserted into the
structure as if ‘out of the blue’.

In discussing word order and grammatical relations, we have already

come across an impressive array of such cases. Some were lexically based, for
instance, lexical gaps filled in by phrases (Gestaltbildung/emergence of
form), while some were due to grammatical differences, as for example
conditions on nominalization of adjectives (das Wesentliche/the crucial
element). The weightier differences, however, were related to information
structure. There was the insertion of lexical elements (like schon, always) to
mark or delimit foci, and there was the extension of phrases into clauses
to disambiguate focus interpretations (mit der Vervollkommnung des
Übersetzungsmechanismus/when the translation mechanism was perfected).

As the difference in structural explicitness is no rare phenomenon, it is

certainly worthwhile looking at these and other cases more systematically.
The challenge will be to identify the language-specific conditions over and
above the grammatically known cases for the use of clauses instead of
phrases, phrases instead of words, and words instead of pro-forms.

The answer will in most cases involve optimal processing in terms of

information structure, which requires, again, a very subtle, if not empathic
analysis of the individual example in its context. We will spend most of the

Structural weight

following efforts in this field. There is one area of processing conditions,
though, which relates to a difference between German and English we have
not yet looked into. It is the area of referential relations, technically
anaphoric relations (in their widest sense), which plays a decisive role in
the choice between pro-forms and phrases. There is one major difference
for the use of anaphoric relations in German and English, which we will
turn to now.

Anaphoric ambiguities

Linguistic elements refer to things or events, including the various aspects
characterizing them. If a previous reference is resumed in a sentence,
we speak of an anaphoric relation between the resumptive element, the
anaphor, and its antecedent. As the interpretation of the anaphor, especially
of pronominal anaphors, like it, depends heavily upon the meaning of the
antecedent, optimal relevance requires us to identify the antecedent quickly.
There are many ambiguities to be resolved, that is, there are many cases
where there is more than one potential antecedent of an anaphor.

In the interest of optimal processing referential ambiguities should be

avoided. But the conditions for referential ambiguities may turn out to be
different in different languages, which is only natural if we think of all the
grammatical and lexical options of a language.

One of the most striking differences in anaphora resolution between

English and German is related to grammatical gender. An antecedent can be
identified much more quickly if we can distinguish between potential
candidates on the basis of their grammatical gender. This is the case in
German, which means that there is also some advantage associated with the
otherwise frustrating grammatical gender in German: der Satz, die Phrase,
das Wort can be unambiguously referred to later in the discourse by er, sie, es.

Except for natural gender, which could be referred to by ‘he’ and ‘she’,

most English nouns are neuter and thus not to be distinguished by pro-
nominal anaphors. To avoid the processing disadvantages arising from this,
English uses fewer pronouns or other pro-forms than German, and does not
mind repeating fully lexicalized noun phrases at relatively short intervals. By
contrast, repeating words or phrases at short intervals is considered
redundant and in most cases poor style in German. Pronouns are meant to
reduce processing effort to a minimum and the Principle of Optimal
Relevance requires us to make use of this option unless we have a special
reason not to do so.

Referential ambiguity is such a reason, and thus an original German

sentence like:

Der zweite Hauptsatz sagt aus, daß die Entropie eines abgeschlossenen
Systems zunehmen muß, bis dieses im Gleichgewicht ist.

(EW 1983a: 175)

Structural weight

becomes more explicit in the translation:

The second law of thermodynamics says that the entropy of a system
will continue to increase until the system achieves equilibrium.

(EW 1983b: 152)

In German, the anaphoric relation between the demonstrative pronoun
dieses and its antecedent System is clear – the next potential antecedent, die
Entropie, is feminine and requires a demonstrative pronoun in the form of
diese. In English, the use of a pronoun instead of the system results in an
anaphoric ambiguity:

the entropy of a system will continue to increase until it achieves
equilibrium.

As System and entropy could both serve as antecedents of the pronoun, the
full lexical form was preferred to the pronoun.

Things are not always that straightforward, though. The discussion of the

next example will illuminate a more complex case of anaphoric ambiguity.

The missing gender

As it is such a beautiful example, we will look from English into German for
this one case.

Way out at sea and all on his own, Hemingway’s Old Man says to himself:

If there is a hurricane, you always see the signs of it in the sky for days
ahead, if you are at sea.

(Hemingway 1952: 67)

This is a comforting thought in view of the white cumulus and the ‘thin
feathers of the cirrus against the high September sky’. The Old Man had:

thought of how some men feared being out of sight of land in a small
boat and knew they were right in the months of sudden bad weather. But
now they were in hurricane months and, when there are no hurricanes,
the weather of hurricane months is the best of all the year.

If there is a hurricane, you always see the signs of it in the sky for days

ahead, if you are at sea.

The following translation into German reorders constituents in the way we
are already familiar with. But it also reduces a phrase with a pronoun and a
lexical head to a pronoun:

Wenn es einen Orkan gibt, kann man das schon Tage vorher am Himmel
sehen, wenn man auf See ist.

Structural weight

The verb phrase in the main clause contains an object and two adverbials,
which are ordered according to their syntactic-semantic hierarchy, extend-
ing from the left peripheral verb in English and the right peripheral verb in
German.

The original has the object closest to the verb, the signs of it, followed by

the local adverbial in the sky, which in turn is followed by the temporal
adverbial for days ahead. The order of the adverbials is reversed in German,
in line with the alternative directionality: the temporal adverbial precedes
the local adverbial schon Tage vorher am Himmel.

As the (pronominal) object is definite it has been moved out of its basic

position before the main verb, and as the topic position is already occupied
by the conditional clause, the object is placed after the finite verb and the
subject: kann man das schon Tage vorher am Himmel. The different order is
grammatically determined; any other order would be marked and hence not
equivalent to the original.

There are three structural differences between the original and the

translation producing a different degree of explicitness. There is, first, the
insertion of the focusing element schon, assigning to the temporal adverbial
a contrastive focus. As the elements following it – the local adverbial and the
main verb – are not yet given, they participate in the main focus of the
sentence. As the subject and object are given, schon marks the border
between background and focus.

Second, there is the insertion of the modal verb, kann, as a sort of verbal

dummy, which secures final position for the main verb within a continuous
focus projection:

kann man das [schon T

AGE

vorher] [am H

IMMEL

SEHEN

In English, the frequency adverb, always, also marks the beginning of the
focus projection. But here, the object is included in the projection:

you [always see the

SIGNS

of it in the sky for days

AHEAD

This is a consequence of the third structural difference between the objects of
the original and of the translation: the signs of it vs. das.

If the signs of it is translated analogously into German, it results in a

structure which is only marginally acceptable:

Wenn es einen Orkan gibt, kann man die Anzeichen davon schon Tage
vorher am Himmel sehen.

Even if we reduce the clumsy object to seine Vorboten, the object is
unnecessarily ‘heavy’:

Wenn es einen Orkan gibt, kann man seine Vorboten schon Tage vorher
am Himmel sehen.

Structural weight

This means nothing else but that the processing cost is too high for what is to
be conveyed. The pronominal object is much better. On the whole, its
interpretation does not really differ from that of the original. What could
you see of a hurricane in the sky for days ahead if not its signs, whatever
these may be for people like the Old Man who can read the sky?

But surely Hemingway must have thought of something when he decided

to use a fully lexicalized phrase rather than a pronoun. If you back-translate
the German das as that or rather it – the demonstrative that is too strong for
the anaphoric function of the object – you will get onto an anaphoric garden
path. In:

If there is a hurricane you can see it in the sky for days ahead.

you will take the hurricane as the antecedent to it and not the future event,
which is the referent of the conditional clause and the clear antecedent of the
German das.

Thus, Hemingway could not have made use of a pronominal version

without creating a processing problem. Although we shall see in Chapter 7
that there may be stylistic reasons for deviating from the Principle of
Optimal Relevance, the processing disadvantage of a pronominal version
cannot be turned into anything advantageous. (But the replacement can! See
below.)

In German, the situation is different. If we wanted to resume the reference

to the hurricane, the pronoun would have to agree with it in gender. Der
Orkan is masculine:

Wenn es einen Orkan gibt, kann man ihn schon Tage vorher am Himmel
sehen . . .

Now, the fascination of this example is due to the extra meaning of signs.
While a mere repetition of a fully lexicalized form, as in the previous
example with system, does not add anything new to the sentence – in very
much the same way as a pronoun – a lexical element which has not yet been
mentioned in the preceding context opens up new interpretational possi-
bilities. It is the difference between the implied, the virtual meaning, and the
explicitly expressed meaning which makes the insertion or deletion of
dummies a sensitive strategy in translation. But if processing conditions
require an additional element in target or source language, the Principle of
Optimal Relevance tells us to restrict surface equivalence by the insertion or
deletion of a dummy.

Tucked away

Dummies disambiguating anaphoric relations may even take on the form of
a clause. The translation of the following example:

Structural weight

Man kann sehr wohl als bloßer Teilnehmer an einem Spiel dessen Regeln
begreifen.

(EW 1983a: 296)

replaces the demonstrative dessen by a postnominal clause:

As participants in a game we can certainly understand the rules by which
it is played.

(EW 1983b: 264)

But could we not stay closer to the original? After all, the version:

As participants in a game we can certainly understand its rules.

is grammatically acceptable and says about the same.

German uses dessen instead of seine to avoid anaphoric ambiguity: both

Spiel and Teilnehmer could serve as an antecedent to seine. But it can only
refer to game, not to participant. The ambiguity does not exist in English.
What made the translator use the clausal paraphrase by which it is played
instead of its?

Now, there is again a difference in the position of an adverbial: als bloßer

Teilnehmer an einem Spiel has been moved from its medial position, where it
would be hardly acceptable in English, to the beginning of the sentence. This
has increased the distance between the possessive pronoun and its ante-
cedent, the game, considerably. In fact, the game, which is only a subordi-
nated phrase within the adverbial, has been moved before the actual English
clause, which begins – canonically – with a subject.

While in German the anaphor and its antecedent are placed next to each

other (Spiel dessen), it takes some processing effort to identify the antecedent
of the shorter English version:

As participants in a game we can certainly understand its rules.

where game is tucked away into something like a parenthesis at the
beginning of the sentence. But wait. The same situation holds true for the
existing translation, where the pronominal subject, it, similarly relates back
to the far away game:

As participants in a game we can certainly understand the rules by which
it is played.

So where is the advantage of the clausal version?

It lies in its structural explicitness. In contrast to the possessive its, the

subject it is a primary relation, and the clause with its predicate played does
not leave much choice for the interpretation of it. Thus the structural

Structural weight

expansion of the pronominal attribute into a relative clause helps us to avoid
the longer backtracking process that would otherwise be necessary.

The last example differs from the preceding ones as the anaphoric

disadvantage of the analogous translation is itself a result of additional
changes that have become necessary for other reasons, as for example, the
different position of an adverbial. But whatever it is that impedes anaphora
resolution, if we adhere to the Principle of Optimal Relevance, optimal
processing requires us to get rid of the obstacle – albeit at the cost of an extra
clause.

Decomposing the unpredictable

Leaving the clearly delineated topic of anaphora resolution, we may easily
lose our way among the countless cases of structural extensions, so let me
first draw some rough lines. We can distinguish three different types of
extensions, all of which could be of a clausal or of a phrasal nature: first,
the original phrase is expanded into a more complex structure; second, the
original phrase is replaced by a more complex structure; third, an additional
phrase or clause is inserted into the original structure. Though the changes
can be triggered by a lexical gap, or by a grammatical gap (as in nominal-
ization options), they can also occur where there are analogous target
language means. The interesting question is when one would use phrasal,
when clausal extensions if one has to fill in some gap, and why one would use
extensions at all if there is no gap to be filled in. As there are just as many
structural reductions in translations as there are structural extensions, we
can ask all these questions once more ‘backwards’. We will look at some
examples of one type or the other, preceding from simple examples to more
complex ones in each case.

There are many instances that do not allow any predictive generalizations,

at least none that could be reached within the confines of the Key. There are,
for example, those simple cases of structural extensions replacing a com-
pound by a syntactic phrase made up of the components of the compound.
Compare the following example:

Schwer zu sagen, wie und wann sich die Idee der Sukzession geschichts-
philosophisch durchgesetzt hat.

(E 1999: 11)

It is difficult to say how and when the idea of succession caught on in the
philosophy of history.

(E 1997: 35)

Although the expressions geschichtsphilosophisch/in the philosophy of
history are not really equivalent, we can accommodate the success story of
the idea of succession from the academic circles of the English version to the

Structural weight

wider public of the German original. And if we are not very pedantic, we will
not even notice the vicinity of a false friend.

Anyhow, the translation could pass as one of the many cases of idio-

syncratic lexical correspondences where the criteria of processing ease and
contextual appropriateness, in themselves, are not very helpful. In most
cases of decomposing we cannot stick to the original so closely: Gestalt-
bildung, Mittelwertbildung, for example, were turned into emergence of
form, process of averaging, which leaves us without hope for a simple
generalization about patterns of decomposing. And even the little we can
trace back to the original may disappear. If we find:

Er verfügt über ein beträchtliches Durchsetzungsvermögen.

(E 1988: 67)

translated as:

He displays considerable determination in getting his own way.

(E 1997: 281)

we cannot identify any element of the English phrase with the components of
the German compound. Capacity to get one’s own way has been replaced by
intention/determination – a modal shift which is not predictable and has to
be accepted as just another one of these idiosyncratic differences between
German and English words.

There are cases of decomposition, though, where considerations of

language-specific processing conditions do play a role. They belong more
often than not to the cases of structural extensions which involve aspects of
information structure and are therefore promising objects for our studies. All
of these extensions could be summarized under the heading of dummy
elements – words, phrases, clauses – used in the interest of processing ease. I
shall begin with the grammatically obvious, and turn to stylistic examples of
ever greater complexity directly afterwards.

Grammatical dummies

A number of structural extensions are necessary for grammatical reasons, as
for example the dummy nouns needed to compensate for the English
restrictions on nominalizations, or a structural anchor for complement
clauses, which may be needed either in English or in German. In most of
these cases, the grammatical dummy is easy to localize, as in the ‘crucial
element’ or ‘the fact that’. But additional changes, as for example a shift in
perspective, can make it difficult to trace a structural extension back to its
grammatical reason. Compare the following example with its structural
extension between the predicate of the main clause and the that-clause
in English:

Structural weight

Dagegen spricht, daß die ‘Überbleibsel der Vergangenheit’ ebenso
unkontrollierbar zu wuchern scheinen wie die Fortschritte.

(E 1999: 11)

But that can be countered with the argument that these remnants of the
past seem to proliferate just as uncontrollably as progress.

(E 1997: 34)

The subordinated clause in German is the subject of the sentence:

Daß die Überbleibsel der Vergangenheit ebenso unkontrollierbar zu
wuchern scheinen wie die Fortschritte spricht dagegen.

In the original, the subject has been replaced by the prepositional object
dagegen in the topic position. The pronominal form of the topicalized
element tells us that it is given and hence requires reordering of the basic
order.

While the German original is an active sentence, the perspective is shifted

to a passive in English: something speaks against something/dagegen vs.
something (here: that) is countered with something. As that and dagegen
have the same reference, the shift in perspective is covert, but it is due to the
same difference between German and English topic possibilities we are
already familiar with.

In this case the passive perspective requires the insertion of a phrase to

which the subordinated clause, which is not any longer the subject of the
sentence, can be attached: the argument that. Although the argument seems
to have more meaning than the previous dummies, things, elements, it does
not really say more than what we can infer from the context: but that can be
countered with . . . that remnants of the past seem to proliferate just as
uncontrollably as progress. The dummy does not contribute anything to the
meaning of the sentence that would not be there without it, but it is needed
for structural reasons.

Grammatical dummies can also serve as structural anchors for stylistic

dummies, used to improve processing conditions, as in the following case:

Das Abgetane feiert, im kleinen wie im großen, auf scheinbar wirre
Weise seine Wiederkehr.

(E 1999: 21)

All the things we had settled and disposed of, the great as well as the
small, are celebrating their tangled comeback.

(E 1997: 44)

The nominalized participle cannot be translated as a dummy noun with
a prenominal modifier, but it could be translated as a dummy noun with a

Structural weight

postnominal modifier, or with two modifiers as one may not be enough to
cover the ambiguity of the original:

All the things settled and disposed of, . . .

But the actual translator has extended the postnominal modifier into a
clause:

All the things we had settled and disposed of,

extending the structural explicitness more than would have been necessary
for grammatical reasons. However, if the participles were not extended into
a relative clause, they would present a canonical case of garden path
structure. For the English reader who does not know the German original,
the first participle at least could be mistaken as the finite verb of the sentence:

All the things settled . . .

And even if the parser began to realize its error when processing the second
participle, disposed of, there is still the apposition:

the great as well as the small,

to get through before the finite verb, are celebrating, would really clarify the
issue. The clausal extension of the modifier disambiguates the structure right
away: the parser can avoid the garden path and will attach the specifying
clause to its dummy head (things) without delay.

In most cases, however, stylistic dummies improve processing conditions

for focus identification. We have already looked at an example where the
nominalization of an adjective served as a structural anchor for yet another
dummy noun, the crucial element in a problem, which secured the right focus
interpretation. But there are many more such cases of phrasal or clausal
expansions contributing to a better understanding of information structure.
We will now look at half a dozen different cases, each of which presents a
unique set of conditions promoting the use of dummies. Grammaticalized
options for structural extensions in the interest of focus identification will be
discussed in the next chapter.

An extra clause for the second focus

Occasionally, we come across structural extensions in translations which
seem to be used per se, without any recognizable reason:

Unübersehbar groß ist die Zahl von möglichen Satzkombinationen.

(EW 1983a: 300)

Structural weight

is translated with a dummy relative clause after the subject:

The number of possible sentences we can form is astronomical.

(EW 1983b: 267)

The original uses a marked word order with a predicate in topic position.
This tells us clearly that the predicate is focused. But how should we read the
rest of the sentence? What about the structural focus exponent at the end of a
German sentence? We can only decide on this with the help of the context.

Having talked about the huge phonetic capacities of languages and the

ongoing assignment of meaning to phonetic sequences, Eigen and Winkler
turn to combining words into sentences, illustrating the astronomical
number of possible sentences in the following with a simple example. As they
use italics to mark new topics in this section, we know from the preceding
context that combining sentences, and hence mögliche Satzkombinationen/
possible sentences in our example, is already given and thus defocused
information. Zahl is the new information. It characterizes the new part of a
discourse topic, which is already given, and need not be the sentence topic, as
the example demonstrates.

Structurally and contextually, we have a marked focus at the beginning of

the sentence, the preposed predicate, unübersehbar groß, and a focus
exponent right after the copula, die Zahl:

Unübersehbar

GROSS

ist die Z

AHL

von möglichen Satzkombinationen.

An analogous translation with a topicalized predicate would be interpreted
differently:

Astronomical is the number of possible sentences.

It could be used as an answer to the question: what is astronomical?, which
presents astronomical as given and therefore does not match the context of
the original. If we use the basic word order:

The number of possible sentences is astronomical.

we can secure the focus on astronomical, but miss the additional focus on
number.

Now, if we use an additional relative clause, we have the possibility of a

second structural focus in the sentence, which will be contextually shifted to
the only new element of the subject, number:

The

NUMBER

of possible sentences we can form is

ASTRONOMICAL

As the number of possible sentences we can form is identical with the number

Structural weight

of possible sentences (if we could not form them, they would not be possible,
and if they were not possible, we could not form them), the meaning of the
extended structure is equivalent to that of the original phrase. Thus, the
clausal expansion of the subject is just another variant of focus spacing,
which secures optimal focus interpretation.

Adverbial expansion

The preceding expansion made use of a relative clause, and in particular a
defining clause, which is an integrative part of the phrase it modifies. But
other types, like for example adverbial clauses, may be helpful, too, as a
language-specific guide to optimal focus interpretation. As a processing aid
they play a different role than defining relative clauses as – in English – they
present relatively autonomous informational units.

Things are different in German, and the difference is again due to the

alternative directionality, which may promote alternative processing
strategies for the integration of clauses into the information structure of the
sentence they belong to. In the following example, a phrasal adverbial in a
German original is expanded into a clausal adverbial in the existing English
translation to secure focus identification in the main clause. In German, the
phrasal adverbial, im Hinblick darauf, contains a nominal dummy that is
grammatically necessary as a structural anchor for the subclause:

Eine Aufklärung dieses Sachverhalts ist vor allem im Hinblick darauf,
welche Kräfte im Bereich der Lebewesen für Ordnung sorgen, von
allergrößter Bedeutung.

(EW 1983a: 175)

The translation expands this adverbial phrase into a conditional clause and
places the predicate before it:

The elucidation of this point is of greatest importance if we want to
understand what forces provide order in the animate realm.

(EW 1983b: 152)

If the translation were to use an analogous adverbial phrase, it could intro-
duce the clausal argument of the original by something like, in regard to or,
with another dummy noun, regarding/for the question:

The elucidation of this point is of greatest importance for the question
what forces provide order in the animate realm.

But this is obviously less transparent than the existing translation.

If we translated the English translation back into German, the result

would also not be as perspicuous as the German original:

Structural weight

Eine Aufklärung dieses Sachverhalts ist von allergrößter Bedeutung,
wenn wir verstehen wollen, welche Kräfte im Bereich der Lebewesen für
Ordnung sorgen.

What is it that promotes the clausal version in English, and the phrasal
version in German?

Let us, again, determine the information structure of the example. We can

read some of its properties off the sentence structure, although the sentence
by itself does not tell us very much about what it refers to:

Eine Aufklärung dieses Sachverhalts ist vor allem im Hinblick darauf,
welche Kräfte im Bereich der Lebewesen für Ordnung sorgen, von
allergrößter Bedeutung.

If we look into the context, we can see that the point which is to be elucidated
refers to an apparent contradiction in the statistical, mechanical interpre-
tation of the second law of thermodynamics. But let us first analyse the
sentence structurally. We can identify the focus exponent of the sentence
by its final position and the superlative: allergrößter Bedeutung. Focus
projection could theoretically include everything except the element, dieses
Sachverhalts, which is clearly given. But there is a second focus marked by
the adverbial, vor allem, which emphasizes the pronominal adverb, darauf,
and through it the interrogative clause. Within the interrogative clause, the
focus exponent would be the next argument to the verb, Ordnung:

[Eine A

UFKLÄRUNG

] dieses Sachverhalts ist [vor allem im Hinblick

DARAUF

], welche Kräfte im Bereich der Lebewesen für [O

RDNUNG

sorgen],

von [allergrößter B

EDEUTUNG

But when we consult the context to see which of the phrases are focused, it
implies forces providing order elsewhere so that O

RDNUNG

has to be

defocused in favour of a contrastive focus on Lebewesen:

[Eine A

UFKLÄRUNG

] dieses Sachverhalts ist [vor allem im Hinblick

DARAUF

], welche Kräfte im Bereich der [L

EBEWESEN

] für Ordnung sorgen,

von [allergrößter B

EDEUTUNG

The focus structure of the translation is similar except for the difference
between the phrasal/clausal link between the matrix clause and the inter-
rogative clause (and the indefinite/definite topic) – structurally:

The elucidation of this point [is of the greatest

IMPORTANCE

] [if we want

UNDERSTAND

] what forces [provide

ORDER

] in the animate realm.

reanalysed contextually:

Structural weight

The elucidation of this point [is of the greatest

IMPORTANCE

] [if we want

UNDERSTAND

] what forces provide order in the [

ANIMATE

] realm.

But what we wanted to study was the difference between the phrasal/clausal
structure that links the interrogative to its matrix clause. The question was:
what makes the clausal link better in English and the phrasal link better in
German?

This is the answer: if we use the phrasal version in English, the superlative

will be defocused by the restriction following it. There is, after all, a differ-
ence between something is of the greatest importance and something is of the
greatest importance for such and such. That is, in the phrasal version, we will
miss the focus on greatest importance despite the superlative. In the extreme,
this could mean that we take everything that precedes the interrogative
clause and its focus as background:

The elucidation of this point is of the greatest importance for the
question what forces provide [

ORDER

] in the animate realm.

But this would take many more steps to be fitted into the context and
therefore not be equivalent to the original, either.

To be able to identify the focus on greatest

IMPORTANCE

we have to

recognize the first part of the sentence as an autonomous information unit
containing a focus of its own before we integrate it with the other parts of the
sentence. This is exactly what the introduction of the interrogative clause
secures. Reaching the conditional clause, we will stop to identify the focus of
the preceding clause before we continue to process the remaining structure:

The elucidation of this point [is of the greatest

IMPORTANCE

] [if we want

UNDERSTAND

] what forces provide order in the [

ANIMATE

] realm.

But this is the effect of the conditional clause in English. If we were to use the
back-translated version in German:

Eine Aufklärung dieses Sachverhalts ist von allergrößter Bedeutung,
wenn wir verstehen wollen, welche Kräfte im Bereich der Lebewesen für
Ordnung sorgen.

we would relativize the superlative by the following conditional clause in
very much the same way as it was relativized by the phrasal version in
English. Relativizing the superlative by the following condition not only
makes us miss the main focus of the sentence, it produces a somewhat
strange reading. The conditional clause is now informationally more rele-
vant than the preceding main clause, which suggests something like: if you
want to understand the forces, the elucidation is important, if we do not
want to understand the forces, elucidation is not important.

Structural weight

Why should there be such a difference in the interpretation of English and

German conditional clauses following their main clauses? A plausible
answer could be this: in a language with a right-peripheral focus expectation
our processing strategy will subordinate preceding clauses informationally
to subsequent ones, unless there is some indicator telling us that we should
not do this. But in a language where we are used to less relevant information
following the focus of the sentence, which is left peripheral within the verb
phrase, our processing strategy will be alternative. We will subordinate the
subsequent clause to the preceding one. But as this means that we could expect
the focus in the preceding clause, we have to check the preceding clause for a
possible focus before we go on processing the subsequent structure.

Of course, there could be formal indicators telling us to drop the default

strategy. Some of the elements connecting clauses – as, for example, defining
relative clauses – will certainly do so. But the default interpretations as well
as the ‘exceptions’ are in many cases language-specific, that is, they may
differ with the type of clause and between source and target language. A
systematic description of the differences between information structures in
German and English clauses is still missing, but translational evidence of the
‘controlled’ type we have been studying could definitely improve our
understanding.

Too short

So far, the structural extensions helped us to identify a focus we might have
missed otherwise. The major problems limiting easy focus identification in
translations with similar word order were again due to the alternative
directionality of German and English (simple and complex) sentences and
the stricter constraints on topicalization in English. The structural extension
identified a focus in the translation that corresponded to a focus in the
original. If the focus exponents happened to be different elements,
equivalence was secured by focus projection (as in the case of stört auf
unerträgliche Weise/presents an insufferable disturbance), or by contextual
defocusing (Zahl/form).

Sometimes the relationship between the original focus and that of the

translations is less transparent and it takes us quite a while to figure out
the information structural equivalence between such obviously divergent sur-
face structures. Compare the following example, where the German original:

Damit ist die Grundlage für eine Evolution geschaffen.

(EW 1983a: 97)

has been translated as:

This is the fundamental process that makes evolution possible.

(EW 1983b: 78)

Structural weight

We can recognize a covert shift in perspective, something like This makes
evolution possible/by this evolution has become possible. We assume that
this and damit refer to the same antecedent, which is confirmed when we
look into the context. The process referred to is the selection of
advantageous mutations, ‘misreadings’, in the reproduction of an organism.
The anaphoric reference to it is grammatically encoded as an adverbial in
German, and as a subject in English.

Additional differences, however, are clearly superimposed on the simple

paraphrase shifting the perspective. One of these is the structural expansion
of Grundlage für eine Evolution/basis for an evolution into fundamental
process that makes evolution possible. The clausal expansion is accom-
panied by a structural extension of the syntactic head, which is made more
dynamic by the insertion of a dummy noun, process.

These are altogether quite massive deviations from the original structure.

What is their basis? The shift in perspective, which may be licensed along the
lines we have worked out before, would also be achieved by the simple
sentence:

This is the basis for evolution.

But there are two aspects of the German original which are not expressed
explicitly in this version, namely, that the basis is a result of what has been
described before – ist geschaffen/has been created – and that evolution is
indefinite:

This has created the basis for (an) evolution.

Evolution is normally conceived of as uncountable in both languages, but
English may be more sensitive, here, and resent the idea of several evolutions
more strongly. Without the indefinite article, everything but the predicate
seems to be given. There is nothing in the form of the sentence that would
attract any attention, and the statement with the copula seems even less
significant. If we put the simple sentence into the context it belongs to, it is
definitely ‘weaker’ than its complex paraphrase:

As we have already mentioned, conservative structures, particularly at
the subcellular, molecular level, play a major role in morphogenesis. The
molecular make-up of the entire information storage system, the double
helix, the spatial placement of functional groups in biocatalysts, such as
enzymes, indeed, the entire structures of organelles and cells make use of
conservative forces that precisely determine the location of every detail.
The invariance of the genetic program is a concrete expression of these
rigid conservative forces. And occasional variations, or mutations, are
nothing more than ‘misreadings’ that occur as the result of thermal
fluctuations on this level. These misreadings are preserved by means of

Structural weight

identical reproduction, and if they prove to be advantageous, they will
be selected. This is the basis of evolution.

versus:

The invariance of the genetic program is a concrete expression of these
rigid conservative forces. And occasional variations, or mutations, are
nothing more than ‘misreadings’ that occur as the result of thermal
fluctuations on this level. These misreadings are preserved by means of
identical reproduction, and if they prove to be advantageous, they will be
selected. This is the fundamental process that makes evolution possible.

Although the messages of the two sentences will become more equivalent if
we spell out all the contextual implications, we may all share the feeling that
the simple sentence is too short and does not really match the context. So
what is it that could make one prefer the more explicit structure over the less
explicit one within the context?

My claim is that one needs more structural weight to show the significance

of the sentence for the discourse. The next section will spell out what this
means in regard to this and other cases.

Textual relevance

As a rule, progress in discourse structures the information of each sentence
into focus and background relative to the immediate context, but there are
also discourse relations that encompass larger segments of a text. That is, the
information of a sentence can participate in macrostructural relations and
therefore be of higher relevance than the elements surrounding it.

Relevance is always relevance relative to something. When we talk about

focus as the most relevant element of a sentence, we think of it as being more
relevant than the other elements of the sentence relative to the cognitive
gains made at a certain stage of the discourse. In most cases this amounts to
new elements versus contextually given ones. If givenness refers to
immediately preceding elements, the concept of relevance is mainly of local
significance. But givenness can also relate back to something that can be
considered a major topic of the text, in which case the relevance of an
element will, of course, surpass local significance. Let us speak of textual
relevance in the latter case.

It is the concept of textual relevance that can explain our preference for

the structurally more explicit translation above. This becomes clear when we
try to determine the meaning of the first element in the sentence. The first
element of this sentence is the demonstrative pronoun this, which can only be
interpreted via its antecedent in the preceding context.

If we take the closest candidate possible, we will identify this with the

process of selection mentioned in the last clause:

Structural weight

they will be selected. This is the basis of evolution.

But the antecedent could also be the last couple of clauses:

if they prove to be advantageous, they will be selected. This is the basis
of evolution.

or the entire complex sentence preceding our example:

These misreadings are preserved by means of identical reproduction,
and if they prove to be advantageous, they will be selected. This is the
basis of evolution.

Or even more sentences?

If we study the whole paragraph more closely, we can discover something

like a contrast between rigid conservative forces and occasional variations,
with the sentence about evolution wrapping up the part about variation. The
subsequent paragraph is opened by a question about the ontogenetic
development of an organism:

But how then are we to explain the ontogenetic development of an
organism that begins its life as a single fertilized egg cell and will grow
into a mature individual made up of billions of somatic cells with
different specific functions?

Thus, the idea of variation is continued, but it is no longer related to the
development of species, that is, to evolution.

The contextual analysis reveals that the antecedent of this is determined

by the two complex sentences which precede it and form a contrastive
relation to the previous part of the paragraph:

Structural weight

And occasional variations, or mutations, are nothing more than
‘misreadings’ that occur as the result of thermal fluctuations on this
level. These misreadings are preserved by means of identical repro-
duction, and if they prove to be advantageous, they will be selected. This
is the basis of evolution.

Rigid conservative forces block variation, and thus secure identical repro-
duction – a process that is occasionally interrupted by misreadings of the
genetic code, which makes philogenetic variation, that is evolution, possible.

The textual relevance of a sentence which sums up a major idea of a

paragraph is quite high. As this itself does not tell us anything about its
textual range, we need other means to help us grasp this macrostructural
discourse relation.

Making it visible

By replacing basis with fundamental process and for evolution with makes
evolution possible, the translator has succeeded in capturing the textual
relevance of the sentence about evolution. This is how it works: by means of
the extensions, we can place a focus on possible, which, in combination with
make, is something like a predicate to evolution:

This is the fundamental process that makes evolution

POSSIBLE

The focused possible helps us see the contrast between identical, invariant
reproduction, where evolution is impossible, and reproduction of occasional
variants, where evolution begins.

Now, nobody could prevent us from making out the textual relevance of

the simple version:

This is the basis of evolution.

within a passage that presents all the information needed for the conclusion.
But in contrast to the extended version there is nothing in the form of the
sentence that could help us along.

100

Structural weight

The processing advantage of the more explicit English version with its text

structuring clues is so obvious that we can only wonder why the German
original is less explicit:

Damit ist die Grundlage für eine Evolution geschaffen.

But there is one more element in the German original we have not yet looked
at, the main verb geschaffen/created. It refers to an achievement, that is,
to something that cannot be reached without effort. Together with the
indefinite form of evolution, the predicate contrasts the achievement to
possible failures. Contrast in this context means leaving the genetic code
unvaried. This takes us back to the conservative forces of the passage,
securing adequate textual embedding of the message.

The German original can thus afford to be more economical than the

English version and is therefore more in line with optimality than a back-
translated form:

Dies ist der grundlegende Prozeß, der Evolution möglich macht.

which suffers also from the ‘wrong’ perspective. The reframed version:

Damit ist die Grundlage geschaffen, durch die eine Evolution möglich
wird.

is unnecessarily explicit, that is, it requires more processing effort (than the
original) without any additional gains.

As English constrains the indefinite use of evolution, the discourse

structuring effect of the predicate in an analogous translation:

This has created the basis for evolution.

is weaker than that of the original or that of the target version: evolution will
be interpreted as given and not contrasted with other possibilities. Once
more the difference in the explicitness of original and the translation can be
said to secure processing ease according to the specific conditions of source
language and target language.

Not all extensions can, however, be licensed by language-specific pro-

cessing conditions.

A comment

With all the structural expansions, insertions and replacements in the
interest of optimal processing, we may eventually cross the border from
translating to commenting.

Structural weight 101

Leicht vergessen Kritiker, daß Gesetzmäßigkeiten immer zwischen den
Ereignissen liegen.

(EW 1983a: 303)

in its brevity an almost cryptic sentence, has been translated as:

Critics forget too readily that while laws explain how events occur, they
cannot explain specific events themselves.

(EW 1983b: 269)

Following the Maxim, this is clearly a comment on what the German
original means and not a translation restructuring the original in line with
the Principle of Optimal Relevance.

However, an analogous translation leaves many questions open:

Critics forget too readily that laws always lie between the events.

The sentence refers to critics of Chomsky’s linguistics, and Eigen and
Winkler say in the following that this type of linguistics applies to language:

in the same way as thermodynamics does to the weather. Weather is
determined by a certain condition of temperature and pressure that
follows laws of thermodynamics. Although we have long understood
these laws, long term weather predictions still depend on luck because
the boundary conditions affecting weather are so complex and so
difficult to ascertain.

(EW 1983b: 269)

Now, this example differs from all the preceding ones as the processing
obstacles in the translation are not presented by the different properties of
the English language, but are already there in the German original, which we
can only understand when we read through the comparison following the
sentence.

What we can understand from the ad hoc metaphor of the original may be

something like: laws are never identical with events. Instead of offering an
explanation for the meaning of the original, one could demetaphorize it:

Critics forget too readily that laws are never identical with events.

and wait for the clarifying exemplification. The criterion of processing ease
allows improving upon analogous translations only when the improvement
is licensed by the different processing conditions of the target language.

The example about evolution has shown the importance of discourse

relations for the question of explicitness and its language-specific nature. If

102

Structural weight

the textual relevance of an element is to become more visible one ought to
find a form that helps the reader identify the macrostructural discourse
relation in which the element participates. Clausal extensions with material
that is contextually implied may provide the appropriate information
structural frame. What in particular one can use to this effect depends upon
the special properties of each individual case, but there are a handful of
lexical and structural means in English and German which follow
grammaticalized patterns whose special function it is to signal textual
relevance of focused elements. Let us take a closer look at these focusing
devices.

Grammaticalized clues

103

The classical focusing devices are what grammar books call ‘cleft sentences’,
which ‘cleft’ a clause into two clauses, making use of the copula ‘be’ and
some pronominal dummies like ‘it’ and ‘what’.

Extending a simple sentence like:

One goes by the face, generally.

into:

The face is what one goes by, generally.

we get Alice’s polite answer to Humpty Dumpty’s reproach:

I shouldn’t know you again if we did meet. You are so exactly like other
people.

(Carroll 1968: 282)

Alice’s answer presents one type of cleft, namely the inverted pseudo-cleft.

Without the inversion, she could have said:

What one goes by, generally, is the face.

which is considered the basic form of a pseudo-cleft. But she could even have
used a third type of cleft:

It is the face you go by, generally.

which is the prototypical cleft, mostly referred to as it-cleft.

However, within the context the sentence is used in, there is only one

optimal cleft: the one used by Lewis Carroll, although until today it is an
issue of debate to say what makes the one or the other cleft a better match for
a certain context.

Regarding translations, the issue becomes more complicated. Even if the

104

Grammaticalized clues

other language, as for example German, offers similar structures in many
cases, they will not always be used. Thus, many of the clefts occurring in
English translations will not have been clefts in the German original. But as
clefts are focusing devices, their use in translations between German and
English is likely to be controlled by the same set of language-specific
processing conditions that played a role in the previous cases of structural
extensions. Let us again proceed empirically and look at some examples.

Talking about television, the ‘zero medium’ as he calls it, Enzensberger

presents a sample of typical television programmes and resumes the medley
with the sentence:

Neu an den neuen Medien ist die Tatsache, daß sie auf Programme nicht
mehr angewiesen sind.

(E 1982: 95)

What’s new about the new media is that they no longer depend on
programs.

(E 1997: 309)

There was, he says, already a similar tendency in the printed media:

Den entscheidenden Fortschritt haben jedoch erst die elektronischen
Medien gebracht.

(E 1982: 95)

However, it was only with the electronic media that the decisive advance
was made.

(E 1997: 310)

Yet, the viewer knows ‘that he is dealing not with a means of communi-

cation but with a means for the refusal of communication’ and:

Gerade das, was ihm vorgeworfen wird, macht in seinen Augen den
Charme des Nullmediums aus.

(E 1982: 100)

In his eyes, it is exactly what it is accused of that constitutes the
attraction of the zero medium.

(E 1997: 315)

In all three examples the translator has introduced ‘cleft sentences’.

It-clefts

The main reason for the introduction of cleft sentences is a difference
between German and English with which you are, by now, already quite
familiar: the restricted options of English for topicalizing an element from its

Grammaticalized clues 105

basic position inside the sentence. As in the first example, German can
topicalize the object to secure focus for the topicalized phrase as well as for
the verb-adjacent subject:

Den entscheidenden F

ORTSCHRITT

haben jedoch erst die elektronischen

EDIEN

gebracht.

Let us first look at the context to see that this is indeed a case of focused

topic. Describing the condition of the zero medium as the true destiny of the
new media, Enzensberger mentions a similar tendency of the old printed
media, especially the tabloid press, picture novels, and illustrated magazines.
Thus, we have to defocus the structural focus exponents and place a narrow
focus on the adjective entscheidend/decisive in the object and on electronic in
the subject:

Den

ENTSCHEIDENDEN

Fortschritt haben jedoch erst die

ELEKTRONISCHEN

Medien gebracht.

But irrespective of this contextual reanalysis we have a focus in the topic and
a focus in the verb-adjacent argument at the end of the sentence, and we have
two linguistic means of indicating these foci in German: topicalization of the
object, and the focusing element, erst/only, before the subject, which, being
definite and contextually given, might be otherwise defocused.

As the focusing element, however, marks the subject as contrastively

focused and it is normal to expect the contextually determined focus in the
classical, verb-adjacent position at the end of the sentence, the basic order
could have been used for this focus structure, too, if it had not been for the
textual relevance of the sentence. The contrast established by this sentence
reaches back across the entire preceding passage, shifting the discourse topic
from some advance to decisive advance. It is this contextual relevance that is
captured by the focusing effect of topicalization in German.

English cannot topicalize the object:

*However, the decisive advance only the electronic media have made.

But sticking to the basic order will yield all sorts of translational or
processing problems. The active version:

However, only the electronic media have made the decisive advance.

puts the definite subject in the background and alters the semantics of the
sentence – advancing are patients and not agents of an event. If one corrects
the perspective:

However, the decisive advance was made with the electronic media.

106

Grammaticalized clues

the additional focus in the topic may be missed. The translation gets no
closer to the original if we add the focusing adverb:

However, the decisive advance was made only with the electronic media.

Still, one could topicalize the adverbial, which is one of the subject–verb
inverting adverbials. But the inversion marks a contrastive focus structure:

Only with the electronic media was the decisive advance made.

which places everything but the first argument in the background. This is
definitely not what the original says. And fronting the adverbial without an
inversion:

Only with the electronic media the decisive advance was made.

produces a very much lop-sided, difficult-to-process structure with two foci
before the finite verb (without anything to mark the second focus) and
nothing after the verb to balance the information structure.

Thus, the cleft which offers a clause for each focus is a welcome

expansion:

However, it was only with the electronic media that the decisive advance
was made.

Linguistically seen, clefts can have a focus in either clause. Whether or not
this is used, each clause has a grammatical focus exponent which is
determined in exactly the same way as in all other clauses with focus, the
extension of focus by projection – that is, wide or narrow focus – included.

Either clause offers one verb-adjacent focus exponent, and restricts the

focus contextually to the respective adjectives:

However, it was only with the

ELECTRONIC

media that the

DECISIVE

advance was made.

Focusing a clausal subject

The second it-cleft is a little more complex since it contains a clausal subject.
The focusing element gerade indicates a focused subject, which leaves room
for another focus – that is, the focus exponent at the end of the sentence:

Gerade das was ihm

VORGEWORFEN

wird, macht in seinen Augen den

HARME

des Nullmediums aus.

This time, the focused topic of the original is the subject of the sentence, and
it is only the particle gerade that indicates its focused status.

Grammaticalized clues 107

Reviewing various complaints about television and its development into a

perfect zero medium that no longer depends upon programmes, Enzensberger
contrasts the idea of the manipulated viewer with the one who manipulates
the medium in order to enforce his own wishes. Of all the things that could
constitute the attraction of television for the viewer it is exactly its lack of
content which the viewer enjoys, despite all critical theories about the
stupefying effects of television. The interpretation of the focused topic
relates back to all the complaints about television presented at the beginning
of the essay, which according to its subtitle tells us ‘Why all complaints
about television are pointless’. That is, the contrast expressed in our example
is the main contrast of the essay, which associates the sentence with a
particularly high contextual relevance.

An analogous, non-cleft translation:

Exactly what it is accused of constitutes in his eyes the attraction of the
zero medium.

(what incorporates the pronominal head of the German version: das, was),
or one which avoids the usual attachment problems with the medial
adverbial:

In his eyes, exactly what it is accused of constitutes the attraction of the
zero medium.

is not only relatively difficult to process with its structural overload before
the finite verb of the main clause, but does not give us any chance to
recognize the textual relevance of the sentence. With the focused subject
exactly what it is

ACCUSED

of we may see no need to look for yet another focus

and interpret the rest of the sentence as background, missing the key contrast
and thus the high discursive significance of the sentence.

Again, clefting is a solution:

In his eyes, it is exactly what it is

ACCUSED

of that constitutes the

ATTRACTION

of the zero medium.

Pseudo-clefts

Referring to ‘medium’ or ‘program’ as ‘anachronistic concepts’, which no
longer apply to the mixture of features announced as ‘programs’ (of which
the preceding passage of the essay presents an impressive sample),
Enzensberger contrasts the new media to the old concept of the medium as a
mediating means, and the original concept of programme as spoken of in
public life ‘when the principles of intended action are announced in advance
in more or less binding form’ (E 1997: 309).

While the old concept of mediating media involves the idea of programme,

108

Grammaticalized clues

the new media no longer depends upon programmes. The contrast is
formulated as a conclusion from the striking illustration in the preceding
passage, but it relates back three passages, to the beginning of this section:

The concept of the medium is an old one; . . .

Thus, the local contrast within the sentence itself participates in a wider
contextual contrast, and the topicalization signals textual relevance of the
contrasted topic as in the other two examples:

an den neuen Medien ist die Tatsache, daß sie auf Programme nicht

mehr

ANGEWIESEN

sind.

An analogous translation with the predicate in initial position could be
interpreted as contrastive, too, but as the following considerations will
show, the contrast:

New about the new media is that they no longer depend on programs.

misses the textual relevance of the original.

Needless to say the basic order:

That they no longer depend on programs is new about the new media.

places the clausal subject in the background, missing the focus structure of
the original. But the reversed order in the cleft does capture the focus
structure of the original with a contrast on new and a focus on the negated
predicate no longer depend:

What’s new about the new media is that they no longer depend on
programs.

Unlike the first two cleft examples, there is no lexical focusing element (such
as only, exactly) that would push everything but its own focus into the
background in a non-cleft version. Yet the cleft is more in line with
optimality. But why? What is it that the cleft adds to the focus structure of
the sentence?

Presupposing and contrasting

Clefts are assumed to presuppose the meaning of their sentence – except for
the focused element which is specified by the cleft.

For our example, this

means that the meaning, there is something that is new about the new media,
is presupposed, with something being specified by the cleft as they no longer
depend on programs.

Grammaticalized clues 109

As the focus is contrastive, the specification contrasts with alternative

possibilities, saying that it is this and nothing else (that is new about the new
media). And the specification of the cleft is, indeed, in contrast to what we
would expect, all the more so as the author had reminded us of the fact that
‘the concept of the medium . . . indicates first of all simply something
middling, mediating’ (E 1997: 308) and the concept of programme is meant
to announce future actions in more or less binding form. Surely, everybody
expects all media, whether printed or electronic, to depend on programmes,
and nobody would expect the novelty of the new media to consist in a lack of
programmes. But the lack of programmes is the defining discourse theme of
the essay about the ‘zero medium’. Thus the specification of what it is that is
new about the new media, namely their independence from programmes,
does, indeed, carry the highest textual relevance imaginable.

Despite their explicitness, clefts offer a very economical device to high-

light the textual relevance of a discourse element for they add no more
content than the identifying relation expressed by the copula ‘be’ and the
pro-forms – a classical structural dummy. With the constraints of a configur-
ational language on focusing by word order, clefts are particularly welcome
in English. No wonder that there are many more cleft-like structures
securing easy focus identification,

as for example this very sentence, with

the copula after there and an -ing-complement after the predicate noun.

Some of these devices exist in German, too, but they are subject to

different conditions of use, and the translator has to decide when to take
which cleft or cleft-like structure and when to make use of marked word
order or of a focusing particle. But the main line is clear, isn’t it? And what
Alice would say is this: processing ease is what one goes by, generally.

In addition to clefts, there are also certain lexical elements that offer

grammaticalized, and in particular lexicalized, means to indicate textual
relevance. All these means constitute a subclass of what I would like to call
‘discourse relators’ (with or without macrostructural relevance).

Lexical discourse relators exist in both English and German, but in

contrast to clefts, they are used more often in German. As this may lead to
structural reductions in translations from German into English, we will now
turn to examples illustrating the other side of the structural weight coin.

Dispensable information

As the following examples will show, there are lexical elements which are
semantic dummies, but which may indicate additional foci in a sentence. The
fact that they are used more frequently in German than in English can, again,
be explained by the parametrized processing conditions of focus identifi-
cation. There are basically two classes of such elements, which will be taken
up in this section and the next: first, modal particles, which underline the
segmentation of a sentence into focus and background; and second, focusing
particles, which assign a contrastive focus to an element in their scope.

110

Grammaticalized clues

The first type are ‘particles’ like ja, denn, eben, eigentlich, nämlich,

durchaus, for which there are hardly any lexical correspondences in English.
Take, for example, the notoriously empty particle, ja, which adds no more to
its sentence than a cooperative nod. A sentence like:

Vielleicht ist er ja nie etwas anderes gewesen . . .

(E 1999: 160)

sounds very much like:

Vielleicht ist er nie etwas anderes gewesen . . .

and its translation into English as:

Perhaps it has never been more . . .

(E 1997: 338)

does not appear to miss anything. But ja cannot appear in a question:

*Ist er ja nie etwas anderes gewesen?

which means that it does have a meaning, incompatible with that of
questions (which have an open epistemic meaning). We can explain the
constraint of ja against questions if we assume that ja presupposes that the
attitudinal object in its scope, that is, the thought it is associated with, had
existed before.

Ja marks the information of its sentence as something that is already

known, even if it is presented as a mere possibility:

Vielleicht ist er ja nie etwas anderes gewesen . . .

At the same time, ja indicates that the element following it carries a focus,
which is obviously an additional focus emphasizing nie in addition to
whatever is presented as focus exponent in the predicate.

Now, we could paraphrase ja by something like:

As we know, it has perhaps never been more . . .

But this would not indicate a focus on never and as such an explicit appeal to
hypothetical knowledge feels definitely redundant, we will readily accept the
translator’s decision to drop it. In contrast to the little ja, the parenthetical
clause is heavily explicit. While ja leaves everything underspecified – except
for the presupposed assumption – the clausal paraphrase cannot but spell out
the predicate and the subject of the attitude with all their grammatical
features: tense, aspect, mood, number, person. Thus, the clausal paraphrase

Grammaticalized clues 111

is naturally overspecified, fixing all sorts of discourse values that are left
variable by ja. And while the reader can easily process the vague and
underspecified particle, (s)he has to invest more effort into processing the
phrase than the cognitive gains of the discourse relator are worth.

The greater explicitness, which retains the semantic function of the par-

ticle in English, but misses the focusing effect, is a processing disadvantage
which is not counterbalanced by any real gains. This was different with all
the previous cases of greater structural explicitness, where the extra effort
helped to avoid anaphoric or information structural garden paths.

In general, it is the attitudinal form of particles for which there are no

corresponding forms in English, and the syntactic paraphrases that could be
used instead will often appear disproportionate. There are other uses of ja
for which there are corresponding expressions in English, as for example:

Sie waren vielmehr verpflichtet, ja gezwungen, der Welt um jeden Preis
. . . ein exorbitantes Schauspiel zu bieten.

(E 1999: 151)

. . . the rich and powerful were obligated, even forced, to offer the world
an exorbitant spectacle at any cost.

(E 1997: 330)

Even, like ja, is used as a focusing particle of degree here, establishing a
hierarchical relation between the element in its scope and a predecessor.
Even has the same (contrastively) focusing function as ja in this context,
which means that English uses lexical discourse relators, too, although
more rarely.

Eben is another prototypical case of the author’s covert dialogue with the

reader. Eben surpasses ja in its discourse relating function as it establishes a
more specific discourse relation. While ja presupposes the meaning of its
sentence as known, eben presupposes the meaning of its sentence as known
in a concluding justification of what was said before.

In an essay on the pastry dough of time – Enzensberger’s ingenious

explanation of anachronism – we can read:

Der Blätterteig enthält eben alle Möglichkeiten . . .

(E 1999: 25)

There are no difficulties identifying the justifying discourse relation between
all possibilities and the preceding context:

The temptation to judge anachronism on moral grounds is difficult to
resist. But perhaps its actual scandalousness lies precisely in its indif-
ference to such judgements. The pastry dough contains all possibilities.

(E 1997: 48)

112

Grammaticalized clues

Demonstrating the wanderings of any given point during the formation of a
pastry dough, Enzensberger shows that a point will always:

land in a transformed context. Thus, the contact between different
layers of time does not lead to a return of the same, but to an interplay
that produces something new each time . . . It is because of time’s
complex layering that every day surprises us with good and evil . . .

(E 1997: 46)

In using eben, the German author appeals to the reader’s knowledge about
the sheer potential of the pastry dough that makes it indifferent to moral
verdicts. At the same time, eben indicates a focus on the object following it,
which is shifted to the quantifier alle by graphic means. (The original
sentence continues with a series of attributes, containing more foci.)

Dropping eben, or rather its disproportionate paraphrase in English, the

translator relies on the reader’s knowledge, too. Knowing the context, as we
do when we read the sentence about all possibilities, we can easily
reconstruct the justification from our discourse knowledge: this is so because
– as you know – The pastry dough contains all possibilities. The focusing
effect is retained graphically.

Eben can be used in other functions, too, for which lexical correspon-

dences in English may exist – as for example the same, which replaces eben
jenes:

Der Apologet des Luxus beruft sich auf eben jenes Gleichheitspostulat,
das seine Kritiker gegen ihn ins Feld zu führen pflegten

(E 1999: 147)

The apologist for luxury calls upon the same notion of equality used by
his critics to attack him.

(E 1997: 326)

The great variety

Particles with a lexical equivalent in English are used less often in English
than the corresponding particles in German. Again, requirements can be
shown to differ according to the language-specific aspects of the contextual
conditions. However, a systematic comparison of the different particles
which exist in both languages suggests that the interaction of the various
components promoting or restricting the use of a special particle does not
admit simple generalizations.

Schon, for example, means something like ‘earlier than expected’ (also

in relation to a scale). It could be retained in the English version as already,
as in:

Grammaticalized clues 113

Verdacht der Dekadenz, der schon den Römern nicht fremd war

(E 1999: 145)

traditional notions of morality, which were already old news during
Roman times

(E 1997: 325)

or be paraphrased as something with a similar meaning as in:

Schon den Biologen des 19. Jahrhunderts war aufgefallen

(E 1999: 148)

As early as the 19th century biologists noted

(E 1997: 328)

or be replaced by something with a similar discourse function as in:

Nun ist es schon aus statistischen Gründen unwahrscheinlich

(E 1999: 108)

Now, on the basis of statistics alone, it is improbable

(E 1997: 197)

or be dropped altogether:

Schon wegen der schieren Menge dieses Materials ist . . . jede andere
Lektüre ausgeschlossen

(E 1999: 111)

The sheer volume excludes any other reading.

(E 1997: 200)

The variety of different correspondences is amazing and some of it is

certainly due to the translator’s endeavour to stay very close to the original.
That is, original texts in English will have fewer discourse relators than
English translations from German – a well-known phenomenon of source
language interference, which can be observed in any area.

The cases of

deletion should be all the more revealing. Here, too, the different focus con-
ditions can explain translations that might otherwise appear quite arbitrary.

Let us look at the last two relatively similar cases. While the translation of:

Nun ist es schon aus statistischen Gründen unwahrscheinlich

replaces schon by alone:

114

Grammaticalized clues

Now, on the basis of statistics alone, it is improbable

the translation of:

Schon wegen der schieren Menge dieses Materials ist . . . jede andere
Lektüre ausgeschlossen

drops the particle:

The sheer volume excludes any other reading.

Schon focuses the first referent of either sentence: in the original, an
adverbial in the form of a prepositional phrase. That is, we have the same
structural conditions in both German sentences – a focused adverbial very
early in the sentence. The initial now of the first sentence can be ignored as it
could be added to the second sentence without altering the conditions:

Now, the sheer volume excludes any other reading.

But there is a decisive difference in the translation of the prepositional
phrases. While the prepositional phrase of the first sentence is retained in the
translation, the prepositional phrase of the second sentence is reframed as a
subject: wegen der schieren Menge . . . ist ausgeschlossen/the sheer volume
excludes . . .. The shift in perspective optimizes processing conditions for the
preverbal position in English. It provides us with a subject, which we can
easily identify as focused. Sheer is sufficient as a focus indicator for the
English subject. The initial adverbial on the basis of statistics, on the other
hand, needs an extra focus indicator.

Focused subjects which do not yet contain a focused specifier may yet

need a focusing particle in English, too, or even more structural weight.
Thus, we find a sentence like:

Gerade der Tatsache, daß er gegen alle Normen des Alltags verstieß,
verdankte der Luxus seinen Eklat und sein Prestige.

(E 1999: 155)

translated as:

It was precisely the fact that luxury broke with all the norms of everyday
life that gave it brilliance and prestige.

(E 1997: 334)

If we ignore the translating ‘liberty’ (Eklat is something like notoriety and
not brilliance), we can appreciate the optimal information structure of the
translation, which makes use of a cleft in addition to the particle precisely.

Grammaticalized clues 115

Particle and cleft help us to identify the focused subject despite its definite
nature (the fact that). If the focusing particle itself is not sufficient and has to
be strengthened by a cleft, the extension indicates a macrostructurally
relevant contrastive focus. Which we find confirmed – luxury, we read in the
preceding context:

took an unexpected and fatal turn. It conquered the field and, in so
doing, dug its own grave. . . . In every society of the past, waste and
excess were rare exceptions to the rule. It was precisely the fact that
luxury broke with all the norms of everyday life that gave it brilliance
and prestige.

Luxury, which is today the rule, used to be the exception in the past. The
textual relevance of the focus of the matrix clause is high enough to warrant
a cleft in English.

There was no cleft in the German original, though. Still, if we take a closer

look at the focused beginning of the German sentence, we find a dative
object topicalized:

Gerade der Tatsache, . . . verdankte der Luxus . . . sein Prestige.

which is, as we remember, a typical way of marking an extra focus in
German. Nevertheless, macrostructural relevance promotes a focusing
particle in German also, over and above the syntactic focus marking.

However, we are back here to examples which are structurally extended

in the translation, while what I promised to concentrate on was: what are the
language-specific conditions for structural reductions?

Processing aid or processing burden

One answer to this is simply a reversal of the answer to the question: what
are the language-specific conditions for structural extensions? If only the
source language needs a full lexical form to optimize anaphora resolution,
we can reduce the corresponding elements, for example by pronominalizing
them, in the target language. You will, certainly, remember the difference
between the signs of it and das in the German translation of the Old Man’s
reflections about hurricanes.

If a dummy is needed for grammatical or stylistic reasons in one language,

it will hardly ever be needed in the other language. ‘Stretching’ subjects,
predicates, attributes and adverbials will be welcome in different places due
to different grammatical parameters: what functions as a processing aid in
the source language will often turn out to be a processing burden in the target
language.

The critical point here is that the translator can never tell in advance

whether a linguistic element has been used mainly for reasons of optimal

116

Grammaticalized clues

processing or whether it is an indismissable element of the message itself. The
question of whether something can be dispensed with in the interest of
processing ease or is indispensable, irrespective of any processing costs, can
only be decided individually by assessing its relevance relative to the
discursive context.

Now, indicators of focus, of discourse relations and textual relevance,

that is, particles, clefts and the like, may seem especially relevant as
contributions to successful discourse.

There is but one disadvantage

associated with them: they have to be processed, too. Thus, they are
themselves subject to the Principle of Optimal Relevance, requiring a
sensible proportion between cognitive gains and processing effort.
Whenever the effort noticeably surpasses the gains, we try to get along
without the extra amount of structural explicitness. The cognitive gains of
focus indicators or discourse relators are measured in terms of easy
processing of focus and discourse relations. In many cases, both aspects of
focus and discourse relations can be computed from the contextually
embedded linguistic structures themselves, without the help of focus
indicators and discourse relators. It is the author’s decision to indicate foci
and discourse relations or leave it to the readers to figure out for themselves.
The author can even choose between more or less explicit forms: full clauses,
phrases, longer or shorter words, such as du weißt schon/you know,
nichtsdestoweniger/nonetheless, überhaupt/after all, also/thus, etc.

Furthermore, instead of asking a question directly, the author can say: this

raises a question; it is by no means clear whether . . .. The decision is very
much a matter of personal style, but if we hold on to the Maxim, we are
ready to respect the author’s decision – relative to the target language
conditions. However, it may not always be so easy to distinguish between the
author’s personal style and the special conditions of the source language. Let
me illustrate this by a more complex example, where the translation has
reduced the original quite radically.

Ambiguities

Es ist nämlich keineswegs klar, wer in Zukunft eigentlich zu den
Nutznießern des Luxus zählen wird.

(E 1999 : 160)

says the author and the translator turns it into a mere question:

Who will count among the beneficiaries of luxury in the future?

(E 1997: 338)

dropping eigentlich in the interrogative clause, nämlich in the matrix clause
and with it the entire matrix clause.

Let us ignore the deletion of the matrix clause for the moment and

Grammaticalized clues 117

concentrate on the particles. Nämlich and eigentlich are discourse relators of
the attitudinal type, that is, they relate the meaning of their sentence to other
attitudes or attitudinal objects. Nämlich tells the reader that this (the
message containing nämlich) is what he has to know to understand a pre-
vious claim. There is no correspondent element for nämlich in English, and if
it is substituted by a clausal paraphrase, it would have to be something like:

One has to know that it is by no means clear who . . .

But the extended matrix clause is overspecified and disproportionate in
English; the clausal paraphrase of nämlich cannot compete with the concise
and cordially vague German particle. Dropping nämlich or, rather, its
awkward paraphrase from the translation is thus licensed by the Principle of
Optimal Relevance.

We are left, then, with eigentlich in the subclause. Eigentlich has also been

dropped from the translation, but it could have been retained in English, at
least for one of the two interpretations it admits. The ambiguity resides in the
focus structure of the sentence. The example belongs to a longer passage
about future luxury, which says that the priorities will shift to the ‘element-
ary necessities of life . . .: quiet, good water and enough space’ (E 1997: 335).
Talking about the future continuation or disappearance of various aspects of
luxury, such as its ‘withdrawal from reality’ or its ‘role in representation’, the
essay turns to yet another aspect arising from the reversal of priorities:

New and bewildering, however, is another question that must be posed
in light of future prospects: Who will count among the beneficiaries of
luxury in the future?

or in the original:

Neuartig und verwirrend ist eine andere Frage, die sich bei solchen
Aussichten stellt. Es ist nämlich keineswegs klar, wer in Zukunft
eigentlich zu den Nutznießern des Luxus zählen wird.

Now, there are basically two focus interpretations of the interrogative

clause, in addition to the inherently focused question word. If the reader
recalls what (s)he knows about the beneficiaries of luxury in the past and the
present, it will more or less coincide with what (s)he has read in the essay. It
was the rich and the powerful, who could or even had to ‘put on orgies of
extravagance’ (E 1997: 330). But since the question implies the possibility of
an alternative, namely, that there are other beneficiaries than those of the
past and the present, it could suggest a contrastive focus on eigentlich:

Es ist nämlich [keineswegs

KLAR

WER

in Zukunft

EIGENTLICH

zu den

Nutznießern des Luxus zählen wird.

118

Grammaticalized clues

This interpretation would need something like an extra real in English that
could attract the contrastive focus:

will count among the

REAL

beneficiaries of luxury in the future?

Obviously, the translator has decided against this option (which the author
tells me would have been his option).

On the other hand, there has been no explicit reference to the rich and

powerful in the context for quite a while. The readers could therefore
consider the beneficiaries as a new aspect of the reflections about luxury in
the future and hence as focused information. Thus, they could also interpret
the sentence as:

Es ist nämlich [keineswegs

KLAR

WER

in Zukunft eigentlich zu den

UTZNIESSERN

des Luxus zählen wird.

This interpretation is not possible for the English version with real, but it is
possible for the English version without a corresponding element for
eigentlich, which is the one the translator has opted for.

Structurally, this version offers only the question word as focus:

will count among the beneficiaries of luxury in the future?

Almost everything else could be given due to its definiteness. In contrast to
the German original, the English version does not offer any processing aid,
hence the focus interpretation has to be figured out contextually. If one used
a subclause without a particle in German:

Es ist nämlich keineswegs klar, wer in Zukunft zu den Nutznießern des
Luxus zählen wird.

one would have the same problem. The processor would take the definite-
ness of the internal argument as an indicator of given information and
restrict the focus to the question word:

. . .

WER

in Zukunft zu den Nutznießern des Luxus zählen wird.

The interpretation has to be reanalysed contextually.

The domino effect

While the particle of the matrix clause was ‘dropped’ in the translation
because English lacks a lexical correspondent, and the clausal paraphrase
violates Optimal Relevance, the particle in the subclause has disappeared
because the translator has opted for the non-contrastive interpretation of

Grammaticalized clues 119

eigentlich. But he has deleted the matrix clause, too, and reduced the
complex sentence of the German original to a mere question. Instead of the
analogous sequence:

New and bewildering, however, is another question that must be posed
in light of future prospects. It is by no means clear who will count among
the beneficiaries of luxury in the future.

we read:

New and bewildering, however, is another question that must be posed
in light of future prospects: Who will count among the beneficiaries of
luxury in the future?

Now, if I pose a question, something is in need of clarification. That is, in
asking the question: Who will . . .? I imply that it is not known who will . . . ,
which, in turn, implies that it is not clear who will . . .. Would the semantic
repetitiveness not be reason enough to translate the question without the
matrix clause?

The first answer is ‘no’. The semantic relations are the same in the German

original, and they did not prevent the author from using the matrix clause.
But then the German matrix clause contains a particle telling us why the
question that must be posed is new and bewildering. Without this discourse
link, the English matrix clause is indeed reduced to a trivial aspect of
questions, which may have prompted the translator to drop the matrix
clause altogether. Raising the status of the interrogative clause from a
subordinated clause to an independent sentence, the translation may even
help us to recognize the focus on beneficiaries, though we could have solved
this problem more elegantly by choosing the contrastive version:

New and bewildering, however, is another question that must be posed
in light of future prospects: Who will count among the real beneficiaries
of luxury in the future?

However, clarifying focus assignment for the question does not change the
situation for the matrix clause. Missing the explanatory link, the matrix
clause feels redundant:

New and bewildering, however, is another question that must be posed
in light of future prospects. It is by no means clear who will count among
the beneficiaries of luxury in the future.

and the translator’s deletion of the matrix clause justified by the domino
effect.

German indicates and spells out discourse relations more often than

120

Grammaticalized clues

English. Should we consider German speakers as being more cooperative
than English speakers? Or as being more in need of processing aid? And only
therefore, in line with Optimal Relevance, more cooperative?

If we think of the basic difference of a subject-verb-object language, like

English, and a subject-object-verb language, like German, we can make out a
processing disadvantage for German which does not exist in English. The
left-peripheral English verb phrase can rely on the verb as a natural, early
clue to the semantic and pragmatic relationships in which the elements of a
sentence participate, including questions of focus. In German, the processor
has to wait till the end of a sentence before it really knows all about the
semantic and pragmatic functions of its constituents and can identify the
structural focus exponent. Thus, additional clues which indicate discourse
relations and the beginning of focus – that is, the border between focus and
background (or mark additional foci) – are welcome in German. But as they
carry their lexical meaning, however faint it may be, retaining/substituting
or deleting them is again an act on the tightrope.

Spelling out the discourse relations and contextual foci in each individual

case may be just as difficult as finding an appropriate substitute, if this is
necessary. Achieving a balanced proportion between ‘cognitive gains’ and
‘processing effort’ presupposes a great deal of highly sophisticated linguistic
knowledge, whether intuitive or ‘learned’. Some general typological insights
into the language-specific, parametrized processing conditions might at least
point the way.

Shifting boundaries

121

Reordering, reframing, extending, reducing – so far we have encountered
an impressive variety of structural changes which were licensed by
parametrized processing conditions determining optimal relevance in
German and English. But even if these changes were cross-clausal, they
occurred within the borders of sentences. It was only in the last example,
with its ‘liberal’ deletion of the matrix clause, that sentence boundaries were
affected. The result of the reduction, the question, had been adjoined to the
preceding sentence by a colon.

Sentence boundaries are determined by full stops. Between two full stops,

linguistic structures contribute to a simple or complex information unit.
Inside this unit, punctuation marks may delimit subunits with different
degrees of cohesion. Commas indicate the highest degree of cohesion,
dashes, colons, semicolons and brackets a much lower one, though colons
and semicolons are punctuation marks signalling a closer relation between
the structures they separate than a full stop.

Now, it is the authors’ decision to present their thoughts in one way or

another, but the Principle of Optimal Relevance restricts their freedom, and
the constraints controlling punctuation are to a certain extent language
specific. Some of the differences between the punctuation of an original and
its translation will merely result from the structural changes necessary in the
interest of easy processing. The last example could have been such a case if it
had not started from the wrong premises. Convincing examples of different
sentence boundaries are by no means rare, however.

Basically, we can distinguish three types of boundary shifts: linking of

independent sentences into more complex sentences, separation of clauses
into independent sentences, and the shift of sentence boundaries which
combines linking and separating. All three types occur in translations from
German into English, and the following examples of each of the three types
clearly result from optimizing processing conditions under the specific
constraints of the target language.

122

Shifting boundaries

Linking

Sometimes we may share the feeling that a sentence does not merit being a
sentence in its own right. Consider for example the sequence:

The names of producers have become a universal code. The label
replaces the object. This has gone so far that customers routinely offer
the companies their bodies as advertising space.

which could leave us a bit breathless around its chopped up appearance in
the middle. This must have prompted the translator to spell out the relation
between the first and second sentences as causal (or temporal) and link the
corresponding clauses accordingly:

The names of producers have become a universal code, as the label
replaces the object. This has gone so far that customers routinely offer
the companies their bodies as advertising space.

(E 1997: 334)

But the German original consists of three independent sentences:

Die Namen der Hersteller sind zu einem universellen Code geworden.
Das Etikett vertritt den Gegenstand. Das geht soweit, daß die
Kundschaft ihren Körper den Lieferanten als Werbefläche zur
Verfügung stellt.

(E 1999: 155)

And for good reason: if we were to link the first two sentences in German, we
would assign too much relevance to the second clause and present it as the
main information of the entire sentence. This applies to the causal link:

Die Namen der Hersteller sind zu einem universellen Code geworden, da
das Etikett den Gegenstand vertritt.

as well as to the temporal link, which fares even worse because the focused
position turns it into a conditional link:

Die Namen der Hersteller sind zu einem universellen Code geworden,
wenn das Etikett den Gegenstand vertritt.

The interpretation with a focused, second clause is a possible view, but it
is definitely not the author’s view. In the original, the second sentence
mentions only another aspect of the same state of affairs.

If one were to insist on the clausal status of the second aspect, one would

have to lower the relevance of the subclause. One could use another con-
nector, marking the discourse relation as an elaboration:

Shifting boundaries 123

Die Namen der Hersteller sind zu einem universellen Code geworden,
wobei das Etikett den Gegenstand vertritt.

Here, however, the resultative aspect of the main clause does not go too well
with the dynamic nature of wobei. Not only does the last version miss the
original idea, it violates selection restrictions. The comparison shows that in
German, separation of the two clauses is much better:

Die Namen der Hersteller sind zu einem universellen Code geworden.
Das Etikett vertritt den Gegenstand.

As English expects the focus near the main verb of the main clause, there are
no difficulties in subsuming the informational value of the second clause to
the first (and interpreting the subclause temporally):

The names of producers have become a universal code, as the label
replaces the object.

For a new sentence, however, less relevance is unexpected and would be
legitimate only if the information could not be incorporated in the preceding
sentence, perhaps on account of its structural length or for other reasons. For
a short and transparent sentence like:

The label replaces the object.

there are no reasons for it to be used independently.

Linking by resumption

Sometimes two sentences can only be linked by taking an element of the first
sentence as a structural anchor for the second sentence. With a sequence like
the following it is unclear how to interpret the second sentence:

The modern concept of time runs aground when confronted with the
most obvious of data. It is almost embarrassing to mention it. Our
genetic code originated billions and billions of years ago; . . .

The major problem is posed by the two its of the second sentence, which

could be anaphors with various antecedents in the first sentence, or – in the
case of the first it – expletive, without any reference. But even if we take the
first it to be an expletive pronoun, we will be at a loss as to the antecedent of
the second it. What in the sentence:

The modern concept of time runs aground when confronted with the
most obvious of data.

124

Shifting boundaries

is referred to by the pronominal object of the second clause:

It is almost embarrassing to mention it?

Yet in terms of punctuation, the English version would be an analogous
translation of the German original:

Das scheitert schon an den unvermeidlichsten aller Gegebenheiten. Man
geniert sich fast, sie zu erwähnen. Unser genetischer Code ist vor
Milliarden und Millionen von Jahren entstanden; . . .

(E 1999: 13)

Now, the existing translation avoids the anaphora problems of the
analogous version by linking the second sentence to the first:

The modern concept of time runs aground when confronted with the
most obvious of data, so obvious that it is almost embarrassing to men-
tion it. Our genetic code originated billions and billions of years ago; . . .

(E 1997: 37)

Repeating the adjective of data and extending it by a connector, so obvious
that, disambiguates the reference. And as an independent version: it is so
obvious that it is almost embarrassing to mention it, is too repetitive even for
English ears, linking the second sentence to the first is a welcome way out.
One can even avoid the explicit relative clause:

The modern concept of time runs aground when confronted with the
most obvious of data, which is so obvious that it is almost embarrassing
to mention it.

by an apposition:

. . . when confronted with the most obvious of data, so obvious that it is
almost embarrassing to mention it.

The anaphoric problem does not exist in German whether one interprets

the pronominal object sie as a plural or as a singular – the latter in cataphoric
anticipation of the genetic code specifying the unvermeidlichsten aller
Gegebenheiten/the most inevitable of all conditions (the most obvious of
data). Thus, one does not have to link the sentences in German in order to
disambiguate the referential relation.

Nor would we want to link the sentences in German. Even if we accepted

the semantic compromise needed for the link and replaced unvermeidlich/
inevitable by offensichtlich/obvious, we could not like the result as much as
the original. The sentence:

Shifting boundaries 125

Das scheitert schon an den offensichtlichsten aller Gegebenheiten, so
offensichtlich, daß man sich fast geniert, sie zu erwähnen.

fails to meet German focus expectations. This is so because the second clause
is of little relevance compared with the first clause. It could be deleted
without any effect whatsoever on the cohesion of the remaining discourse:

The modern concept of time runs aground when confronted with the
most obvious of data. Our genetic code originated billions and billions
of years ago; . . .

Placing an informational element of such little relevance in the canonical
focus position of a German sentence violates the Principle of Optimal
Relevance. The processor may mistake the focus of the complement clause:

so offensichtlich, daß man sich fast geniert, sie zu erwähnen.

for the focus of the whole sentence, and will have to correct this afterwards.
It is clearly the superlative, den offensichtlichsten/unvermeidlichsten aller
Gegebenheiten, which is the contextual focus.

Linking the first and the second sentence produces an informational unit

with decreasing informational values. While this is a distribution to which
English – with its left-peripheral verb phrase and corresponding focus
expectations does not object – it will in most cases be restructured in
German; if reordering with or without reframing cannot be used, one will
rather separate the discursive structure and use independent sentences as the
original does.

Separating

While lower relevance of a subsequent sentence in the German original may
suggest the linking of two independent sentences in the English translation,
an alternative distribution of informational values, that is, a steady increase
of relevance, may create exactly the opposite condition for sentence
boundaries.

The following two sentences:

On the other hand, the unemployed, the elderly, and refugees, who in the
future will make up the majority of the world’s population, usually
dispose of their time as they like. But it would be sheer mockery to call
that a privilege.

(E 1997: 338)

go back to one sentence in the original:

126

Shifting boundaries

Umgekehrt können Arbeitslose, Alte und Flüchtlinge, die zusammen in
naher Zukunft die Mehrheit der Bevölkerung ausmachen werden, in der
Regel beliebig über ihre Zeit verfügen, aber es wäre der blanke Hohn,
darin ein Privileg zu sehen.

(E 1999: 160)

If we translated this as one sentence in English, too, it would not only be
difficult to process, but the contrast in the end would be devalued:

On the other hand, the unemployed, the elderly, and refugees, who in the
future will make up the majority of the world’s population, usually
dispose of their time as they like, but it would be sheer mockery to call
that a privilege.

The main focus is expected earlier in the sentence and assigned to the finite
verb of the previous clause and its argument. If one wants to assign more
relevance to the last clause under the condition of a left-peripheral focus
structure, the final clause has to be separated.

In German, the situation is the opposite. Separating the last clause

from the sentence attributes a higher value to the preceding clause. This
would make it difficult to recognize the real importance of the second
sentence:

Umgekehrt können Arbeitslose, Alte und Flüchtlinge, die zusammen in
naher Zukunft die Mehrheit der Bevölkerung ausmachen werden, in der
Regel beliebig über ihre Zeit verfügen. Aber es wäre der blanke Hohn,
darin ein Privileg zu sehen.

The original attributes the highest value to the last clause, which distributes
the relevance of the contrastive clauses optimally:

Separation alone is not enough if the critical segment is no clause. Thus,

the second sentence in:

What began as a struggle between the ancient and modern ages grew
into a protracted battle between the inherited past and the revolutionary
new, between tradition and the modern, and finally became a given. It
has survived into the present as an either/or, first in cultural setting, and
then in the political arena.

(E 1997: 35)

Shifting boundaries 127

corresponds to a nominal apposition after a dash:

Was damals als Streit zwischen Antike und Neuzeit thematisiert wurde,
hat sich dann immer weiter entfaltet, bis der Kampf zwischen dem
Althergebrachten und dem umwälzend Neuen, zwischen Tradition und
Moderne, zur Selbstverständlichkeit wurde – einem Entweder-Oder,
dem zunächst in der Kultur, bald aber auch in der Politik eine lange, bis
heute andauernde Karriere beschieden war.

(E 1999: 11)

The nominal head of the phrase is a nominalized compound of two
connectors: Entweder-Oder, with a long and highly involved attribute
clause. The phrase has been extraposed behind the finite verb due to its sheer
length. It is part of the prepositional phrase, zur Selbstverständlichkeit, and
shares its preposition; zu einem Entweder-Oder is formally the last and most
important argument of the verb within the temporal clause that spells out the
development of the protracted battle between the inherited past and the
revolutionary new.

Besides rephrasing the matrix clause and the adverbial of the first

sentence, English separates the postposed apposition into a second sentence.
This requires filling in the subject and the verb of the original apposition. The
translator has opted for a wider paraphrase, replacing dem eine bis heute
andauernde Karriere beschieden war by it has survived into the present.

The changes secure easy focus identification in both sentences. If the

second sentence were linked to the first, for example as a relative clause, or as
a participle phrase, we might not be able to grasp the discursive relevance of
this attribute. There are two possibilities. With a restrictive clause for the
specification of the survival, the preceding information, the head of the
attribute, will be devalued:

What began as a struggle between the ancient and modern ages grew
into a protracted battle between the inherited past and the revolutionary
new, between tradition and the modern, and finally became a given
which has survived/surviving into the present as an either/or, first in
cultural setting, and then in the political arena.

With a non-restrictive clause, the attribute itself will be devalued:

What began as a struggle between the ancient and modern ages grew
into a protracted battle between the inherited past and the revolutionary
new, between tradition and the modern, and finally became a given,
which has survived/surviving into the present as an either/or, first in
cultural setting, and then in the political arena.

Separating the attribute from its head as an independent sentence allows us

128

Shifting boundaries

to assign an adequate focus to both sentences and express the progressive
discourse at the same time.

What began as a struggle between the ancient and modern ages grew
into a protracted battle . . . and finally became a given. It has survived
into the present as an either/or, first in cultural setting, and then in the
political arena.

Felicitous contrasts

The different word order conditions in English and German may dictate
further structural changes of the segments separated into independent
sentences. Thus:

Fast cars and gold watches, cases of champagne and perfume are
available on every street corner; they are not scarce, rare, expensive or
desirable in this age of raging consumerism. Instead, it is the elementary
necessities of life that come at a great price: quiet, good water, and
enough space.

(E 1997: 335)

has reordered the first sentence in line with the subject-before-verb principle.
The German original begins with the predicate:

Knapp, selten, teuer und begehrenswert sind im Zeichen des wuchern-
den Konsums nicht schnelle Automobile und goldene Armbanduhren,
Champagnerkisten und Parfums, Dinge, die an jeder Straßenecke zu
haben sind, sondern elementare Lebensvoraussetzungen wie Ruhe, gutes
Wasser und genügend Platz.

(E 1999: 157)

But then the apposition to the subject things which are available on every
street corner has been syntactically and informationally raised to replace the
original predicate of the sentence. The latter is separated into a clause of its
own after a semicolon:

Fast cars and gold watches, cases of champagne and perfume are
available on every street corner; they are not scarce, rare, expensive or
desirable in this age of raging consumerism.

If one contented oneself with reversing the subject and predicate of the
original, one would get:

Fast cars and gold watches, cases of champagne and perfume, which are
available on every street corner, are not scarce, rare, expensive or
desirable in this age of raging consumerism.

Shifting boundaries 129

By subordinating the availability of the traditional luxury goods in a
preverbal position, one spoils the author’s point because it is the general
availability that explains why those goods are no longer rare and desirable.

The German original has this information at the end of the first

contrastive segment and thus in its focus position, whose value is only
surpassed by the subsequent element of the contrast:

nicht . . . Dinge, die an jeder Straßenecke zu haben sind, sondern
elementare Lebensvoraussetzungen . . .

But this is the optimal form of the contrast in German, and it requires much
restructuring to become optimal in English, too.

The predicate and the apposition of the original are in a contrastive

relation to each other, the negated part of which is normally used first, the
affirmative part afterwards. Exactly this is secured in German by using the
negated predicate in topic position and the affirmative part in form of an
apposition after the subject at the end of the first clause.

The subject-before-verb principle does not allow the same solution in

English, but one could achieve a similar effect by using the normal order
within the predicate:

Fast cars and gold watches, cases of champagne and perfume are not
scarce, rare, expensive or desirable in this age of raging consumerism,
but are available on every street corner.

Why did the translator not make use of this option, but reversed the normal
order of the contrast, referring to the availability of fast cars before negating
their scarcity?

Fast cars and gold watches, cases of champagne and perfume, which are
available on every street corner, are not scarce, rare, expensive or
desirable in this age of raging consumerism.

The answer is that the first sentence with its contrast is embedded into
another contrast with the second sentence. It is not the availability of the
traditional luxury goods, but the negated scarcity and expensiveness which
is contrasted with the elementary necessities of life that come at great cost.
That is, the second clause contrasts with the predicate of the first clause and
not with the apposition. This means that the content of the first clause should
be easily accessible when we reach the second clause, which will not be the
case if we use the first contrast in its normal order.

The solution in the translation is the separation of the negated phrase of

the contrast into an independent clause after a semicolon, where it becomes
immediately accessible to the successive contrast. But as the semicolon has
already reached the highest level of internal separation, the subsequent

130

Shifting boundaries

contrast needs an even higher level of separation. It is thus used as an
independent sentence, marking the contrastive topic at the beginning of the
sentence by a cleft. This cuts out a bigger chunk of the preceding context as
the domain of its contrast:

If the last clause were not separated from the clause it contrasts with, it
would not show its hierarchical level despite the cleft:

Fast cars and gold watches, . . . are available on every street corner; they
are not scarce, rare, expensive or desirable in this age of raging
consumerism, but it is the elementary necessities of life that come at a
great price: quiet, good water, and enough space.

It is more difficult in a clause to find one’s way back to the fast cars, which
form the contrast with the elementary necessities of life, than in an
independent sentence.

The problem does not arise in German, where the contrastive topics

follow each other directly. The intervening apposition is itself a part of the
contrasted discourse topic:

Knapp, selten, teuer und begehrenswert sind . . . nicht schnelle
Automobile und goldene Armbanduhren, . . . Dinge, die an jeder
Straßenecke zu haben sind, sondern elementare Lebensvoraussetzungen
wie Ruhe, gutes Wasser und genügend Platz.

Shifting sentence borders

Sometimes both operations are combined, that is, clauses are separated into
an independent sentence and a clause which is linked to a subsequent
sentence. The following is a relatively transparent example, although it
involves a change from a declarative to an imperative sentence at the end.
Both the original and the translation consist of two sentences, but the
separation occurs in different places.

The translation reads as follows:

What would happen if we were to apply this simple mathematical model
to time or, more precisely and more modestly, to historical time? Just out
of curiosity and, if you like, for entertainment, as an alternative to the
linear model of time used in classical physics, let’s do the experiment.

(E 1997: 39)

Shifting boundaries 131

Except for the final part, it has stayed very close to the original:

Wie wäre es, wenn wir dieses mathematische Modell auf die Zeit,
genauer und bescheidener gesagt, auf die historische Zeit anwenden
würden – vorerst nur aus Neugier und meinetwegen zum Vergnügen, als
Alternative zum linearen Zeitmodell der klassischen Physik? Es käme
auf den Versuch an.

(E 1999: 15)

The postverbal adverbial after the dash has been adjoined in the translation
to the second sentence – an imperative – as an adverbial onset. As the
experiment in the imperative refers to the application of the mathematical
model to time, which was mentioned in the first sentence, the adverbial
modifies the same event whether we use it in the first sentence or in the
second. The translation can therefore be considered equivalent.

But why did the translator shift the sentence boundary at all? What does

one get if one translates the sentence analogously?

What would happen if we were to apply this simple mathematical model
to time or, more precisely and more modestly, to historical time – just
out of curiosity and, if you like, for entertainment, as an alternative to
the linear model of time used in classical physics? Let’s do the
experiment.

The end of the first sentence seems awfully long. It is rather difficult to sort
out all constituents of the adverbial and integrate them adequately into the
matrix structure.

Now, the original matrix sentence presents an idiomatic request in form of

a question: wie wäre es, wenn/how about . . . which is very similar to the
imperative: let’s do it. But reformulating the real question: what would
happen if . . .? as an indirect request would not alter the different conditions
for sentence boundaries too much:

How about applying this simple mathematical model to time or, more
precisely and more modestly, to historical time – just out of curiosity
and, if you like, for entertainment, as an alternative to the linear model
of time used in classical physics? Let’s do the experiment.

Although this version is better than the previous one as the affirmative
nature of the adverbial no longer clashes with the interrogative nature of the
matrix structure, one might easily miss the textual relevance of the last part
of the sentence.

Another short look at the original:

Wie wäre es, wenn wir dieses mathematische Modell auf die Zeit,

132

Shifting boundaries

genauer und bescheidener gesagt, auf die historische Zeit anwenden
würden – vorerst nur aus Neugier und meinetwegen zum Vergnügen, als
Alternative zum linearen Zeitmodell der klassischen Physik? Es käme
auf den Versuch an.

tells us that the alternative between the simple mathematical model and the
linear model of time is of textual relevance, and the final position in the
German original helps us to realize it directly. In fact, the interpretation of
the conditional is not different if we take the extraposed information back
into its basic position:

wenn wir dieses mathematische Modell – vorerst nur aus Neugier . . . –
auf die historische Zeit anwenden würden . . .

provided the parenthetical character of the information is signalled by
dashes.

In English, the focus of the conditional (or gerund phrase) is restricted to

the verb and its internal argument – that is, the matrix structure establishes a
contrast between apply to historical time/not apply to historical time. The
adverbial extensions after the dash do not participate in this alternative:

The focus interpretation is restricted to the focus exponent and does not
include the focus projection after the dash, although the structure after the
dash covers contextually new information, too.

In German, the parenthetical nature of the information does not prevent

the processor from identifying the extraposed, right-peripheral material as
an important part of the entire sentence, participating in the contextual focus
projection. However, if we use the German structure in English, there is
nothing to tell us that all this, after the dash at the end of the sentence, is of
textual relevance. Except that the mere quantity of the information, which
contradicts the weak position, will cause a processing confusion.

In separating the modifiers from the first sentence, one can avoid this

confusion. However, unlike previous cases, the adverbial phrase need not be
extended into a clause, because it can adjoin the next, rather short sentence.
The price for the boundary shift is the use of an imperative instead of the
declarative sentence since only the former can counterbalance the complex
modifying phrase and secure equal relevance to the modifier and the main
clause:

What would happen if we were to apply this simple mathematical model

Shifting boundaries 133

to time or, more precisely and more modestly, to historical time? Just out
of curiosity and, if you like, for entertainment, as an alternative to the
linear model of time used in classical physics, let’s do the experiment.

In an English sentence, as you know, the left-peripheral position is much
more prominent; thus, the shift secures an adequate degree of relevance to
the contrasted models of time.

Extrasentential reordering

Shifts of sentence boundaries can be directed backwards, too, and can come
as a consequence of changes which are necessary for other reasons – for
example those of filling a lexical gap.

Earlier on in the Key (p. 8), there was an extreme example of structural

extension confronting us with the question of whether such a deviation from
the original did not mean a change of style altogether:

Mittelwertbildung ist gleichbedeutend mit Informationsverlust.

was translated by:

Whenever we represent a system by an average we necessarily lose some
information about the total system.

that is, by clausal extensions of the subject and the predicate and a
conditional relation instead of the equivalence relation expressed by ist
gleichbedeutend/ means the same. Having gathered so much evidence about
language-specific processing conditions in English and German, we may
now be ready to cope with the example.

What we can see straight away is that there is a lexical gap which the

translation has tried to fill. Actually, Mittelwertbildung is a discourse topic
of the chapter the sentence belongs to, and thus the term occurs quite
frequently in the text. It is translated in various ways depending upon the
particular context it occurs in. We have already looked at one of these
translations (the process of averaging) some time ago.

Now, an analogous translation of the original would be something like:

Forming an average means losing some information.

How would this version fit into the context?

Entropy, which is complementary to temperature, is a measure of
information. It tells us how the total energy is distributed among the
various quantum states of the system. Forming an average means losing
some information.

134

Shifting boundaries

The idea of forming an average may seem to occur somewhat abruptly, but
as the sentence opens a new paragraph, we can accept a shift in topic.
Actually, the idea of an average was a recurring theme in the last but one
paragraph, explicitly mentioned in four of its nine sentences. Thus, the focus
is the predicate of the sentence, and the focus exponent Informationsverlust/
losing information. There are apparently no problems, then, with the
analogous translation. So why did the translator think it necessary to put so
much structural weight into his version?

Perhaps we find a clue in the subsequent context. How does the paragraph

initiated with the sentence continue?

Mittelwertbildung ist gleichbedeutend mit Informationsverlust. Wo
immer man mittelt, bedarf es einer Spezifikation, auf welche Detail-
zustände sich die Mittelung erstreckt, das heißt einer Zahl, die angibt
wieviel Information bei der Mittelwertbildung verloren gegangen ist.

Relatively close to the second sentence of the original would be the English
version:

Whenever we form an average, we need a specification of how many
individual cases are included in the average, that is, a number which tells
us how much information was lost in the process of averaging.

This version may be a bit too nominal for the parsing strategy of a language
with left-peripheral verb phrases, and one could recategorize some of the
nouns as verbs to avoid attachment problems:

Whenever we form an average we have to specify how many individual
cases are included in the average and determine how much information
was lost in the process of averaging.

Now, if one adds a textual link between the first and the second sentence, the
resulting paragraph is not too bad:

Forming an average means losing information. Thus, whenever we form
an average we have to specify how many individual cases are included in
the average and determine how much information was lost in the
process of averaging.

or, dropping one of the repeated averages:

Forming an average means losing information. Thus, whenever we form
an average we have to specify how many individual cases are included
and determine how much information was lost in the process of
averaging.

Shifting boundaries 135

When we look at the different beginnings of the first and second sentence, we
get the feeling that the translator resented the repetition of the phrase form
an average. The phrase got into the translation in the first sentence by
recategorizing Mittelwertbildung as a non-finite phrase. To avoid the
repetition the translator simply replaced the non-finite subject of the first
sentence by the conditional clause of the second sentence.

If we apply the same to our even more economical version, we get:

Whenever we form an average, we lose some information. Conse-
quently, we have to specify how many individual cases are included and
determine how much information was lost in the process of averaging.

So far, we can follow the translator. And one thing is already quite clear –
although we still lack an explanation for the remaining extensions, the
difference between the original and the translation is definitely much less
striking now than when we compared the sentences first. The translation no
longer suggests a deviation from the stylistic register of the original. What
appeared to be a structural extension, the conditional clause, is in reality a
case of extrasentential reordering, which not only filled the lexical gap of
Mittelwertbildung, but allowed the number of unnecessary repetitions to be
cut down.

What about the remaining differences between the more analogous

paraphrase and the existing translation:

Whenever we form an average, we lose some information.

Whenever we represent a system by an average, we necessarily lose some
information about the total system.

The translator relates the average and the loss of information to a system and
makes the loss a necessary consequence of the average. The first extension
signals the textual relevance of the topic, by underlining the shift from the
particular to the general, that is from the specific system of temperature,
which was the discourse topic of the last two paragraphs, to any system
whatsoever (the subsequent example speaks of jetliners and average weight
for passengers).

The second extension, the modal adverb, necessarily, has an information

structuring function. Necessarily indicates that there is a focus in the main
clause, which is something that might be missed otherwise, since whenever is
one of those focus attracting elements which could push everything else into
the background. As the modality of necessity is implied in the original by ‘ist
gleichbedeutend mit’, the messages of the original and translation are
equivalent.

Having looked at differences in linearization, syntactic relations,

structural explicitness and sentence boundaries which distinguish English

136

Shifting boundaries

translations from German original sentences in line with the language-
specific conditions for optimal processing, we have concentrated on a basic
but tiny set of translation ‘problems’ between two alternatively para-
metrized languages. The only excuse for dealing with such a small subset at
all is its general character. Questions of word order, grammatical relations
and structural explicitness are part of any linguistic encoding. Nevertheless,
there are many more differences between the systems and uses of languages
to deal with, even if one ignores the all-pervasive question of lexical
differences. What most translation scholars will have missed in the previous
chapters is a reference to the great variety of translational problems and
translational norms which are not controlled by the criterion of optimal
processing. Although I have used existing translations to demonstrate the
language-specific conditions for optimal processing in discourse, I did not
access the vast reservoir of existing translation which does not, for whatever
reasons, follow the translational maxim of the Key. Although the Principle
of Optimal Relevance says that language is perceived normally as if it were
optimally relevant for the perceiver, we know that there are very many ways
of violating this principle – including the alternative presumption that
something which is directly addressed to me is totally irrelevant for me. But
even if we ignore all sorts of interactive idiosyncrasies, there are quite
enough cases of conventionalized constraints on the Principle of Optimal
Relevance. Translational norms themselves could be among them. Does this
mean that I have to surrender the criterion of optimal processing or the
concept of optimal translation based on it?

Let us take a closer look at some examples illustrating such cases.

Relativizing optimality

137

When we set out walking the tightrope, we agreed to close our eyes to any
special uses of language we might encounter in translating. Even so, the
examples we have been looking at were colourful enough. They were all
subject to the default version of the Principle of Optimal Relevance,
however. That is, linguistic means were used so as to optimize processing
conditions in the original as well as in the translation. The use of language,
though, can also be subject to other principles, which may in one way or
another relativize optimality.

It may be the characteristics of individual style or of whole genres that

cause the conditions of optimal processing to be flouted, but even in such
cases, enough remains coherent to safeguard comprehension – unless a
message is emotional or dadaistic or whatnot almost to the exclusion of
everything informative. In general, stylistic peculiarities may be character-
ized by the nature and extent of deviations from optimal processing
conditions.

Two alternative cases of such special uses of language may suffice to

illustrate the consequences for translation: first, a typical example of legal
language and, second, some classical examples of literary translation. Both
cases demonstrate the limits of translatability, when differences between
target and source language cannot be compensated for by any redistribution
of information.

A preamble

Special purposes generate special discourse conditions. They will in many
cases appear as arbitrary as last year’s fashion, yet they may be strictly
conventionalized, and the conventions of target and source language may or
may not coincide. Even when they do coincide, the specific properties of
source and target language will promote variations within otherwise similar
patterns.

The conventional discourse patterns of legal texts in English and German

are not so far apart in their ‘macrostructural’ architecture. Compared with

138

Relativizing optimality

the default type of language use, legal texts are in both languages strongly
marked as text types for which the principle of optimal processing is lifted
more than once.

Thus, if we have a sentence starting with a participle adverbial which has

four infinitive complements between one and four lines long and afterwards
only a three-word subject and a two-word predicate, normal processing
conditions for English and German sentences are violated alike. But as an
opening statement to a treaty, we will consider even a lop-sided structure
such as the following as stylistically well formed, and think of all the articles
after it as holding it in balance:

Based on the intention

to reduce the state’s commercial activity as quickly and extensively as
possible through privatization,

to make as many businesses as possible competitive, thus protecting
existing jobs and creating new ones,

to provide real estate for economic purposes,

to make it possible, after taking stock of state property, its productivity,
and after its prior use for structural adaptation of the economy and
stabilization of the national budget, to grant investors a vested right to
shares in state property for a sum reduced through the monetary
conversion of July 2, 1990,

the following law is passed:

Article 1: . . .

(Jarausch and Gransow 1994: 165)

The fact that we are accustomed to the sight of such stylistic figures does not
take away anything of the processing burden. The preamble is after all
intended as one, however complex, information unit. Without the graphic
aid cutting the whole into relatively autonomous subsegments, we would
hardly be able to process it. The internal structure of the long final segment is
particularly difficult as it does not profit from the graphic layout. But this
difference is a matter of the microstructural distribution of the information,
which we will look at in more detail later on.

As long as we stay with the macrostructural aspect, the German and

English conventions are surprisingly similar. Except for the parametrization
of the left- and right-peripheral verb positions and the left- and right-
directional structural extensions of the verbs, the German version is built up
very much in the same way as the English version:

Relativizing optimality

139

Getragen von der Absicht,

die unternehmerische Tätigkeit des Staates durch Privatisierung so rasch
und so weit wie möglich zurückzuführen,

die Wettbewerbsfähigkeit möglichst vieler Unternehmen herzustellen
und somit Arbeitsplätze zu sichern und neue zu schaffen,

Grund und Boden für wirtschaftliche Zwecke bereitzustellen,

. . .

wird folgendes Gesetz erlassen:

§ 1

(Gransow and Jarausch 1991: 185)

Although topicalization is normal in German, it is not only the solemn
nature of the participle phrase getragen von der Absicht, but also the
complexity of the topicalized series which makes it an extraordinary
beginning in German, too. Processing ease is clearly not the dominating
criterion when conceiving a law.

No liberalism

If we look at the microstructural architecture, we can see that the
subsegments carved out by the layout are structurally supported by the
syntactic parallelism between the segments. However, the English version is
here more consistent than the German version, where the fourth segment is
presented as a clausal complement:

daß nach einer Bestandsaufnahme des volkseigenen Vermögens und
seiner Ertragsfähigkeit sowie nach seiner vorrangigen Nutzung für
Strukturanpassung der Wirtschaft und die Sanierung des Staats-
haushaltes den Sparern zu einem späteren Zeitpunkt für den bei der
Währungsumstellung am 2. Juli 1990 reduzierten Betrag ein verbrieftes
Anteilsrecht an volkseigenem Vermögen eingeräumt werden kann, . . .

Despite all its complexities, this segment is easier to process in English than
in German, which may come as a surprise seeing that the German version –
as you may have noticed from the content – is the original.

When we think of what may have prompted the additional complexity in

German, we can make out a language-specific difficulty in the verbal part of
the infinitival alternative. It is the right-peripherality of the German verb
structure which results in an additional constraint on the use of the modal
verb. By itself, the complement:

140

Relativizing optimality

den Sparern ein verbrieftes Anteilsrecht am volkseigenem Vermögen
einräumen zu können

is a grammatically perfect version, but not in the intentional context:

*getragen von der Absicht, den Sparern ein verbrieftes Anteilsrecht an
volkseigenem Vermögen einräumen zu können.

Grammatically seen, the situation would be somewhat better with möglich
machen:

Getragen von der Absicht, es möglich zu machen, den Sparern ein
verbrieftes Anteilsrecht an volkseigenem Vermögen einzuräumen

but stylistically, things have become even worse – especially if we think of the
many additional conditions yet to be integrated into the segment.

One could, of course, avoid the entire problem by dropping the modality:

getragen von der Absicht, den Sparern ein verbrieftes Anteilsrecht an
volkseigenem Vermögen einzuräumen

but that would promise the investors more than the legislator has intended.
The difference may be negligible in the everyday use of language, but not for
the degree of commitment adhered to in legal texts. For legal texts are not
only peculiar in that they follow certain formal conventions requiring an
extra amount of processing effort, but also they are less tolerant in terms of
specificity as the legislator cannot afford potential ambiguities or merely rely
on the readers’ inferential capabilities. Still, you could suspect the legislator
of overdoing it. After all, the intention of doing something does not entail its
success – the intention of granting a right does not entail the fact of granting
it, but it should entail trying to create the possibility of granting it. So, why
mention the possibility at all?

Obviously, jurisdiction is not going to let us off with this sort of everyday

reasoning. However difficult the formulation may be, if the legislators decide
upon it, they will claim to have had reasons for doing so. No liberal deletion
of information in the interest of easy processing!

Quite normal

In general, however, the Principle of Optimal Relevance is valid even in a
legal text. It goes without saying that the grammatical parameters are not
floated in the translation, even if they are stretched to their extremes as in the
original. Let us begin with the normal cases, such as the microstructural view
of the simpler segments.

In the original, all three start with the object, but each of the objects is

Relativizing optimality

141

located in a different configuration. The differences between German and
English secure adequate processing conditions for a normal focus projection.

The simplest case is the third complement:

Grund und Boden für wirtschaftliche Zwecke bereitzustellen

where the prepositional object has been placed in the position of the focus
exponent, near the verb. The idea of real estate/Grund und Boden is implied
by the concept of state property referred to in the title of the law. The order
of the two objects is exactly what we expect it to be in a discourse-sensitive
language.

In the English translation, the verb-adjacent focus position is filled by the

direct object:

to provide real estate for economic purposes.

The translation unscrambles the German original in line with the English
grammar and equivalence is achieved via a discursive reanalysis identifying
the prepositional object as focus, as in any normal text.

The first complement is not much more complicated than the third:

die unternehmerische Tätigkeit des Staates durch Privatisierung so rasch
und so weit wie möglich zurückzuführen,

to reduce the state’s commercial activity as quickly and extensively as
possible through privatization, . . .

It is merely a question of locating the adverbials according to their scope,
which is nothing but the verb in the case of the modifying adverbial, but the
verb with its modifier in the case of the causal adverbial – causal before
modal in a left-branching language, and modal before causal in a right-
branching language.

Things are less simple with the second complement:

die Wettbewerbsfähigkeit möglichst vieler Unternehmen herzustellen . . .

where the first predicate is in the original more nominal than in the trans-
lation. The head of the German object is a nominalized adjective, which is
translated as a predicate adjective in English: Wettbewerbsfähigkeit/(make)
competitive:

to make as many businesses as possible competitive . . .

The recategorization has its effect on the grammatical realization of the
arguments associated with the syntactic heads. There is the attributive

142

Relativizing optimality

genitive of the German noun, (Wettbewerbsfähigkeit) vieler Unternehmen,
while in English, the underlying subject of competitive is at the same time the
overt object of the predicate, (make) . . . many businesses . . . (competitive).

Both versions secure an adequate focus interpretation. Unternehmen/

businesses can be contextually inferred from commercial; thus, there is a
focus on competitive, in addition to the lexically inherent focus of the
superlative möglichst viele/as many . . . as possible.

Formally seen, these are, again, cases where both languages have both

options. That is, we could have an analogous form of the English version:

möglichst viele Unternehmen wettbewerbsfähig zu machen

as well as an analogous translation of the German original in English:

to achieve/warrant competitiveness of as many businesses as possible.

Yet in German, the nominalized version of the original:

die Wettbewerbsfähigkeit möglichst vieler Unternehmen herzustellen . . .

could present another case of focus-spacing, taking the intended quality and
quantity of businesses apart. English achieves the same effect with its small-
clause version of the predicative adjective:

to make as many businesses as possible competitive . . .

which puts a syntactic border between the predicate and the matrix struc-
ture. The verbal frame blocks this possibility in the analogous German
structure, calling for a different means of focus spacing, while in English, the
nominal form seems to carry one focus only – perhaps due to the weight of
the as-many-as-possible structure. Whatever the precise differences between
the focus structures of English and German noun phrases may be, they
belong to the general properties of both languages and are not text-type-
specific.

As a rule, nominalization in itself is no processing obstacle. It can help

improve processing through focus spacing, as was the case in the example
about competitive businesses. Nominalization may even be necessary in
a left-branching language, to secure contextual appropriateness of the
structural focus itself. Thus, the English version:

The Council of Ministers appoints the Trusteeship Agency to implement
the necessary measures.

ends on an infinitive complement that goes back in German to a prepo-
sitional phrase with a deverbal noun:

Relativizing optimality

143

Der Ministerrat beauftragt mit der Durchführung der entsprechenden
Maßnahmen die Treuhandanstalt.

The direct object is the focus exponent, and it is the contextual focus because
the law refers to the Trusteeship Agency here for the first time after the title,
which is separated from this sentence by the whole of the preamble and the
first two sentences of article 1. The Council of Ministers and the necessary
measures are given or entailed by the immediately preceding sentence.
Thus, the prepositional object mit der Durchführung der entsprechenden
Maßnahmen is of lower relevance than the direct object, and the order of the
two objects is contextually adequate.

If we were to use an infinitive instead of the deverbal noun – that is, a

structure analogous to the English version – we would get the wrong focus
structure:

Der Ministerrat beauftragt die Treuhandanstalt, die entsprechenden
Maßnahmen durchzuführen.

English, on the other hand, needs just this order to get the Trusteeship
Agency into the structural focus position after the verb:

The Council of Ministers appoints the Trusteeship Agency to implement
the necessary measures.

But in English, a nominal version would not allow any other order, either:

The Council of Ministers . . . the Trusteeship Agency with the
implementation of the necessary measures.

Yet as the nominal version is a derived structure, it is more difficult to
process than the verbal structure, and as there are no other gains associated
with the difficulty it is felt to be unnecessary and clumsy.

In German, the nominalized version improves processing in terms of focus

structure, but this is as far as the legal text goes in optimizing processing
conditions.

Statuesque and dense

In an everyday text of German, one might find a more balanced distribution
of information, using a concave pattern with the partitive topic:

Mit der Durchführung der entsprechenden Maßnahme beauftragt der
Ministerrat die Treuhandanstalt.

In the legal example, however, except for the preamble, all sentences begin

144

Relativizing optimality

with the subject. That is, the stylistic principle of a balanced focus structure
with an even distribution of the more relevant elements around those of
lowest relevance is flouted in a legal text. The statuesque impression of the
genre is clearly due to the rigid handling of the topic position in a language
which is normally highly discourse sensitive.

The effect is reinforced by the repetitive nature of the initial subjects,

which the Principle of Optimal Relevance would normally require to be
pronominalized:

Das volkseigene Vermögen ist zu privatisieren . . .

Volkseigenes Vermögen . . .

Der Ministerrat trägt . . .

Der Ministerrat beauftragt . . . die Treuhandanstalt.

Die Treuhandanstalt wird . . .

The unnecessary explicitness adds an extra weight to each subject.

The rigid and repetitive nature is all the more remarkable as the high

degree of explicitness can fill the verbal frame with much information, which
would be easier to process in a more balanced form. Thus, the English
version:

In certain cases to be determined by law, state owned property can also
be conveyed to the ownership of local governments, cities, districts, and
states, as well as to the public.

is in the German original:

Volkseigentum kann auch in durch Gesetz bestimmten Fällen
Gemeinden, Städten, Kreisen und Ländern sowie der öffentlichen Hand
als Eigentum übertragen werden.

With the adverbial squeezed into the middle field and its specifier in
prenominal position, in durch Gesetz bestimmten Fällen, the German
version makes for tough reading, which may demonstrate to the legal subject
that the assertion of the law is no matter of discourse.

The sentence would be easier to process if the adverbial were topicalized

and the prenominal attribute spelled out postnominally:

In bestimmten Fällen, die durch Gesetz festzulegen sind, kann volks-
eigenes Vermögen Gemeinden, Städten, Kreisen und Ländern sowie der
öffentlichen Hand als Eigentum übertragen werden.

Relativizing optimality

145

This is very much the form of the English version, which does not try to
reproduce the authoritarian, statuesque style of the original. A translation
that merely informs us about a law regulating affairs in another country can
afford a greater degree of processing ease. The result is still bad enough,
though, where the original is as crammed full with information as in our
fourth complement of the preamble.

Concentrating on a particularly ill-shaped original is really no fun, yet we

cannot close our eyes to what translators have to cope with. Let us approach
the fourth complement in its ‘liberalized’ English form. Here is the trans-
lation once more:

to make it possible, after taking stock of state property, its productivity,
and its priority usefulness for structural adaptation of the economy and
stabilization of the national budget, to grant investors a vested right to
shares in state property for a sum reduced through the monetary
conversion of July 2, 1990, . . .

Indeed, one could easily choke on the amount of information squeezed into
what is, after all, just an infinitive complement. But it is an infinitive
complement with an infinitive complement inside, with a seemingly endless
number of modifiers and arguments of arguments of arguments . . .

The density of the structure can become even worse, as the German

original demonstrates:

daß nach einer Bestandsaufnahme des volkseigenen Vermögens und
seiner Ertragsfähigkeit sowie nach seiner vorrangigen Nutzung für
Strukturanpassung der Wirtschaft und die Sanierung des Staatshaus-
haltes den Sparern zu einem späteren Zeitpunkt für den bei der
Währungsumstellung am 2. Juli 1990 reduzierten Betrag ein verbrieftes
Anteilsrecht an volkseigenem Vermögen eingeräumt werden kann, . . .

While the English version has at least two finite and two non-finite verbs
distributed over the whole passage, which is in this way subdivided into
several clause-like segments:

to make it possible after taking stock . . .

to grant investors . . .

a sum reduced . . .

the German original is nominal almost throughout. Unlike the English post-
nominal participle, even the prenominal present participle shortly before the
end is of no help. The German participle closes a complex prenominal phrase
consisting of two prepositional phrases which are embedded into yet another
prepositional phrase:

146

Relativizing optimality

für den bei der Währungsumstellung am 2. Juli 1990 reduzierten Betrag
. . .

The entire prepositional object is, in turn, squeezed between a temporal
adverbial and a complex direct object before the predicate:

zu einem späteren Zeitpunkt für den . . . Betrag ein verbrieftes Anrecht
an . . . eingeräumt werden kann . . .

and the temporal adverbial is preceded by as many as twelve noun phrases,
which together constitute a short direct object and another, ‘endless’
temporal adverbial.

Overspecifying

The second temporal adverbial seems redundant since the first temporal
adverbial specifies the preconditions for the right to shares in state property
extensively, namely in two coordinated prepositional phrases, where each
nominal head is again extended by a coordinated complement:

nach einer Bestandsaufnahme des volkseigenen Vermögens und seiner
Ertragsfähigkeit sowie nach seiner vorrangigen Nutzung für Struktur-
anpassung der Wirtschaft und die Sanierung des Staatshaushaltes . . .

As the implementation of all these conditions will take time, the next step
can only take place at a later point of time. That is, the meaning of the second
adverbial is implied by the meaning of the first adverbial. In the interest of
optimal processing, one could simply leave the second temporal adverbial
implicit – which is what the translator has indeed opted for.

This would, in German, also allow us to prepose the direct object as it is no

longer needed to cover up the redundancy of the second temporal adverbial
by putting some distance between the first and the second adverbial:

daß den Sparern nach einer Bestandsaufnahme des volkseigenen
Vermögens und seiner Ertragsfähigkeit sowie nach seiner vorrangigen
Nutzung für Strukturanpassung der Wirtschaft und die Sanierung des
Staatshaushaltes . . .

But the density of the structure is still so high that the processing advantage is
hardly noticeable. In any case, we can be sure that the legislators did have
second thoughts about the temporal specifications. Just as they insisted on
adding the modal verb to the predicate, they had their judicial reasons for
making the temporal implication explicit. We, the legal subjects, should not
get the wrong idea and claim our right to shares in state property too early. It
is clear that the legislators are, again, overspecifying their case to be on the
safe side, fending off unjustified claims right from the beginning.

Relativizing optimality

147

Now, the translator has decided against overspecification, which is

understandable in view of the informationally overloaded complement, but
the reduction is not licensed by our Translation Maxim. As the discussion
of this example shows, the convention of a special register may demand
relativizing the Principle of Optimal Relevance, that is, the optimal
proportion between processing ease and cognitive gains may deviate from
the default case of discourse.

In our case, the judicially motivated degree of specification, combined

with a low degree of discourse sensitivity in terms of topicalization and
pronominalization, restricts processing ease in a genre which is characterized
by a high degree of informational density, that is, by passages where one
discourse segment may comprise an extreme amount of information.
The constraints are a mixture of rational and arbitrary, or irrational,
conditions, with the latter dominating. While the trend to overspecification
may be understandable from the legislator’s point of view, the constraints
on the discourse-sensitive use of language, on topicalization and pro-
nominalization, and on a lower degree of information density seem to be
arbitrary.

However, there is one common denominator to all three constraints – they

all guarantee a very low degree of context dependency. That is, a discourse
segment in a legal text has a high degree of contextual autonomy. Its
interpretation does not depend upon the preceding context, in contrast to the
easier-to-process alternatives.

It is clear that anaphors are by definition context-dependent, and an even

distribution of information according to its discourse relevance is highly
context-dependent, too. For a lower degree of informational density we
would have to use several sentences that would depend upon each other for
their interpretations.

After all, if we think of the function of laws as regulating social matter, we

can recognize some rationale behind the trend towards context indepen-
dency. But the desired self-sufficient, unambiguous and firm declaration is,
as the example demonstrates, easily in danger of becoming an overburdened,
indigestible piece of discourse.

Normalizing the extraordinary

Optimality can be relativized for various reasons, whether general or
individual and idiosyncratic. Individual reasons are accorded high regard
when they can be attributed to the characteristics of a writer’s literary style.

To the extent that literature wants to inform, its special use of language is

also governed by the Principle of Optimal Relevance, but as we all know, it is
the artistic use of language which accounts for much of the literary
fascination, and processing ease may well be sacrificed for it. Yet whenever a
literary style stretches the potential of its language, violating the default
conditions of optimal use, the translator faces a real problem. Would

148

Relativizing optimality

stretching the target language in an analogous way merely be ‘blamed’ on the
translator, who will be assumed to have failed optimality under target
language conditions?

The Austrian writer Thomas Bernhard is not only famous for his deep-

black irony, but also for his endless variations and repetitions of thoughts,
which only gradually make way for new thoughts. His sentences can be said
to have a very low degree of informational density. We have to process a
great number of nearly equivalent sentences, in which Bernhard approaches
his object slowly and stubbornly, very much like a painter who draws ever
new lines around his figures. Even the title of a whole book may be subjected
to this procedure. The following examples are taken from the novel,
Auslöschung. Ein Zerfall, which is something like Extinction. A Case of
Dissolving. The double title emphasizes the destructive nature of the book
by a near tautology. (The English translator makes use only of the first
concept, already demonstrating outside the text his readiness to ‘normalize’
the author.) This is what we find at the beginning of the English version:

On the twenty-ninth, having returned from Wolfsegg, I met my pupil
Gambetti on the Pincio to discuss arrangements for the lessons he was to
receive in May, writes Franz-Josef Murau, and impressed once again by
his high intelligence, I was so refreshed and exhilarated, so glad to be
living in Rome and not in Austria, that instead of walking home along
the Via Condotti, as I usually do, I crossed the Flaminia and Piazza del
Popolo and walked the whole length of the Corso before returning to my
apartment in the Piazza Minerva, where at about two o’clock I received
the telegram informing me that my parents and my brother, Johannes,
had died.

(B 1995: 3)

Except for the position of the parenthetical clause, writes Franz-Josef
Murau, the English version does not differ from any other beginning of a
story, working its way step by step, clause by clause towards the climax at
the end of the sentence.

This is by no means the case with the German original. While the

translation uses parallel structures:

. . ., having returned . . . and impressed once again . . ., I was so refreshed
and exhilarated, so glad . . .

the original coordinates two conjuncts within a relative clause, and intersects
the first and the second part of the passage with unusually located paren-
thetical clauses:

Nach der Unterredung mit meinem Schüler Gambetti, mit welchem ich
mich am Neunundzwanzigsten auf dem Pincio getroffen habe, schreibt

Relativizing optimality

149

Murau, Franz-Josef, um die Mai-Termine für den Unterricht zu
vereinbaren, und von dessen hoher Intelligenz ich auch jetzt nach meiner
Rückkehr aus Wolfsegg überrascht, ja in einer derart erfrischenden
Weise begeistert gewesen bin, daß ich ganz gegen meine Gewohnheit,
gleich durch die Via Condotti auf die Piazza Minerva zu gehen, auch in
dem Gedanken, tatsächlich schon lange in Rom und nicht in Österreich
zuhause zu sein, in eine zunehmend heitere Stimmung versetzt, über die
Flaminia und die Piazza del Popolo, den ganzen Corso entlang in meine
Wohnung gegangen bin, erhielt ich gegen zwei Uhr mittags das
Telegramm, in welchem mir der Tod meiner Eltern und meines Bruders
Johannes mitgeteilt wurde.

(B 1989: 7)

The sentence is extremely difficult to parse for there is no formal indicator

that would help the parser recognize the structural attachment of the two
conjuncts in the relative clause. Let me present it once more in a segmented
format to help you get through:

Nach der Unterredung mit meinem Schüler Gambetti,

mit welchem ich mich am Neunundzwanzigsten auf dem Pincio

getroffen habe, schreibt Murau, Franz-Josef,

um die Mai-Termine für den Unterricht zu vereinbaren,

und von dessen hoher Intelligenz ich auch jetzt nach meiner Rückkehr
aus Wolfsegg überrascht, ja in einer derart erfrischenden Weise
begeistert gewesen bin, daß ich

ganz gegen meine Gewohnheit, gleich durch die Via Condotti auf die
Piazza Minerva zu gehen,

auch in dem Gedanken, tatsächlich schon lange in Rom und nicht in
Österreich zuhause zu sein, in eine zunehmend heitere Stimmung
versetzt,

über die Flaminia und die Piazza del Popolo, den ganzen Corso
entlang in meine Wohnung gegangen bin,

erhielt ich gegen zwei Uhr mittags das Telegramm, in welchem mir der
Tod meiner Eltern und meines Bruders Johannes mitgeteilt wurde.

The sentence would be easier to process if the two parentheses were

in a grammatically normal position (and the name not in its bureaucratic
form):

150

Relativizing optimality

Nach der Unterredung mit meinem Schüler Gambetti, schreibt Franz-
Josef Murau, mit welchem ich mich am Neunundzwanzigsten auf dem
Pincio getroffen habe . . .

daß ich, auch in dem Gedanken, tatsächlich schon lange in Rom und
nicht in Österreich zuhause zu sein, in eine zunehmend heitere Stimmung
versetzt, ganz gegen meine Gewohnheit, gleich durch die Via Condotti
auf die Piazza Minerva zu gehen . . .

But Bernhard wants his readers to get out of breath, lose their bearings, when
walking with Murau through Rome in an elevated mood before the bad
news hits.

It is the linguistic obstacles to comprehension, the exhilarated but bumpy

road into the story, which is missing in the translation. But what, indeed,
could the translator do, in a language that depends upon configurationality
and cannot stretch its positional variability as much as the non-
configurational German?

Artistic redundancies

Normalizing the author in questions of order, a sensitive property of the
English language, is necessary. Normalizing violations of optimal relevance
which would not result in grammatically unacceptable translations is not to
be excused – at least not under the Translation Maxim we have chosen. The
elimination of a whole clause, for example, is not justified even if the clause is
the umpteenth repetition of preceding thoughts.

The following translation not only reorders the original in the interest of

focus structure, but it also drops the last subclause of the original:

The thinking person who is idle appears as the greatest threat to those
for whom idleness means simply doing nothing, who actually do nothing
when they are idle.

(B 1995: 24)

Der Nichtstuer als der Geistesmensch ist tatsächlich in den Augen
derer, die unter nichts tun, tatsächlich nichts tun verstehen und die als
Nichtstuer auch tatsächlich gar nichts tun, weil in ihnen während des
Nichtstuens gar nichts vorgeht, die größte Gefahr und also der Gefähr-
lichste.

(B 1989: 39)

Now, the German original is heavily marked by its repetitiveness, and by

the fact that the heavy and complex relative clause has not been extraposed.
It would also be more normal in German, too, to have the predicate before
the local adverbial:

Relativizing optimality

151

Der Nichtstuer als der Geistesmensch ist tatsächlich die größte Gefahr
und also der Gefährlichste in den Augen derer, die unter nichts tun,
tatsächlich nichts tun verstehen und die als Nichtstuer auch tatsächlich
gar nichts tun, weil in ihnen während des Nichtstuens gar nichts vorgeht.

In English, however, we have to adapt the linear order to the grammatical
constraints of a configurational language. However, not being Thomas
Bernhard, we could be tempted into ‘normalizing’ the sentence further,
cutting down on the merely repetitive parts of the structure. We would then,
for example, replace at least one of the three tatsächlich by some other
adverb, and we would definitely drop the highly synonymous conjunct of the
predicate, also der Gefährlichste, and if we can help it, pronominalize at
least one of the references to Nichtstun/being idle:

Der Nichtstuer als Geistesmensch ist tatsächlich die größte Gefahr in
den Augen derer, die unter nichts tun wirklich nichts tun verstehen, und
die, wenn sie nichts tun tatsächlich nichts tun, . . .

This is what the translator has done in English (besides reordering the
structure and bridging lexical gaps). The translator has also dropped the last
clause, which is in fact a third repetition of the very same idea within half a
page or so. Talking about his parents, Murau says that:

they hated idleness . . . because . . . they did not know what to do with it,
for when they were idle there was actually nothing going on . . . My
parents’ idleness was of course genuine idleness for when they did
nothing there was nothing going on in them . . .

weil sie mit ihrem tatsächlichen Nichtstun gar nichts anfangen konnten,
weil in ihrem Nichtstun tatsächlich gar nichts vorging . . . ihr Nichtstun
war ein tatsächliches Nichtstun, denn es tat sich in ihnen nichts, wenn sie
nichts taten . . .

‘Repetitiveness’ is, as both passages of the original show, a major charac-
teristic of Bernhard, and if we want to render Thomas Bernhard in English,
we ought to retain his characteristic relativizations of optimality:

The thinking person who is idle appears indeed as the greatest danger
and the most dangerous person in the eyes of those for whom idleness
means really doing nothing, who when they are idle do nothing at all
because there is nothing going on in them when they are idle.

No choice

What if a writer’s style is mainly marked by violations of optimal relevance
which are simply not possible in the target language? Writing about Sunday,

152

Relativizing optimality

13 August 1961 in East Berlin, Uwe Johnson says about his heroine in Two
Views:

Before going to work she took the nearly one-hour walk to her room.
The street did not have a Sunday air; no radio music came from the
windows, balconies protruded emptily. More strollers than usual
seemed to be waiting for an event that rumor anticipated, like fireworks,
a thunderstorm.

(Johnson 1967: 38)

But this is, of course, what Johnson’s translator says. In the original, we read:

Vor dem Dienst noch fast eine Stunde lang zu Fuß lief sie zu ihrem
Zimmer. Die Straßen waren nicht sonntäglich, aus den Fenstern kam
nicht Radiomusik, die Balkons hingen leer. Mehr als sonst Spazier-
gänger schienen zu warten auf ein Ereignis, das gerüchteweise ausstand,
wie auf ein Feuerwerk, ein Gewitter.

(Johnson 1965: 50)

The German original describes the same situation as the English translation,
but while the English version reads quite normally, the German original is
marked heavily, mainly by extreme word order deviations. The first sentence:

Vor dem Dienst noch fast eine Stunde lang zu Fuß lief sie zu ihrem
Zimmer.

has three adverbials in topic position, before the finite verb, specifying the
modality, duration and point of time of the event. This is a violation of the
grammatical rule that normally restricts the topic position to one part of
speech.

The sentence sounds highly manneristic, and the effect is to some extent

kept up in what follows. Instead of the nominal negation, keine Radiomusik,
Johnson uses sentence negation, aus den Fenstern kam nicht Radiomusik,
attributing thereby a strong, contrastive focus to the negation; uses hängen
without its obligatory local adverbial; uses the obligatory postnominal
complement to mehr prenominally – Mehr als sonst Spaziergänger instead of
Mehr Spaziergänger als sonst; and places a predicate after the comparative
and thus in its contrastive scope, which should not be interpreted contrast-
ively. The prenominal position of the complement can be seen as a means of
structurally separating the subject from its finite verb, which would other-
wise demand something like Mehr Spaziergänger als sonst waren auf den
Straßen, so as to separate the subject from the following predicate and avoid
the erroneous implication that at other times fewer strollers were expecting
such an event. Normalized, the sentence would have to be extended into
something like:

Relativizing optimality

153

Mehr Spaziergänger als sonst waren auf den Straßen, sie schienen auf ein
Ereignis zu warten, das . . .

Nor is the predicate ‘harmless’. The prepositional object is extraposed
together with its relative clause:

schienen zu warten auf ein Ereignis, das gerüchteweise ausstand, . . .

which is modified by an ad hoc adverb, gerüchteweise, and followed by a
comparative phrase with two asyndetically coordinated noun phrases,
whose attachment – to the prepositional phrase before the relative clause –
requires yet more processing effort:

schienen zu warten auf ein Ereignis, das gerüchteweise ausstand, wie auf
ein Feuerwerk, ein Gewitter.

Except for the somewhat unusual statement about the balconies,

balconies protruded emptily

and the asyndetic sequence at the end,

for an event that rumor anticipated, like fireworks, a thunderstorm

which would be more normal with an or in between:

for an event that rumor anticipated, like fireworks or a thunderstorm,

none of the peculiarities of the German original can be found in the English
translation. (One of the consequences is that it is more difficult to avoid the
unintended implication of the last sentence.)

The translator really does not have any choice if the author’s style offers

no peculiarities which can be recreated in the target language. We can only
stretch the potential of a language as far as it goes. Relativizing optimality is
parametrized, too. And, surely, no violation – say, redundancy instead of
word order – is an adequate substitute for any other violation.

Leaving the tightrope for the trapeze

In one respect the tightrope is similar to firm ground: it does not swing. But
parametrized processing conditions are only one set of the linguistically
controlled differences between target and source language. There are also all
those innumerable idiosyncratic conditions of the lexicon, and then, of
course, the many cultural differences that leave less ground under our feet
than a swinging trapeze. There are no parameters to hold on to, just indi-
vidual problems of translations to be solved individually. A few examples

154

Relativizing optimality

may suffice to remind you of what you already know only too well, namely
that any attempt at predictive generalizations along the lines of the Key is
futile, but each individual case can boost our expertise as translators.

Explaining the applicability of entropy as a measuring tool, Eigen and

Winkler suggest considering ‘a printed passage consisting of one hundred
symbols’:

By symbols, we mean the twenty-six letters of the alphabet plus three
punctuation marks and the space between words. We are using, then, a
total of thirty different symbols.

(EW 1983b: 145)

But in reality they said:

Unter Symbolen verstehen wir die neunundzwanzig Buchstaben unseres
Alphabets (unter Einschluß der Umlaute Ä, Ö, Ü) sowie ein Abstands-
symbol, einen Zwischenraum, der zur Abgrenzung von Wörtern dient.
Insgesamt benutzen wir also dreißig Klassen von Symbolen.

(EW 1983a: 168)

It is clear that the translator has adapted the statement about the German
alphabet to the English alphabet, which needs some topping up, being three
symbols shorter than the German alphabet. The compensation is necessary
as the thirty symbols provide a relevant referent for the ensuing discourse.

What the English version does not need is an extra explanation for the

space between words, which is quite an affair in the German original:

ein Abstandssymbol, einen Zwischenraum, der zur Abgrenzung von
Wörtern dient.

In German, however, the space between words is not normally conceived

of as a symbol, and the strict logic of physicists cannot ignore the contextual
inadequacy, however small it may be for the layman. Thus, they compose an
Abstandssymbol and explain it afterwards. In English, though, where the
translation has extended the symbols of the alphabet by punctuation marks,
the conceptual class has already been shifted to the general domain of
graphic elements. Thus, the list can include the space between words without
any further comment.

Illustrating the future of luxury regarding time, Enzensberger talks about

some professions subjected to ‘regulations that limit their temporal sover-
eignty to a minimum’:

Workers are tied to the pace of their machines, housewives (in Europe)
to absurd shopping hours, . . .

(E 1997: 336)

Relativizing optimality

155

But, of course, the German original did not have to restrict the geographical
domain of the housewives:

Arbeiter hängen von Maschinenlaufzeiten, Hausfrauen von absurden
Ladenschlußzeiten . . . ab, . . .

(E 1999: 157)

as it is in Europe, especially Germany, where shopping hours are absurd.

Comments like the above are necessary to secure adequacy and equiva-

lence; without them the translation would carry a different message for the
English readers. Comments like these are not to be confused with para-
phrases of an original which fill no linguistic or cultural gap between source
and target texts, as was the case with an example way back in the Key (p.
101), where the translator offered his interpretation of an, admittedly,
nebulous original instead of translating it.

Pre-programmed losses

If we can fill a linguistic or cultural gap as left-handedly as in the examples
above, we are lucky. Often enough, we have to sacrifice some feature of the
original because there is nothing to compensate it with. We have frequently
touched upon the limits of translatability, but before we mounted the
trapeze, we had always been able to rescue the really relevant features.
Extraordinary use of word order was the first feature we could not rescue in
the translation, but there are, as everybody knows, many more aspects of an
original which are relevant but not fully translatable – such as metaphors,
puns, regionalisms, etc.

There is, for example, no hope for the Pomeranian dialect bringing Grass’

personae alive; when Kriwe, the ferryman in Dog Years, shows the village
mayor Amsel’s exhibition of scarecrows, he says:

There you are, friend. See what I mean.

(Grass 1989: 37)

But what Grass really made him say was:

Liebärchen. Daas send se nu, wo ech häd gemaint.

(Grass 1963: 47)

which is in German/English something like:

Mein Lieber, das sind sie nun, (die Vogelscheuchen,) die ich gemeint
hatte./ Here they are, the scarecrows I had told you about.

But the original is, of course, much more than this. It represents a whole

156

Relativizing optimality

world, bygone yet still encapsulated in the dialect. Although the book
provides us with a detailed account of this world and its people in the
translated version, too, its readers can never participate in the homely beauty
of the regional way of speaking.

The fate of puns, or the play on words, is in most cases similar. In his

nightmarish poem Security Considerations, placed between the ninth and
the tenth Canto of his poem sequence The Sinking of the Titanic,
Enzensberger presents himself as trapped in a crate with a heavy lid, which
he eventually manages to open by a chink, but only for a brief moment.

In the translation the first verse starts with the neutral

I am trying to lift the lid,

the third verse accelerates into:

. . . I knock,
I hammer against the lid
I call out More light, I gasp,
logically, pounding away at the hatch.

which is once more intensified in the seventh verse:

Thus I break my very own back
against the lid. Now!
A chink, a narrow gap! Ah!
Marvelous! The open country
outside, . . .

to culminate in the ninth, the last verse:

And hence I cry: I express
my regrets, woe to me,
my very own regrets,
while with a hollow plop
the lid, for security reasons,
comes down again
over my head.

(E 1989: 30f)

But in German the ninth verse is:

Also, Ich drücke, rufe ich
mein Bedauern aus, wehe mir!
mein eignes Bedauern,
während mit dumpfem Pflupp

Relativizing optimality

157

der Deckel sich wieder,
aus Sicherheitsgründen,
über mir schließt.

(E 1978: 41)

Ich drücke mein Bedauern aus draws the reader into a lexical garden path as
drücken, which we encounter first without its affix aus-, seems to continue
stemmen, pressen gegen/press against, before the next line turns it into
ausdrücken/express. It is the poet with tongue in cheek, the playful
philosopher, who balances the nightmare with his witty reflections and plays
on words. Yet not even the poet himself – who happens to be the (much-
admired) translator – could get the pun across.

Intended ambiguity, redundancy, repetitiveness – there are as many ways

of relativizing optimality as there are different ways of using a language over
and above the default use of conveying a message. Multiplied by the different
concepts and images associated with the lexical elements of the languages,
the number of translational problems awaiting us beyond the grammatically
parametrized differences between source and target language grows
exponentially, keeping up tension for the rest of our professional lives.
Where idiosyncrasies dominate, the translator seems doomed to the life of
Kafka’s trapeze artist:

A trapeze artist – this art, practiced high in the vaulted domes of the
great variety theaters, is admittedly one of the most difficult humanity
can achieve – had so arranged his life that, as long as he kept working in
the same building, he never came down from his trapeze by night or day,
at first only from a desire to perfect his skill, but later because custom
was too strong for him.

And although:

he drew a stray glance here and there from the public . . . the manage-
ment overlooked this, because he was an extraordinary and unique
artist.

(Kafka 1988: 231)

Generally, though, conditions are mixed and parametrized optimality

offers a hold that is relatively firmly rooted in the typological characteristics
of the languages translators work with. Reordering, reframing, reducing,
extending, recategorizing, relexicalizing and so on and so forth, they can,
often enough, get quite close to the original:

Ein Trapezkünstler – bekanntlich ist diese hoch in den Kuppeln der
großen Varietébühnen ausgeübte Kunst eine der schwierigsten unter
allen, Menschen erreichbaren – hatte, zuerst nur aus dem Streben nach

158

Relativizing optimality

Vervollkommnung, später auch aus tyrannisch gewordener Gewohnheit
sein Leben derart eingerichtet, daß er, so lange er im gleichen
Unternehmen arbeitete, Tag und Nacht auf dem Trapeze blieb.

and although:

hie und da ein Blick aus dem Publikum zu ihm abirrte . . . verziehen ihm
dies die Direktionen, weil er ein außerordentlicher, unersetzlicher
Künstler war.

(Kafka 1970: 155)

Agreed?

Flouting the Maxim

Whatever the conventionalized or individual constraints on optimal process-
ing of the original were, the translations were more or less in line with the
Maxim of Translation, as set out in the Key, securing optimal processing
within the constraints of the corresponding target language options. How-
ever, the maxim of staying as close to the meaning and style of the original as
is possible within the constraints of the target language can be flouted either
partly or altogether – which is by no means a rare case in existing
translations. Although optimal processing may also play a role in the final
product, tight bonds to the original may be loosened or severed for reasons
other than processing ease. Thus, for example, as Toury (1995) demon-
strates, adaptations to certain literary norms may become a controlling
factor and this could result in a relativized optimality which was not present
in the original. The history of translation abounds with examples where the
ties to the meaning and style of the original were cut more radically than
would have been necessary for optimal processing in the target language.
Many of the more drastic cases cited in the literature present obvious
violations of the default Maxim of Translation – however justified their own
codex may be within the discourse norms of the time. But these are clearly
topics of descriptive/historical translation studies and their fascinating
aspects lie beyond the domain of processing conditions.

With the exponential growth of pragmatic translations and the advent of

machine (-aided) translation, the Maxim is flouted for practical reasons: lack
of time or technical capacities. ‘Quick and dirty is better than perfect and too
late’ (quoted in Stoll 2000: 250) is the slogan that replaces the Maxim for
texts which are ‘ephemeral . . . no more than transitory manifestations . . .
of a temporally bounded social relationship. When the “social moment” of
such a text passes, its . . . significance . . . dissipates’ (Shreve 2000: 219).

Nevertheless, all the culture-specific conventionalized or idiosyncratic

deviations from the original which are not in line with the default Maxim
will always, to a greater or lesser extent, be embedded in linguistic encodings

Relativizing optimality

159

of the more general type, where the yardstick of parametrized processing
could determine the starting point for an assessment of the problems and
solutions of an existing translation. The Maxim can also be used successfully
if the purpose of the original has been changed in the translation

as long as

the skopos is associated with a conventionalized form of linguistic encoding
and/or a characteristic type of relativized optimality. Flouting the Maxim
will in these cases simply mean replacing the original subcode by a subcode
appropriate to the new purpose. At the level of abstract models this presents
no problem, but, as the Key has shown, we still have a long way to go to fill in
the language-specific details.

Reviewing the scene

160

Having looked at several dozens of examples, interpreted the original in its
context and compared systematically varied translation versions with regard
to their contextual appropriateness, we have reached a point where we
should step back and review our findings. The crucial question I asked at the
beginning of the Key was: which paraphrases should be chosen to overcome
the differences between the languages? What I wanted to find were
predictive statements, lending themselves to empirical verification. That is, I
was looking for statements of the type ‘if x then y’, and the question I have to
answer now is: have I come anywhere near this goal?

I had asked you to agree on a default Maxim of Translation, against which

we could measure the quality of a translation; the Maxim required the
translation to retain as much of the meaning and style of the original as was
compatible with an appropriate use of the target language. I had suggested
basing appropriateness on the Principle of Optimal Relevance as a universal
principle governing the use of language through the presumption of an
optimal trade-off between cognitive gains and processing effort. This trade-
off was claimed to be controlled by language-specific conditions on optimal
processing.

Three major areas of language-specific differences were thematized in the

Key: order, within the sentence and beyond it; perspective (the projection of
semantic roles onto syntactic functions); and explicitness (the use of overt
linguistic structures versus implications or implicatures). The analyses of the
examples revealed a high degree of interdependence between order and
explicitness, but also some interdependence between order and perspective.

Parametrized focus structures and theoretical implications

The general result of the analyses confirmed that, despite innumerable idio-
syncrasies, the linguistic structures in source and target language that are
related to each other through optimal translation differ in regular ways
which can be explained by parametrized processing conditions. In
particular, two grammatical parameters were identified with alternatively
set values underlying such observed preferences as an alternative order of

Reviewing the scene

161

elements in English and German, and different degrees of structural explicit-
ness and/or different perspectives. They are the parameters of directionality
and configurationality, which jointly constrain the freedom of word order
and the language-specific ways of compensating for the processing dis-
advantages arising from these constraints.

Processing ease in identifying focus structures emerged as the dominating

aspect, unifying most of the differences. There were two more aspects that
played a major role, preconditions, as it were, for focus interpretation since
they belong to the more basic stages of language processing – parsing
(structural attachment) and anaphora resolution (identifying referential
antecedents).

Concentrating on focus interpretation, I suggested adopting the linguistic

assumptions about a syntactically determined, neutral focus position for
German and English. Although a sentence may have several foci, the main
focus is assumed to be prototypically associated with the position of the
verb: it is, as a rule, verb-adjacent. Due to the different directionality of
German and English, verb-adjacency can entail different positions within the
sentence as a whole. In contrast to traditional assumptions (as for instance in
Quirk et al. 1985) and in a very much simplified way, English was
characterized as a mid-focus language, the focus being expected close to the
beginning of the verb phrase. German is an end-focus language, the focus
being expected at the end of the verb phrase (with or without verbal elements
after it).

With regard to the semantic-pragmatic side of focus, the structurally

focused elements were interpreted as the most relevant elements relative to
their sentence-internal context. In a well-written text, including translations,
the sentence-internal interpretation was said to match the one following
from the sentence-external context. But due to the parametrical differences
between left- and right-peripheral verb phrases, analogous translations can
result in a mismatch between the structural focus and the contextual focus.
The typologically determined processing difficulties of analogous trans-
lations were shown to carry over from the simple to the complex sentence
and even beyond sentence boundaries to sequences of sentences.

A main psycholinguistic assumption in the Key was that focus interpreta-

tions are first read off from the linguistic form of a sentence before they are
integrated with the information of the preceding discourse. If an analogous
translation results in a mismatch between structural and contextual focus,
the processor will find itself, at least temporarily, on a garden path regarding
information structure. Although the interpretations can still be adapted, that
is, reanalysed in line with contextual conditions, the need for such reanalyses
violates the Principle of Optimal Relevance and hence also the Maxim of
Translation pursued in the Key. Meeting the Maxim of Translation means
restructuring the analogous version into a paraphrase which secures optimal
processing conditions, especially for structural focus identification.

The same applies whenever there are two foci in a sentence which are not

162

Reviewing the scene

clearly separated from one another. The processor, by default, expects only
one focus and will miss the other focus unless there is an extra formal clue.
Garden paths are avoided by focus separation (or additional lexical means).

In general, the parametrized processing conditions of German and English

promote different solutions in many cases. Here are some of the differences
we have encountered:

(a) Material that has been topicalized (or scrambled) in German in the

interest of end-focus is presented in its basic position in English (even
if this results in the order focus-background; recall the examples in
Chapters 2 and 3).

(b) Sentences that are used in a passive (-like) perspective in German to

secure neutral focus interpretation or focus separation are presented in
an active structure in English, and initial adverbials may be reframed as
subjects (with associated changes in the entire case frame and the
predicate) to secure easy parsing as well as neutral focus interpretation
(examples in Chapter 4).

interest of focus separation is presented in its basic position in English
but at the same time extended by a dummy structure, for example, by
clefts (compare examples in Chapters 5 and 6).

(d) If reordering, reframing or extending/reducing of structure cannot

secure focus identification, sentence boundaries are re-set. A sequence
<2,1>, with 1 as the higher value, which is a classical focus structure of
end-focus German, will be avoided in ‘mid-focus’ English by reordering
or, if need be, by separating. If the alternative case, <1,2>, is separated in
German, it may be linked in English. Each version secures the presen-
tation of focus in its classical position in English/German (compare
examples in Chapter 7).

The first group of changes serves to secure grammatical acceptability in

English. It places elements of lower relevance in final position, ignoring
discourse appropriateness in the sense of end focus, which governs topical-
ization and scrambling in German. The other groups of changes serve to
optimize discourse appropriateness in English, too. The changes in perspec-
tive help to avoid attachment problems and narrow focus interpretation in
left-peripheral, right-branching English, while the choice of passive (-like)
structures in German helps to avoid case-ambiguities and narrow focus
interpretation under alternative conditions. Structural extension in English
serves to indicate additional foci, especially in cases of greater textual
relevance. The differences are mere trends, subject to the idiosyncratic
conditions of each individual case.

Although the generalizations suggested here are supported by a wide

variety of similar cases, they all require further research. As the discussion of
the individual examples shows, there are basically two lines of research

Reviewing the scene

163

which ought to be pursued. There is, first, the discourse-based analysis of
information structure which should enable us to identify the contextual foci
of sentences and their significance relative to each other and to the sentence-
external context. The second line of research has to identify the language-
specific means of formally indicating these foci, including the constraints
limiting the use of similar means in translations and promoting or blocking
the use of substitutes.

Both lines of research are, of course, interrelated. Let me point out two

cases. First, we need a better understanding of the various types of foci that
can be distinguished on the basis of discourse relations like parallelity or
contrast. For example, different discourse topics may constitute alternative
sets excluding each other but forming no contrast. Thus, a shift from one
element to another would be assigned neutral focus but no contrastive focus.
But if such an element is presented in topic position, its focus cannot project –
unlike a non-contrastive focus in the comment. Only contrastive foci justify
the use of marked focus structures in English. Do we then have to distinguish
between a projecting and a non-projecting neutral focus?

Second, we have seen that new information can be interspersed with given

information. When do we speak of a discontinuous focus, when of several
foci? Focus-spacing or additional focus-marking devices are only appro-
priate in the case of several foci.

There are many more questions like these concerning the linguistic and

psycholinguistic assumptions needed for predictive generalizations of the
individual analyses. As each assumption has to fit in with all the other
assumptions covering related ground, we need the patience of a chess player
for each move. And as all of these questions concern the system and use of
languages, should we not just lean back and wait for the linguists and
psycholinguists to come up with answers? Moreover, if we think of all the
translation problems we have ignored, focusing on questions of information
structure and optimal language processing, we may wonder whether the
approach is worth the effort. After all, when there are so many more dis-
course functions other than informing, why restrict attention to the latter?

The answer to the last question is rather simple: informing is a most basic

discourse function and of prime importance for a wide range of texts. Thus,
if we can identify the basic strategies and language-specific conditions of
informing, we should have access to a wide variety of frequently occurring
phenomena, strengthening the predictive potential of the generalizations we
are searching for. The concentration on information structure does not deny
that there are other important topics of relevance to translation, but the
findings about information structure may to some extent even influence the
questions asked elsewhere.

On the other hand, information structure is, indeed, a major topic for

linguists studying the use of language in discourse, and processing of
information structure could be a major topic for psycholinguists, no less
fascinating than parsing or interpreting semantic roles. Both disciplines

164

Reviewing the scene

could also study information structure contrastively, which might even
include comparisons of original and translated texts. Yet so far, translations
tend to be avoided as an empirical basis in (psycho-) linguistics (except,
perhaps, in corpus linguistic studies) as there are too many additional factors
involved in the data – not the least being ‘translationese’, that is, poor
translations due to source language interferences. But then any research has
to come to terms with the problem of ‘impure’ data, and translations are,
in essence, no different from other data of language use – they, too, can be
subjected to any control procedure necessary.

Still, whatever their attitudes towards translations as an empirical basis,

linguists and psycholinguists can afford to ignore translations and concen-
trate on their traditional data. Translation studies, however, cannot ignore
translations. And as the basic principles controlling the distribution of
information are involved in any use of language, whether neutral or marked,
original or translated, translation studies also has to attend to the issue of
information structure, even if possible generalizations look like nothing in
comparison to the translational idiosyncrasies (of half a million words alone
in English). But if typological properties of a language set the frame for an
optimal distribution of information, research into the special conditions
obtaining in source and target language can help us understand some of the
more basic, intuitive strategies of language use at work in felicitous
translations, or in felicitous stylistic encoding in general. And this, I believe,
is worth the effort.

Glossary

Technical terms seen through the Keyhole

165

Acceptability

grammatical correctness. It is you, the speaker of the

language, who decides what is acceptable and what is not relative to,
say, standard English. Although there is a great deal of agreement
among speakers concerning the form of words, phrases and clauses,
judgements may vary – there are different degrees of grammatical
acceptability, and the border between grammatical and stylistic aspects
is fuzzy; languages consist of a great variety of subcodes; and languages
change, with words taking on new meanings and thus opening up new
combinatorial possibilities.

In everyday life we need not worry about grammatical acceptability.

All those sentences that linguists mark with asterisks or question marks
as grammatically more or less deviant do not occur in written texts, and
if they occur in oral speech we ignore them as speech errors. However,
there are, of course, situations in which we do worry about the accept-
ability of linguistic structures, as for example when we learn a new
language, including variants of our own language which we do not
master. Unless we do this at a very early stage in our lives, our automatic
language acquisition capacities will have dwindled away, leaving us at
the mercy of other people’s explicit knowledge about grammatical
acceptability. We are grateful for any grammatical regularities that can
be pointed out to us, helping us to avoid the production of grammatic-
ally unacceptable forms.

But how do those other people know what is grammatically accept-

able and what is not? If it is their own language, their implicit knowledge
is just as intuitive as our own, but any explicit knowledge, whether
native or non-native, presupposes linguistic research, which is at its best
when it draws upon the implicit knowledge of native speakers. So here
we are back full circle. It is you, the native speaker of a language, who
decides grammatical acceptability.

This view is, as we know, too liberal for authoritarian or pedantic

people, but you and I, who are interested in stylistic appropriateness, are
willing to base our translation studies even on the stylistic intuitions of

166 Glossary

native speakers tested by the method of systematically varied control
paraphrases.

Adverbials

mostly parts of speech modifying adjectives, verbs, verb

phrases or clauses. Some verbs take adverbials as arguments. These
adverbials are necessary and can be distinguished from free adverbials
by the criterion of deletability. If a necessary argument is deleted, the
sentence becomes grammatically unacceptable. For example, the intro-
ductory sentence to chapter VIII in Alice in Wonderland could not be
used without its local adverbial (the asterisk within parentheses is meant
to symbolize this):

A large rose-tree stood (*near the entrance of the garden).

Most adverbials, however, are free extensions of a verb, verb phrase or
sentence.

Like all other parts of speech, adverbials can consist of one word only

– even of a pronominal form – of simple and complex phrases, and of
clauses. Adverbials could be adverbs: ‘perhaps’, ‘thoroughly’, ‘most’,
‘always’, ‘apparently’, ‘necessarily’ and so on; they could be noun
phrases or prepositional phrases: ‘last week’, ‘by an average’; or they
could be clauses: ‘as the mechanism was perfected’.

Adverbials play a special role in translations between English and

German because there are systematic differences between their positions
in both languages, which are subject to the parameter of directionality,
and also because of the English preference for primary relations (see
Perspective).

Adverbials are ordered according to the position of the constituents

they modify, or the heads they belong to. What constituent they modify
is determined by the meaning of the adverbial: manner before place
before time is a well-known pattern resulting from this. Adverbials are
not only reordered in the translation, but very often ‘reframed’ as other
parts of speech, especially as subjects. (See examples in Chapter 4.)

Analogy similarity of forms in terms of word class, word forms, word

order, grammatical relations, structural explicitness, number and type
of clauses, number and order of sentences, and so on.

Grammatically acceptable analogous or near-analogous translations

are the starting point in our search for an optimal translation.

Anaphors, anaphoric relations

referential identity between a referent and a

contextually preceding element. The latter can be considered the –

Antecedent

to an anaphor. The element which refers to or specifies a

subsequent referent, as for example the hurricane which specifies it in
the signs of it in Hemingway’s sentence: If there is a hurricane you
always see the signs of it in the sky (discussed in Chapter 5).

Appropriateness

fitting the context, locally and ‘globally’. Major factor for

an appropriate use of language is the Principle of Optimal Relevance.

Glossary

167

Special uses of language are characterized by constraints on this prin-
ciple (see Chapter 8).

Argument

The meanings of words open up positions which are to be

specified by the syntactic context of the word. Most variable in this
respect are verbs, which can take up to three or even four arguments,
directly or via prepositional mediators: ‘someone gives someone some-
thing’, or ‘someone gives something to someone’. But such meaning-
based ‘valency’ is not restricted to verbs. Nouns and adjectives, adverbs
and prepositions show the same combinatorial effect, though their
variability is somewhat restricted in comparison with verbs: ‘a book
about something’, ‘a challenge for someone’, ‘an interest in something’,
‘to be similar to something’, ‘to be tired of something’, ‘to be far off
something’ etc.

What arguments are filled into the slots is also predetermined by their

syntactic heads or ‘governors’: phrases with or without prepositions,
clauses, animate or inanimate elements, agents or patients, etc. And as
all this may be different in different languages, including additional
formal distinctions in the forms of words expressing case, the choice
of the grammatical head predetermines the linguistic structure of a
sentence to a great extent.

Attachment

adding a structural element to the ‘partial current phrase

marker’, which is built up when parsing sentence structure. The Prin-
ciple of Minimal Attachment makes the processor choose the lowest
node possible, that is the most immediate syntactic phrase admitting
attachment (see the corresponding section in Chapter 4).

Attributes

modify noun phrases; they may consist of a word or a phrase or

a clause, used before or after their heads. Their positions relative to their
heads and to each other is predetermined by the class they belong to,
very much as the position of adverbials is predetermined by their
relations to each other and their common head.

Background

complementary to focus, limits the extension of focus

projection. The part of the message that is presented as if it were already
known to the addressee. As definite phrases signal givenness, the author
can present something as background in the very first sentence of a
story. Thus, for example, the first sentence in Iris Murdoch’s novel The
Sacred and Profane Love Machine:

The boy was there again this evening.

refers to a time, a place and an individual as if we were already familiar
with them. The only new element, again, refers to a recurrence of an
event, presupposing at least one similar event preceding it – the boy had
already been there some other time – which means that everything but
the recurrence of the event belongs to the background of the sentence.

Case

the syntactic role of arguments within the structural hierarchy of a

168 Glossary

sentence. Often expressed by certain word forms (as for example by
German nominative, accusative, dative, genitive) or by certain structural
configurations of the syntactic hierarchy. In the latter sense, case
corresponds to syntactic functions or grammatical relations (like
subject, object).

Category

word class; the major (lexical) categories are nouns, verbs,

adjectives/adverbs and prepositions. They determine their ‘argument
structure’; minor (grammatical) categories are articles, pronouns,
auxiliaries and the like.

Clefts

syntactic focus indicators with specific structural characteristics

compensating for topicalization restrictions in English. There are three
canonical types of clefts: it, wh-, inverted wh-clefts and many more cleft-
like structures. Most of them mark a second focus by doubling the
clausal structure. A sentence like:

It is the context that fixes the meaning

allows a focus on meaning or on context or on both, depending upon the
context – as the sentence says.

The prototypical cleft is characterized by a presupposition with a

variable which the cleft specifies by a contrastive focus, that is, a focus
that rejects any alternative specifications of the variable. For our
example this means the presupposition – there is an x which fixes the
meaning – and the assertion – x is the context (compare Chapter 6).

Comment part of sentence meaning commenting upon the topic.
Compound

complex words, particularly frequent in German, for example,

Bestandsaufnahme, Ertragsfähigkeit, Strukturanpassung, Währungs-
umstellung, Anteilsrecht. Most of them are decomposed in the
translation.

Configurationality

grammatical parameter of languages, determines case

by structural configurations; for example, in English, object as verb
phrase internal constituent, subject as verb phrase external constituent;
restricts positional variations.

Constituent

structural part constituting the syntactic hierarchy of a

sentence.

Constituent hierarchy

syntactic architecture of sentence structure, groups

words into ever larger phrases, with at least one noun phrase and one
verb phrase constituting a sentence. The entire structure can be viewed
as a tree with phrasal branches (a so-called phrase marker). Each phrase
consists of a lexical category specifying the head of the phrase plus
structural extensions, especially complements, filling in the structural
slots (arguments) predetermined by the meaning of the head, and
specifiers like articles (compare section ‘Linguistic knowledge’ in
Chapter 1). Modifiers may be added freely, at various hierarchical levels
without raising the level of hierarchy. The properties of constituent

Glossary

169

structure are universal – that is, they determine the architecture of
sentences in all languages – but on them are superimposed parametrical
variations, as for example the parameters of directionality (determining
the direction of the structural extension of the head) and configur-
ationality (constraining deviations from the basic word order).

Context

surrounds an element in discourse, ‘locally’, within the same or

the preceding sentence, or ‘globally’, in preceding and subsequent
passages of a text, within the situation in which the text participates, and
within the cognitive model we have gained from the text and from all the
knowledge we can bring to bear upon it. The big theoretical question is
how the various contextual factors interact with each other and the
element for which they form the context.

Contrast

discourse relation between features excluding each other.

Contrastive focus

contrastively interpreted focus. One of the two or three

major types of focus interpretation (see also ‘narrow focus’)

Copula

‘be’ used as a main verb, yielding syntactically somewhat puzzling

structures, especially in a language like German where the subject is
identified by its nominative case and is not bound to the position before
the verb. Thus in a sentence like:

Der Gärtner ist der Mörder.
The gardener is the murderer.

we cannot really say which of the nominative cases is the subject of the
sentence. Frequent reorderings in translations suggest that the initial
elements in German sentences are often predicates topicalized for
discourse reasons.

Consider for example the German translation of:

The Moon is the natural object as our first source of space
materials.

as:

Unsere erste Rohstoffquelle im Weltall ist natürlich der Mond.

where English uses a predicational sentence, predicating of the moon
that it is our first source of space materials, while German uses a
specifying sentence, saying that there is a first source of space material
which is the moon.

In a context about space materials, reference to the moon has to be

considered new information and is used in a structural focus position in
German, while English prefers the predicational view in favour of
primary relations.

Decomposing

translating compounds as syntactic phrases making use of

their morphological or semantic build-up.

170 Glossary

Definiteness

reference to a certain set of one or more elements; in

opposition to indefinite. Prototypically expressed by articles: the for
definiteness and a or a bare noun for indefiniteness. A boy, the boy,
boys, the boys, three boys, the three boys . . . Definiteness signals
givenness, but not necessarily when used for concepts referring to unique
objects, as the moon; indefiniteness signals novelty, except, perhaps, for
cases referring to abstract objects, like evolution or creation. Both
exceptions are cases of underspecification, which can only be decided
with the help of context.

Defocusing reinterpreting the structural focus as background. Definite

noun phrases in canonical focus positions will have to be reinterpreted
as background if there is another element marked as focus. Consider:

There are a number of arguments that support this hypothesis,

which marks its subject as focused by there are, suggesting that we
should defocus the object after the main verb. The definiteness of the
object reaffirms the decision because it signals givenness (see back-
ground).

Diachronic

refers to the development of languages, complementary to

synchronic.

Directionality

grammatical parameter determining word order to a large

extent, by the position of categorial heads at the right or left periphery of
their structural extensions. While verbal and prepositional phrases are
left-peripheral in English, German phrases are less consistent: verbal
phrases are basically right-peripheral, prepositional phrases are basic-
ally left-peripheral. The different directionality is associated with
different focus expectations (see structural focus).

Discourse

usually a longer text about a major topic, addressed to a real or

virtual partner, if need be to oneself. A discourse is structured into larger
segments with relations to each other, constituting the macrostructure
of the entire unit. Macrostructural units are made up of sequences of
sentences, which in turn are characterized by the microstructural
relations between their elements. While written forms of discourse
develop their topic in an orderly fashion, oral forms of discourse are
more likely to progress in an erratic way. Although literal translation
involves much of the erratic type of discourse, holding one’s balance on
the tightrope is difficult enough with the more orderly type of discourse.
Thus, the Key focuses on written discourse. As the last chapter shows,
there are many ways of relativizing optimality, yet processing in dis-
course can be an enlightening concept for special uses of language, too.

Discourse relations

semantic or pragmatic relations between discourse

segments, like contrast, continuation, elaboration etc. Of utmost
importance for optimizing discourse linking (in terms of information
structure, anaphoric relations and the like).

Glossary

171

Dummies

elements with little meaning used to fill a syntactic structure for

grammatical or stylistic reasons. For example, to disambiguate ana-
phoric relations or to promote a correct focus interpretation.

Equivalence

similarity of meaning and style despite different linguistic

encodings in different languages or within the same language (see
Paraphrase). If there is an additional formal similarity, the Key speaks of
‘surface equivalence’. Surface equivalence is mostly sacrificed in the
interest of grammatical acceptability and optimal processing. Equiva-
lence in the wider sense, including implicit meaning based on the mental
model built up from linguistic and extralinguistic knowledge associated
with a message, is a defining criterion of translations. The fact that
surface equivalence has to be given up in the interest of contextual appro-
priateness makes the overall equivalence depend upon a wide range of
contextual factors related to the various text types, stylistic registers and
translational purposes. This can even lead to the apparent paradox that
formal analogy will be given priority over equivalence to some extent (as
when translating poems, for example). The different nature of
contextual factors made translation theorists come up with various
concepts of equivalence (for an exhaustive survey, see Koller 1992).

Eventuality

cover term for any type of event or state.

Explicitness

the part of a message encoded linguistically. Only part of what

a message says is said explicitly, that is, with the lexical and grammatical
means of a language. Quite a lot of what is said is said between the lines,
implied, suggested, presupposed. Some of this indirect meaning is
associated with the linguistic meaning; it is ‘conventionalized’. A great
deal of the implicit meaning is added on the basis of what we know
about the states of affairs referred to, knowledge which – ideally – is
shared by those participating in the exchange of a message.

Focus

presented as most relevant for the further progress of discourse.

Theoretically, we are free to choose any element as focus, but practically
we are constrained by the context and the language-specific conditions
for focus assignment – at least, as long as we heed the Principle of
Optimal Relevance.

It is mostly the preceding context which decides upon the relevance of

an element, as new versus given information or as contrasted informa-
tion, excluding alternative possibilities. Focus assignment should be in
line with the contextual conditions. Focus assignment is, as a rule,
realized prosodically: one word, or rather syllable of the focus is marked
by pitch accent. If the focus is restricted to this word, that is, to the focus
exponent, it is a ‘narrow’, restricted focus and can be interpreted
contrastively; if the focus consists of more than the focus exponent, it is a
wide, projecting focus. The extension of the focus projection is deter-
mined by the context (see Background). Written texts require recon-
structing the prosodic accent. Psycholinguists speak of phonological
coding (‘silent voice’).

172 Glossary

There are two formal devices that help us to identify the focus at an

early stage of sentence processing: lexical elements (focusing elements)
which, by their meanings, focus an element in their scope, and certain
structural configurations, including the position of an element relative
to its syntactic head. There are certain grammatical rules governing the
assignment of focus or pitch accent to a certain element of the sentence.
The main focus of a sentence is normally associated with a focusable
element in a verb-adjacent position. See the sections from ‘Structural
focus’ to ‘Focus projection’ in Chapter 2 for a more detailed presentation.

A sentence may contain more than one focused element. If there is

more than one focus in the sentence, the Principle of Optimal Relevance
requires us to use additional marking devices and distribute the foci
evenly in the sentence, i.e. to make use of focus spacing.

Gap

lexical or grammatical element of the original missing in the target

language.

Garden path

processing error by taking the wrong ‘path’ when analysing

(parsing) or interpreting linguistic structures, including erroneous
discourse linking of anaphors and focus identification.

Generic

complementary to particular, referring to all possible, i.e. also

virtual, instances of eventualities talked about.

Grammar

language competence, in its widest sense comprising lexical

knowledge, that is, knowledge about the meanings and forms of words
and their syntactic properties that permits them to be inserted into the
constituent hierarchies and structural derivations (transformations)
available to a particular language.

Illocutionary, illocutive part of sentence meaning determining the speech

act in which a proposition can be used.

Implication strongly implied meaning.
Implicature weakly implied meaning.
Information structure discursive partioning of sentence structure into two

or more, bi- or tripartite layers: topic and comment, focus and back-
ground, corresponding to two or three subsegments with different
information values.

Informational density

proportion between total meaning and explicit

information. Distributing the same amount of information onto more
linguistic elements (for example onto more clauses) lowers density; vice
versa, distributing the same amount of information onto fewer linguistic
elements (for example onto more deverbal nouns) increases density
(compare section ‘Statuesque and dense’ in Chapter 8).

Interpreting oral form of translating, where the Principle of Optimal

Relevance is heavily constrained by the lack of control phases during
perception and production. (Lambert and Moser-Mercer 1994 present
an impressive collection of psycholinguistic and neurophysiological
studies on the effects of these constraints.)

Language processing

the process of perceiving and producing linguistically

Glossary

173

encoded messages. The Key concentrates on the former perspective. (For
a comprehensive presentation of the latter cf. Levelt 1989). The linear
structure of a message is processed incrementally, that is, step by step.
All the various layers constituting the message on the syntactic and
semantic levels of the linguistic structure are processed in a parallel way.
When we perceive a sentence, the semantic and the pragmatic interpre-
tation lag slightly behind the syntactic analysis. The integration of the
message into the context is performed at certain wrapping-up junctures
(for example, at full stops and semicolons, but also at shorter intervals).
The processing mechanism is carried out by a specialized component of
our mind, the processor, which follows its own strategies determined by
the specific nature of language and the general principle of economy. As
languages differ in their forms and the way they are used, processing
strategies meet with language-specific processing conditions. The basic
requirement for stylistic appropriateness takes account of the language-
specific conditions for efficient processing. If the first-pass analysis
breaks down (see Garden path), the structure is reanalysed.

Language-specific

characteristic of an individual language. There are

innumerable features distinguishing individual languages in terms of
syntax, semantics and phonetics. The basic architecture of languages,
however, varies only in a few grammatical parameters, like those of
directionality and configurationality. These have far-reaching conse-
quences for questions of style because they determine optimal processing
conditions in discourse. The fact that German contrasts here with most
European languages, including English, makes it a rewarding object of
contrastive and translation studies.

Linearization

order of linguistic elements, ‘word order’. Determined by

syntactic features of the individual lexical elements: category (word
class), argument structure of the word, scope. The universal principle of
‘lexical projection’ is parametrized by basic grammatical properties of
the individual language (directionality, configurationality), determining
the basic linear order of sentence structure and transformational
options. Languages differ not only in their grammatically obligatory
transformation (as, for example, moving the verb into second position in
German main clauses), but also in their discourse-based transformations
(scrambling or topicalization of given elements in German). Constraints
on word order variability may be compensated to some extent by a
greater lexical variability (see perspective).

Modifier

complementary to argument (see Adverbials, Attributes).

Optimal relevance

an optimal proportion between cognitive gains and

processing efforts. Cognitive gains are made on the basis of existing
assumptions, which can be confirmed, rejected or extended by new
information. Processing effort is determined by the difference between
the linguistically encoded input and background knowledge available
to the processor (that is, the background knowledge needed for the

174 Glossary

interpretation of a message), but also by the general processing con-
ditions of the particular situation and the processing conditions of the
linguistic coding.

Parameters

basic grammatical properties of languages determining their

typological characteristics in regard to a small set of alternative
possibilities (see Directionality and Configurationality). Parametrical
variations are highly relevant for the language-specific processing
conditions.

Paraphrase

contextually equivalent sentence but rarely equally appro-

priate sentence. In its wide sense determined by the overall interpre-
tation of a message within a certain context, irrespective of the amount
of temporary undecidedness and reanalyses necessary to get there.
Systematically, that is, minimally, varied sets of paraphrases – ‘control
paraphrases’ – are used as an empirical method of controlling the
contextual appropriateness of linguistic encodings in discourse.

Parsing

processing of syntactic structure. Building up the constituent

hierarchy of a sentence, it is necessary to hypothesize about the syntactic
category of an element and its syntactic relation to preceding elements,
as well as about possible future elements of the linear chain of words
processed incrementally (that is, step by step). The processor has to
attach incoming material while anticipating its syntactic position and its
syntactic function within the entire constituent hierarchy of the
sentence. There is some evidence, such as garden path effects, that the
processor follows certain strategies like Minimal Attachment in solving
structural ambiguities, and in reanalysing the outcome should the
analysis break down, that is, make no sense at some later point.

Particular

complementary to generic, referring to a specific instance.

Perspective

relation between the semantic roles and syntactic functions of

constituents; compare for example the difference between an active and
a passive perspective, where the role of an agent or the role of the patient
is projected onto the subject:

Someone is translating the book.

The book is being translated.

or the transitive/intransitive difference:

Someone opened the door.

The door opened.

or ‘middles’ like:

The book sells well.

Glossary

175

and some other verb classes, all of which raise lower roles to the highest
rank in the syntactic hierarchy, thus shifting the perspective of the
eventualities presented. Languages differ in their lexical possibilities for
such a shift in perspective. Thus, for example, many of the intransitives
projecting the role of the patient into the subject in English are reflexives
in German. Compare:

Die Tür öffnet sich.
The door opens.

Das Buch verkauft sich gut.
The book sells well.

In many cases, the projection of the lower role into the subject is blocked
by selection restrictions. Consequently, circumstances and causes are
projected into adverbials rather than subjects. With English preferring
primary relations, German adverbials may be reframed as English
subjects and vice versa to improve processing conditions on either side.

Phrase marker

constituent hierarchy of sentence structure. At any stage

of parsing, the processor proceeds from the ‘partial current phrase
marker’ it has built up so far, attaching incoming structures according to
expectation.

Pitch accent

stress associated with a shift in pitch level, for example, from

high to low level or low to high level, signalling focus. The assignment of
pitch accent to a constituent of a sentence is determined by grammatical
rules concerning the position of the element in the constituent structure
of the sentence or with respect to a focusing lexical element. If we assign
pitch accent to any other constituent than the grammatically predeter-
mined one, we will get a marked focus interpretation.

Postponing

moving an element to a position after its head; for example,

placing the relative clause of a subject at the end of the sentence.

Presuppositions

implicit meaning exempt from speech act performance.

Thus:

Why didn’t you tell him?

presupposes that you didn’t tell him and asks for the reason; that is, the
negated proposition is presented as a fact and is not included in the
question.

Primary relations

syntactic functions of arguments; complementary to

secondary relations, that is, modifiers. English prefers primary relations
as it is a configurational language.

Processing see Language processing.
Pro-forms

grammatical words with little meaning of their own, replacing

meaningful words, phrases or clauses, whose referential properties they
inherit (for instance, ‘it’, ‘she’, ‘they’, ‘there’, ‘one’) (see also Anaphors).

176 Glossary

Proposition semantic unit consisting of at least an argument and a

predicate related to it. Propositions are objects of operators quantifying,
negating, modalizing them, as well as of attitudes, evaluating them and
transforming them into objects of illocutions.

Prosody

intonation and stress patterns associated with utterances (also by

the ‘inner voice’ of silent reading); particularly important property:
pitch accent.

Psycholinguistics

discipline studying the acquisition and use of language,

in particular, first language or second language acquisition, language
processing in perception and production, and deficient language use,
like speech errors and aphasia (linguistic pathology). There are various
schools of thought with extremely divergent theoretical positions aiming
to explain the empirical (as a rule, experimental) data gathered by
strictly controlled methods, including neurophysiological measurements
of brain potential – a more recent development – testing theoretical
assumptions about the cognitive aspect of language use.

Reanalysis

adjusting grammar- or discourse-incompatible results of

processing. Pursuing processing strategies like Minimal Attachment, the
processor may be led up the garden path by temporary ambiguities of the
incoming elements, which it has to identify and integrate into the
syntactic hierarchy under construction (the ‘current partial phrase
marker’). In its attempt to adjust problematic structures, the processor
checks our grammatical (including lexical) and discursive competence
for alternative analyses that would avoid the garden path. It works
along the lines of grammatical dependencies and discourse linking and
is controlled by the economic Principle of Optimal Relevance as any
first-pass analysis. Minimizing processing effort in translations means
reducing the need for reanalyses. (Compare Fodor and Ferreira 1998,
for a representative range of current approaches to reanalysis.)

Referent

individual or eventuality identified by linguistically encoded

concept. In line with the many different encodings (for example as noun
phrase, non-finite verb phrase, that-clause etc.), there are many different
types of referents, including all sorts of abstract objects.

Relevance

importance for progress in discourse, relativized to effort in the

Principle of Optimal Relevance.

Role

semantic or thematic role of referents in the event they belong to, as

agent, patient, instrument, place, time etc. The projection of roles into
syntactic functions determines the perspective of a sentence.

Scope

domain of quantifying or modifying elements, determines which

elements are affected by the quantifier or modifier and thus, to some
extent, the basic order of these elements.

Scrambling deviating from basic word order by swapping positions. In

German quite freely, in the interest of discourse linking; much more
constrained in configurational English.

Selection restrictions

semantic constraints on the combination of words.

Glossary

177

Violations of these constraints may be intentional, as in the literary use
of language or in translations extending the expressive potential of the
target language, but in most cases they will be unintentional, resulting in
semantic mismatches of a more or less distorting nature.

Semantics

the meaning of linguistic structures. Although we have given

priority to pragmatic concepts like optimal relevance and contextual
appropriateness, semantic aspects are highly relevant for translations
since translations are by definition characterized by (a certain degree of)
equivalence between the meanings of the original and the translation,
and a large part of equivalence is determined semantically.

There are various layers of semantic meaning to be distinguished. The

inner core of the various layers is the referential meaning of all, at least
potentially, referring elements. It consists of two basic types: individuals
or events (called eventualities to cover states, too) and predicates, which
take us from individuals to eventualities. On the side of the semantic
form we speak of arguments and predicates forming propositions, which
can, in turn, be used as arguments to form more complex propositions.
The basic semantic types are subjected to all sorts of additional specifi-
cations and modifications, determining among other things the quantity
of the reference and the modality of the predication in terms of truth
or likelihood. On the layer of quantified and modified propositional
meaning is superimposed the layer of propositional attitudes, expressing
the speaker’s opinions, often relative to other opinions; the attitudinal
meaning includes emotional evaluations of the eventualities referred to.
The layer of evaluated propositional meanings is overlain by the layer of
illocutionary, performative meaning, determining the type of speech act
(like question, order, assertion etc) performed by the utterance of the
linguistic forms carrying all these meanings.

Some of the linguistically carried meaning is expressed explicitly,

some implicitly through implications or presuppositions. But the mean-
ings of sentences are not only determined linguistically by the semantics
of the words and their syntactic relations to each other, but also by our
common knowledge about the individuals and eventualities talked
about and the discursive context, situational or textual, to which the
sentence belongs. This includes all the inferences we can draw on the
basis of our knowledge, presuppositions, implications and implicatures
(the weak type of implications, based only on likelihood). In translations,
we can compensate for the differences between the linguistically deter-
mined meanings not only by composing meaning through different
combinations of the means of the target language, but also by redistri-
buting meaning between the explicit and implicit, linguistic and non-
linguistic components of knowledge. (For a theoretically comprehensive
and ‘hard’ approach to the vast area of semantics cf. Wunderlich and
Stechow 1991.)

Speech acts

acts performed by uttering sentences; determined by the

178 Glossary

illocutionary meaning of a sentence and the context. Thus, for example:

Can you tell me the time?

can be a real question to a child, asking about its cognitive capacities,
but a request to someone whom we can expect to have these capacities.

Style conventionalized or individual variety of language use. Lexical or

grammatical options are either ‘neutral’ or marked for special stylistic
levels (casual, substandard, etc) and registers (literary, academic, etc),
including regional, social or aesthetic variants. The Principle of Optimal
Relevance, together with the parametrized grammatical conditions of a
language, determines the general stylistic profile that distinguishes a
language from other languages (as for example the greater positional
freedom or discourse sensitivity of German word order versus the
greater lexical and prosodic freedom of English subjects, as discussed in
Chapter 4).

Synchronic

at one and the same stage of development.

Syntax grammatical component determining the combination of words in

word groups, clauses and sentences. There is a universal inventory of
categories (word classes), and universal principles for the extension of
these categories into phrases and their completion into clauses or
independent sentences. Languages differ in the basic direction of the
structural extensions, requiring the categorial head at the left or right
periphery of its phrase. According to the position of the verb – the
dominant category in sentence structure – English can be said to be a
left-peripheral language, extending its verb phrase to the right, while
German extends its verb phrase to the left, whether the verb is in its basic
right-peripheral position as in the subclause, or in the left-peripheral
position of the main clause. The latter has to be seen as derived from
basic structures. There are universal and language-specific constraints in
the derivation of sentence structures, concerning, for example, the par-
ticularities of what can move where. Moving elements to the beginning
of sentences, for example, is much more constrained in English than in
German. But the greatest difference in the syntactic options of languages
are the particular constraints laid down in the syntactic properties of the
individual lexical element. There is the set of major categories, such as
verbs, nouns, adjective, adverbs and prepositions, which determine the
types of arguments they are used with, their number, case and order.
Then there is the set of grammatical word forms, inflections, carrying
the grammaticalized meanings of tense, mood, voice, gender, number
and person, which differ greatly from language to language.

Text

linguistically formed unit of discourse; written or oral whole of

varying length unified under one major topic.

Textual relevance

relevance of individual elements beyond the immediate,

local context; determines the importance relative to each other of

Glossary

179

elements participating in discourse relations, structuring the text into
larger segments.

Topic (of sentence or discourse) the element the sentence or discourse is

about, which need not be the same.

Topic in the sentence is the element that the sentence comments upon.

Together, topic and comment constitute one layer of information
structure. Although topics may have their own focus, the main focus of a
sentence will mostly reside in the comment, so that topic is often simply
juxtaposed to focus.

Topicalization moving an element into structural topic position, which

amounts to the initial position of a sentence. Initial position can, how-
ever, mean different things in terms of the syntactic hierarchy of
sentence structure, as, for example, a sentence-internal position in
German versus a sentence-external position in English, where the
subject-before-verb condition pushes the landing site for topicalization
further out. The difference makes itself felt in the greater markedness of
the topicalized element in English.

Transformations

structural changes turning basic syntactic structures into

derived structures, mostly by moving elements to other positions (see
also Topicalization, Scrambling, Postponing).

Translation studies, translation theory

discipline focusing on translation

and translating, often including interpreting. Compared to genetics,
computer linguistics, biochemistry and other more recent disciplines,
translation studies is quite of age now, but much less accepted in
academic circles. For most of the time, theories on translation have
focused on literary translation, including bible translation, where
linguistic aspects of translation are heavily dominated by aesthetic and
cultural aspects. Thus, although translation studies has largely followed
the developments in linguistics, starting with a transformational
approach (Nida, Catford), via a communicative approach (for example
Kade, Neubert, Reiß and Vermeer) towards a cognitive, psycholinguistic
approach (Gutt), linguistic issues have been increasingly neglected or
explicitly rejected as genuine topics of translation studies. The major
tendency today places translation studies among cultural studies and
looks for the special norms controlling translations or even influencing
the target language/target culture through translations (see for example
Toury, Snell-Hornby). Yet whatever the merits of the cross-cultural
approach may be or will be in the future, there are vast areas of
translational problems, including those of machine translation (as spelt
out in Kay et al. 1994), that cannot be tackled, theoretically or prac-
tically, without recourse to linguistics – as the Key amply demonstrates.

Notes

180

1 Setting the scene

1 A high degree of equivalence will thus come very close to the concept of ‘optimal

equivalence’ as defined – and strikingly illustrated – in Toury 1983: 117.

2 This contrasts with Baker 1992, who discusses many of the aspects to be taken up

in the Key and often in a similar vein, but who says: ‘It is in fact virtually
impossible . . . to draw a line between what counts as a good translation and what
counts as a bad one’. House 1997, on the other hand, vigorously advocates the
possibility of quality assessments, albeit in a much wider sense of the convention-
alized properties of source and target texts.

3 Comprising all the aspects spelled out in Speech Act Theory; cf. Searle 1969.

2 Questions of order

1 Whether they are of the more traditional type, like Rochemont and Culicover

1990, or of the strictly generative type like Rosengren 1993; whether they take
‘given’ as the basic semantic concept, like Schwarzschild 1999, or ‘new’, like the
above and, in fact, most linguists. There are, however, not a few linguists, like
Lambrecht 1994, or Vallduví and Engdahl 1996, who warn against a simple
identification of focus and new.

2 Compare Selkirk 1984, for a standard approach.
3 Abraham 1992 presents this idea in a unified generative account, extending it

onto non-verbal categories.

4 Bierwisch 1963 was the first to make this assumption.
5 An aside for the observant reader: the analogous order of the objects in the

examples above is not representative of the alternative directionality and has
made many theoretical linguists miss the basic difference in the direction of the
structural extensions which is so manifest in the more complex translational data.
If one compares predicates, obligatory and free adverbials, the parallel direction-
ality of the objects has to be seen as the exception – very much like other
irregularities found with frequently used elements. The particular role of the
complementary prepositional objects, pointed out by Collins 1995, is an addi-
tional argument in favour of the alternative-direction hypothesis.

6 See Lambrecht 1994, for a renowned expert on information structure, who

supports this idea, too.

7 Applying half a dozen syntactic and semantic tests, Frey and Pittner 1999 distin-

guish five classes of adverbials linearized alternatively in English and German.
For a corpus-linguistic study of English adverbials in Swedish translations on the
basis of Quirk et al.’s comprehensive subclassification from 1985, which illustrates
the right-directional trend of English adverbials extensively, cf. Lindquist 1989.

Notes

181

8 For a detailed account of the interaction between syntax and the lexicon, the

‘lexical projection’, see Bierwisch, for example, 1996.

3 Complex sentences

1 Except for some sporadic comments, as in Rosengren 1993, and some cases of

syntactic focus structures as for example, cleft-sentences, which we will turn to in
Chapter 6.

2 Chafe 1987 distinguishes between active, semi-active and inactive information,

which would allow us to classify role division as inactive, translation mechanism
as semi-active.

3 Büring, for example, 1998, speaks of a ‘partitive topic’ in a case like this, which

we may expect to be quite frequent considering the processing advantage of focus
spacing.

4 It may be interesting to note that the clausal version meets the Principle of End-

Weight, although in a wider sense than traditionally seen. Hawkins 1991–2
considers it to be one of the basic principles of linearization, but this is a
generalization which has to be relativized by a wide range of restructuring
patterns, like the one illustrated above.

4 In favour of primary relations

1 Cf. Altmann 1989, or Tanenhaus 1988, for concise surveys of psycholinguistic

research in the last decades.

2 In fact, the difference has been measured by neurophysiologists (as A. Friederici

1998: 180), who say that we may discover a grave syntactic error up to 200 msec
faster than a lexical-semantic error. Compare also Crocker 1996 for a
computational model on incremental processing.

3 See Frazier 1988, 1999, or Frazier and Clifton 1996, on the garden path theory.
4 We have already encountered the effect in topicalization and scrambling; in

general, this is a well-known phenomenon in focus theories; see for example
Stechow and Uhmann 1986.

5 For the impact of morphological case on processing, compare for example Bader

1998.

6 See Doherty 1996, which reports on a two-year research project on this topic.
7 Hawkins 1986 suggests that this is one of the many cases of a general difference

between English and German, where the nature of the relation between syntax
and semantics is more homomorphous in German than in English.

8 See Tanenhaus et al. 1989.
9 Halliday 1985 speaks of a grammatical metaphor in such cases.

5 Structural weight

1 For a systematic presentation of the various factors involved in the resolution of

anaphoric ambiguities, cf. for example Preuss et al. 1994; for a general theory of
referential movement in discourse Klein and Stutterheim 1991.

2 Fabricius-Hansen 1999 formulates this as a principle constraining the intro-

duction of new referents and new conditions in a text.

3 A major example in the German forerunner to the Key; cf. also Doherty 1999c,

for an English presentation.

4 Defining relative clauses are likely to be relatively autonomous information units,

which may promote contextual integration right away. Compare Marcus and
Hindle 1990, for the impact of sentence-internal boundaries on processing.

5 Asher 1993 presents an elaborate theory on the various types of abstract objects

182 Notes

constituted by anaphoric relations, which could provide the semantic-pragmatic
frame for the following observations. For first applications to problems of trans-
lation, cf. Fabricius-Hansen 1996, and Behrens 1998.

6 Kade 1968 referred to the latter as a case of heterovalente Übersetzung, con-

trasting it with equivalent translation.

6 Grammaticalized clues

1 For more detailed linguistic presentations of English clefts, see for example Prince

1978; Declerck 1988; Collins 1991; Delin 1995. A comparative study of English
clefts and their German translations is presented in Erdmann 1988, 1990.

2 For a suggestion explaining Carroll’s preference, see Doherty 2001a.
3 Rooth 1999 says that it is an existential presupposition which is associated with

clefts.

4 A very productive pattern with dummy head nouns are sentences like:

The main purpose behind the development of these techniques is to study
how the activity of genes is controlled.

which often correspond to simple sentences with marked topics in German:

Die verschiedenen Verfahren der Genübertragung wurden vor allem für die
Erforschung genetischer Steuerungsmechanismen entwickelt.

(cf. Doherty 2001b)

5 Cf. Doherty 1987, for a detailed semantic analysis of a handful of German

particles.

6 For a comparative study of focusing particles, cf. König, for example, 1991.
7 Cf. for example, the corpus-linguistic findings in Johansson and Oksefjell 1998.
8 Together with marked word order they belong to the linguistic clues which

psycholinguists like Flores d’Arcais 1988 expect to promote processing in a
language-specific way.

7 Shifting boundaries

1 For a generative approach to the linguistics of punctuation, see Nunberg 1990.

8 Relativizing optimality

1 As Grice 1975 demonstrated in his seminal paper on logic and conversation.
2 Some of the ‘individual’ cases may, however, belong to conceptual differences of a

more general nature, which can play a role in a wide variety of contexts. For an
impressive collection of such culture-specific perspectives, as for example the
different concepts associated with left and right in German and English, compare
Schmitt, for example, 1998.

3 In the sense of the ‘skopos’ theory by Reiß and Vermeer 1984.

Bibliography

183

Sources

Bernhard, T. (1989) Auslöschung. Ein Zerfall, Berlin: Volk und Welt.
—— (1995) Extinction, London: Penguin.
Bredekamp, H. (1993) Antikensehnsucht und Maschinenglauben, Berlin:

Wagenbach.

—— (1995) The Love of Antiquity and the Cult of the Machine, Princeton: Markus

Wiener.

Carroll, L. (1968) Alice in Wonderland, Through the Looking Glass, New York:

Penguin.

Cooreman, A. (1988) ‘Ergativity in Dyirbal’, Linguistics, 26(5): 730f.
Davies, D. (1989) ‘Traveller’s tales’, Far Eastern Economic Review, 28. September.
Eigen, M. and Winkler, R. (1983a) Das Spiel, München and Zürich: Piper.
—— (1983b) Laws of the Game, New York: Penguin.
Enzensberger, H.M. (1978) Der Untergang der Titanic, Frankfurt am Main:

Suhrkamp.

—— (1982) Politische Brosamen, Frankfurt am Main: Suhrkamp.
—— (1988) Mittelmaß und Wahn, Frankfurt am Main: Suhrkamp.
—— (1989) The Sinking of the Titanic, London, Sydney and Auckland: Paladni.
—— (1997) Zig Zag, New York: The New Press.
—— (1999) Zickzack, Frankfurt am Main: Suhrkamp.
Gransow, V. and Jarausch, K. (1991) Die deutsche Vereinigung, Cologne: Verlag

Wissenschaft und Politik.

Grass, G. (1963) Hundejahre, Neuwied am Rhein and Berlin: Luchterhand.
—— (1989) Dog Years, Orlando, FL: Harcourt Brace Jovanovich.
Jarausch, K. and Gransow, V. (1994) Uniting Germany, Providence, RI and

Oxford: Berghahn Books.

Hemingway, E. (1952) The Old Man and the Sea, New York: Scribners.
—— (1995) Der alte Mann und das Meer, Frankfurt am Main: Suhrkamp.
—— (1965) Zwei Ansichten, Frankfurt am Main: Suhrkamp.
—— (1967) Two Views, London: Jonathan Cape.
Kafka, F. (1970) ‘Ein Hungerkünstler’, in F. Kafka Sämtliche Erzählungen,

Frankfurt am Main: Fischer, 155–63.

—— (1988) ‘A hunger artist’, in F. Kafka The Metamorphosis, The Penal Colony

and Other Stories, New York: Schocken Books, 231–4.

184 Bibliography

Literature

Abraham, W. (1992) ‘Clausal focus versus discourse rhema in German: a program-

matic view’, Language and Cognition, II: 1–19.

Altmann, G.T.M. (1989) ‘Parsing and interpretation: an introduction’, Language

and Cognitive Processes, 4(3/4): 1–19.

Asher, N. (1993) Reference to Abstract Objects in Discourse, Dordrecht, Boston,

MA and London: Kluwer.

Bader, M. (1998) ‘Prosodic influences on reading syntactically ambiguous

sentences’, in J.D. Fodor and F. Ferreira (eds) Reanalysis in Sentence Processing,
Dordrecht, Boston, MA and London: Kluwer, 1–46.

Baker, M. (1992) In Other Words. A Coursebook on Translation, London and New

York: Routledge.

Behrens, B. (1998) Contrastive Discourse: An Interlingual Approach to the

Interpretation and Translation of Free ING-participial Adjuncts, Dr. Art
Dissertation, Department of Linguistics, University of Oslo, Norway.

Bierwisch, M. (1963) Grammatik des deutschen Verbs. Studia Grammatica II,

Berlin: Deutsche Akademie der Wissenschaften.

—— (1996) ‘Lexical information from a minimalist point of view’, Studia Gram-

matica, 40: 227–266.

Büring, D. (1998) ‘Topic’, in P. Bosch and R. van der Sandt (eds) Focus, Linguistic,

Cognitive, and Computational Perspectives, Cambridge: Cambridge University
Press, 142–65.

Carlson, G.N. and Tanenhaus, M.K. (1988) ‘Thematic roles and language

comprehension’, in W. Williams (ed.) Thematic Relations. Syntax and Semantics
21, San Diego, CA: Academic Press, 263–88.

Carston, R. (1988) ‘Language and cognition’, in F.J. Newmeyer (ed.) Linguistics:

The Cambridge Survey III, Cambridge: Cambridge University Press, 38–68.

Catford, J.C. (1965) A Linguistic Theory of Translation, London: Oxford

University Press.

Chafe, W. (1987) ‘Cognitive constraints on information flow’, in R.S. Tomlin (ed.)

Coherence and Grounding in Discourse, Amsterdam and Philadelphia, PA:
Benjamins, 21–51.

Collins, P.C. (1991) ‘Pseudocleft and cleft constructions: a thematic and infor-

mational interpretation’, Linguistics, 29(3): 481–520.

—— (1995) ‘The indirect object construction in English: an informational

approach’, Linguistics, 33(1): 35–49.

Crocker, M.W. (1996) Computational Psycholinguistics: An Interdisciplinary

Approach to the Study of Language, Dordrecht: Kluwer.

Declerck, R. (1988) Studies on Copular Sentences, Clefts and Pseudo-clefts, Leuven:

Leuven University Press.

Delin, J. (1995) ‘Presuppositions and shared knowledge in it-clefts’, Language and

Cognitive Processes, 10(2): 97–120.

Delisle, J. and Woodsworth, J. (eds) (1995) Translators Through History.

Benjamins Translation Library, Amsterdam and Philadelphia, PA: Benjamins.

Doherty, M. (1987) Epistemic Meaning. Language and Communication, Berlin:

Springer Verlag.

—— (1996) ‘Passive perspectives; different preferences in English and German – a

result of parametrized processing’, Linguistics, 34(3): 591–643.

Bibliography

185

—— (1997a) ‘“Acceptability” and language-specific preference in the distribution

of information’, Target, 9(1): 1–24.

—— (1997b) ‘Textual garden paths – parametrized obstacles to target language

adequate translations’, in C. Hauenschild and S. Heizmann (eds) Machine
Translation and Translation Theory, Berlin and New York: de Gruyter, 69–89.

—— (1998) ‘Clauses or phrases – a principled account of when-clauses in trans-

lations between English and German’, in S. Johansson and S. Oksefjell (eds)
Corpora and Cross-linguistic Research, Amsterdam and Atlanta, GA: Rodopi,
235–54.

—— (1999a) ‘Clefts in translations between English and German’, Target, 11(2):

289–315.

—— (1999b) ‘Position and explicitness – language-specific conditions for the use

of adverbial clauses in translations between German and English’, Studia
Grammatica, 47: 112–48.

—— (1999c) ‘Processing and directionality in German and English’, Languages in

Contrast, 1(1): 23–43.

—— (2001a) ‘Discourse functions and language-specific conditions for the use of

cleft(-like) sentences: a prelude’, Linguistics, 39(3): 457–62.

—— (2001b) ‘Cleft-like sentences’, Linguistics, 39(3): 607–38.
Erdmann, P. (1988) ‘On the Principle of “Weight” in English’, in C. Duncan-Rose

and T. Vennemann (eds) On Language: Rhetorica, Phonologica, Syntactica. A
Festschrift for Robert P. Stockwell From his Friends and Colleagues, London:
Routledge, 325–39.

—— (1990) ‘Fokuskonstruktionen im Deutschen und Englischen’, in C. Gnutzmann

(ed.) Kontrastive Linguistik. Forum Angewandte Linguistik 19, Frankfurt am
Main: Lang, 69–83.

Fabricius-Hansen, C. (1996) ‘Informational density: a problem for translation and

translation theory’, Linguistics, 34(3): 521–65.

—— (1999) ‘Information packaging and translation: aspects of translational

sentence splitting (German-English/Norwegian)’, Studia Grammatica, 47: 175–
214.

Flores d’Arcais, G. (1988) ‘Language Perception’, in F.J. Newmeyer (ed.)

Linguistics: The Cambridge Survey III, Cambridge: Cambridge University Press,
97–123.

Fodor, J.D. and Ferreira, F. (eds) (1998) Reanalysis in Sentence Processing,

Dordrecht, Boston, MA and London: Kluwer.

Frazier, L. (1988) ‘Grammar and language processing’, in F.J. Newmeyer (ed.),

Linguistics: The Cambridge Survey II, Cambridge: Cambridge University Press,
15–34.

—— (1999) On Sentence Interpretation, Dordrecht, Boston and London: Kluwer.
Frazier, L. and Clifton, C. (1996) Construal, Cambridge, MA: MIT Press.
Frey, W. and Pittner, K. (1999) ‘Adverbialpositionen im deutsch-englischen

Vergleich’, Studia Grammatica, 47: 14–40.

Friederici, A.M. (1998) ‘Diagnosis and reanalysis: two processing aspects the brain

may differentiate’, in J.D. Fodor and F. Ferreira (eds) Reanalysis in Sentence
Processing, Dordrecht, Boston and London: Kluwer, 177–200.

Grice, P. (1975) ‘Logic and conversation’, in P. Cole and J.L. Morgan (eds) Speech

Acts. Syntax and Semantics 3, New York: Academic Press, 41–58.

Gutt, E.-A. (1991) Translation and Relevance, Oxford: Blackwell.

186 Bibliography

Halliday, M.A.K. (1985) An Introduction to Functional Grammar, London:

Caulfield.

Hawkins, J.A. (1986) A Comparative Typology of English and German, London

and Sydney: Croom Helm.

—— (1991–2) ‘Syntactic weight versus information structure in word order

variation’, in J. Jacobs (ed.) Informationsstruktur und Grammatik. Linguistische
Berichte, Sonderheft 4, Opladen: Westdeutscher Verlag, 196–219.

House, J. (1997) Translation Quality Assessment. A Model Revisited, Tübingen:

Narr.

Jacobs, J. (1992) ‘Neutral stress and the position of heads’, in J. Jacobs (ed.)

Informationsstruktur und Grammatik. Linguistische Berichte, Sonderheft 4,
Opladen: Westdeutscher Verlag, 220–44.

Jakobson, R. (1959) ‘On linguistic aspects of translation’, in R.A. Brower (ed.) On

Translation, Cambridge, MA: Harvard University Press, 232–9.

Johansson, S. and Oksefjell, S. (eds) (1998) Corpora and Cross-linguistic Research,

Amsterdam: Rodopi.

Kade, O. (1968) Zufall und Gesetzmäßigkeit in der Übersetzung, Leipzig: Verlag

Enzyklopädie.

Kay, M., Gawron, J.M. and Norvig, P. (1994) Verbmobil: A Translation System for

Face-to-Face Dialog. CLSI Lecture Notes 33, Stanford, CA: Center for the Study
of Language and Information.

Klein, W. and Stutterheim, C. von (1991) ‘Text structure and referential movement’,

Sprache und Pragmatik, 22: 1–32.

Koller, W. (1992) Einführung in die Übersetzungswissenschaft, Wiesbaden: Quelle

and Meyer.

—— (1995) ‘The concept of equivalence and the object of translation studies’,

Target, 7(2): 191–222.

König, E. (1991) The Meaning of Focus Particles: A Comparative Perspective,

London and New York: Routledge.

Lambert, S. and Moser-Mercer, B. (eds) (1994) Bridging the Gap, Amsterdam:

Benjamins.

Lambrecht, K. (1994) Information Structure and Sentence Form. Cambridge

Studies in Linguistics 71, Cambridge: Cambridge University Press.

Levelt, W. (1989) Speaking: From Intention to Articulation, Cambridge, MA: MIT

Press.

Lindquist, H. (1989) English Adverbials in Translation. A Corpus Study of Swedish

Rendering, Lund: Lund University Press.

Macheiner, J. (1995) Übersetzen. Ein Vademecum, Frankfurt am Main: Eichborn.
Marcus, M. and Hindle, D. (1990) ‘Description theory and intonation boundaries’,

in G.T.M. Altmann (ed.) Cognitive Models of Speech Processing:
Psycholinguistic and Computational Perspectives, Cambridge, MA: MIT Press,
483–512.

Neubert, A. and Shreve, G. M. (1992) Translation as Text, Kent, OH and London:

Kent State University Press.

Nida, E.A. (1986) ‘Barriers and bridges in translating’, in K. Kummer (ed.)

Proceedings of the 27th Annual Conference of ATA, Medford, NJ: Learned
Information, 3–14.

Nunberg, G.D. (1990) The Linguistics of Punctuation. CSLI Lecture Notes 18,

Stanford, CA: Center for the Study of Language and Information.

Bibliography

187

Pinker, S. (1994) The Language Instinct, London: Penguin.
Preuss, S., Schmitz, B., Hauenschild, C. and Umbach, C. (1994) ‘Anaphora

resolution in machine translation’, in W. Ramm (ed.) Text and Context in
Machine Translation: Aspects of Discourse Representation and Discourse
Processing, Luxembourg: Office for Official Publications of the European
Communities, 29–52.

Prince, E.F. (1978) ‘A comparison of WH-clefts and it-clefts in discourse’,

Language, 54: 883–906.

Quirk, R., Greenbaum, S., Leech, G. and Svartik, J. (1985) A Comprehensive

Grammar of the English Language, London: Longman.

Reiß, K. and Vermeer, H.J. (1984) Grundlegung einer allgemeinen Translations-

theorie. Linguistische Arbeiten 147, Tübingen: Niemeyer.

Rochemont, M.S. and Culicover, P.W. (1990) English Focus Constructions and the

Theory of Grammar, Cambridge: Cambridge University Press.

Rooth, M. (1999) ‘Association with focus or association with presupposition?’, in

P. Bosch and R. van der Sandt (eds) Focus, Linguistic, Cognitive, and Computa-
tional Perspectives, Cambridge: Cambridge University Press, 232–46.

Rosengren, I. (1993) ‘Wahlfreiheit mit Konsequenzen – Scrambling, Topikal-

isierung und FHG im Dienste der Informationsstrukturierung’, in M. Reis (ed.)
Wortstellung und Informationsstruktur. Linguistische Arbeiten 306, Tübingen:
Niemeyer, 251–312.

Schmitt, P.A. (1998) ‘Wo ist eigentlich “links”? Beobachtungen zum Übersetzen

lokaldeiktischer Referenzmittel im Deutschen und Englischen’, Lebende
Sprachen, 1/98: 2–9.

Schwarzschild, R. (1999) ‘Givenness, Avoid F and other constraints on the place-

ment of accent’, Natural Language Semantics, 7: 141–71.

Searle, J.R. (1969) Speech Acts, Cambridge: Cambridge University Press.
Selkirk, E. (1984) Phonology and Syntax: The Relation between Sound and

Structure, Cambridge, MA: MIT Press.

Shreve, G.M. (2000) ‘Translation at the millennium: prospects for the evolution of a

profession’, in P.A. Schmitt (ed.) Paradigmenwechsel in der Translation,
Tübingen: Stauffenburg, 217–34.

Snell-Hornby, M. (1988) Translation Studies: An Integrated Approach, Amsterdam

and Philadelphia, PA: Benjamins.

Sperber, D. and Wilson, D. (1986) Relevance, Oxford: Blackwell.
Stechow, A. von and Uhmann, S. (1986) ‘Some remarks on focus projection’, in

W. Abraham and S. de Meij (eds) Topic, Focus, and Configurationality.
Papers from the 6th Groningen Grammar Talks, Amsterdam: Benjamins, 295–
320.

Stoll, K.-H. (2000) ‘Zukunftsperspektiven der Translation’, in P.A. Schmitt (ed.)

Paradigmenwechsel in der Translation, Tübingen: Stauffenburg, 235–64.

Tanenhaus, M.K. (1988) ‘Psycholinguistics: an overview’, in F.J. Newmeyer (ed.)

Linguistics: The Cambridge Survey III, Cambridge: Cambridge University Press,
1–37.

Tanenhaus, M.K., Carlson, G. and Trueswell, J.C. (1989) ‘The role of thematic

structures in interpretation and parsing’, Language and Cognitive Processes, 4
(3/4): 211–34.

Toury, G. (1983) ‘Sharing relevant features. An exercise in optimal translating’,

Meta, 28(2): 116–29.

188 Bibliography

—— (1995) Descriptive Translation Studies and Beyond, Amsterdam and

Philadelphia, PA: Benjamins.

Vallduví, E. and Engdahl, E. (1996) ‘The linguistic realization of information

packaging’, Linguistics, 34(3): 459–519.

Wunderlich, D. and Stechow, A. von (eds) (1991) Semantik. Handbücher zur

Sprach- und Kommunikationswissenschaft Bd. 6, Berlin and New York: de
Gruyter.

Index

189

acceptability 7, 26, 162, 165–6
adverbials 34, 47, 60, 66, 74, 78, 162,

166, 175

analogy 79, 101, 161, 166, 171
anaphors 82, 87, 147, 161, 166, 170–1
antecedent 52, 76, 82, 161, 166
appropriateness 16, 20–1, 160, 162,

166, 171, 173–4

argument 30, 32, 34–5, 166, 167, 168,

173, 175–6

attachment 63–4, 67, 161–2, 167,

174–5

attributes 39, 66, 167

background 16, 29, 97, 109, 120, 167,

169, 172

Carston, R. 10
case 24, 60, 162, 167–8, 169, 178
category 14–15, 80, 141, 168, 173–4,

178

Catford, J.C. 178
clefts 103–4, 106, 108–9, 116, 162,

168

comment 16, 25, 30, 168, 172, 179
compound 81, 87, 168, 169
configurationality 22, 24, 38, 60, 109,

161, 168, 172–3, 175–6

constituent 15, 37, 46, 63, 166, 168,

175

constituent hierarchy 168–9, 174–5
context 2–3, 15–16, 23–4, 28, 147,

161, 166, 169, 171, 177

contrast 37–8, 66, 108, 169, 170, 171
contrastive focus 38, 106, 109, 115,

163, 168–9

copula 103, 109, 169

decomposing 81, 88, 168–9

definiteness 29, 32, 56, 167, 169–70
defocusing 36, 43, 45, 50, 95, 169
Delisle, C. and Woodworth, J. 51
directionality 20, 22, 31, 38, 56, 92,

161, 170, 173

discourse 20, 23, 109, 116, 137, 147,

163, 170, 170–1, 178–9

discourse relations 97, 101–2, 111,

116, 119–20, 169–70

dummies 46, 85, 88–90, 109, 115, 162,

171

equivalence 9, 25–6, 48, 155, 171, 174,

177

eventuality 171, 176–7
explicitness 14, 16, 81, 100–1, 116,

160–1, 171–2, 177

focus 38, 45–6, 49–50, 95, 120, 161–2,

167, 169, 171–2

Fodor, J.D. and Ferreira, F. 176
Frazier, L. 63

gap 6, 81, 87, 133, 155, 172
garden path 64–5, 161–2, 172,

176

grammar 21, 38, 172
Grice, P. 16
Gutt, E.-A. 179

illocution 172, 176–8
implication 16, 26, 172, 177
implicature 16, 172, 177
information structure 23, 25–6, 31, 39,

48, 163, 170, 172, 179

informational density 147–8, 172
interpreting 2, 172, 179

Jakobson, R. 3

190 Index

Kade, O. 179
Kay, M. et al. 179
Koller, W. 171

Lambert, S. and Moser-Mercer, B. 172
language processing 17, 64, 161,

172–3

language-specific 18, 38, 61, 95, 115,

160, 173–4

Levelt, W. 173
linearization 34, 151, 173–4

modifiers 14, 34, 65–6, 74, 166–8, 173,

175–6

optimal relevance 10, 13, 22, 121, 136,

147, 166, 172, 173, 178

parameters 20, 22, 29, 31, 38, 115,

160–1, 168, 170, 173–4

paraphrase 6, 12–13, 15–16, 38, 174
parsing 63, 74, 76, 79, 161, 172,

174–5

perspective 57, 59–61, 74, 76, 160–1,

162, 174, 176

phrase marker 167–8, 175–6
Pinker, S. 2
pitch accent 171–2, 175–6
postponing 175
presuppositions 16, 108, 111, 168,

171, 175, 177

Preuss, S. et al. 181
primary relations 14, 74, 76–7, 175
pro-forms 81–2, 109, 175
proposition 15, 176, 177
prosody 36, 171, 176
psycholinguistics 17, 63–4, 161, 176,

179

Quirk, R. et al. 161

reanalysis 36, 43, 161, 174, 176
referent 82, 166, 175, 176–7
Reiß, K. and Vermeer, H.J. 179
relevance 10, 97, 116, 125, 171, 176,

178

role 60, 77, 167, 174–6

scope 34, 37, 109, 172–3, 176
scrambling 42, 46, 162, 173, 176
selection restrictions 18, 76–8, 175–6
semantics 15, 64–5, 120, 170, 173,

176–7

Shreve, G.M. 158
Snell-Hornby, M. 179
speech acts 172, 177–8
Sperber, D. and Wilson, D. 10
Stoll, K.-H. 158
style 6, 8–9, 18–19, 21, 81, 137, 144,

147, 171, 173, 178

syntactics 60, 63–4, 167–8, 172–4,

176, 178

text 28, 140, 163, 169–70, 178
textual relevance 2, 97, 99, 102, 105,

109, 116, 162, 178

topic 16, 25, 27, 69–70, 73, 91, 152,

163, 170, 172, 178–9

topicalization 56, 95, 105, 162,

179

Toury, G. 1, 158, 179
transformations 173, 179
translation studies 3, 158, 164–5, 173,

179

Wunderlich, D. and Stechow, A. von

177

Document Outline

Preliminaries
Contents
1 Setting the scene
2 Questions of order
3 Complex sentences
4 In favour of primary relations
5 Structural weight
6 Grammaticalized clues
7 Shifting boundaries
8 Relativizing optimality
9 Reviewing the scene
Glossary: technical terms seen through the Keyhole
Notes
Bibliography
Index

Wyszukiwarka

Podobne podstrony:
Language Processing in Discourse A Key to Felicitous Translation M Doherty (2002)
A key to the exercises contained in Adle
A key to the exercises contained in Adle
class 6 key to exercises in handout 3
class 6 key to exercise 2 in handout 1
Text, Context, Pretext Critical Isssues in Discourse Analysis Language in Society
Postlexical Integration Processes in Language Comprehension
Reichelt, Melinda English language writing instruction in Poland Adapting to the local EFL context
LH55 Niebling Symbols (Translation key to German Terms and Symbols for Kunststricken, Decken, Garnit
Erratum to Revised Transliminality scale
Proces pielęgnowania omówienie co to jest
German only, language rule in German schools
inżynieria bioprocesowa, IN[1].BIOPROCESOWA, Co to jest węglomol
Percepcja jest to proces-folia, Percepcja jest to proces selekcjonowania, wartościowania, organizowa
The Key to Women
Excellence is the?st key to fighting racism
Gee, J P Social Linguistics and Literacies Ideology in discourses

więcej podobnych podstron