Coherence

COHERENCE IN SPOKEN
AND WRITTEN DISCOURSE

Pragmatics & Beyond
New Series
Editor:
Andreas H. Jucker
(Justus Liebig University, Gessen)
Associate Editors:
Jacob L. Mey
(Odense University)
Herman Parret
(Belgian National Science Foundation, Universities of Louvain and Antwerp)
Jef Verschueren
(Belgian National Science Foundation, University of Antwerp)
Editorial Address:
Justus Liebig University Giessen, English Department
Otto-Behaghel-Strasse 10, D-35394 Giessen, Germany
e-mail: andreas.jucker@anglistik.uni-giessen.de
Editorial Board:
Shoshana Blum-Kulka {Hebrew University of Jerusalem)
Chris Butler (University College of Ripon and York)
Jean Caron {Université de Poitiers); Robyn Carston (University College London)
Bruce Fraser (Boston University); John Heritage (University of California at Los Angeles)
David Holdcroft {University of Leeds); Sachiko Ide {Japan Women's University)
Catherine Kerbrat-Orecchioni (University of Lyon 2)
Claudia de Lemos (University of Campinas, Brasil); Marina Sbisà {University of Trieste)
Emanuel Schegloff (University of California at Los Angeles)
Paul O. Takahara (Kobe City University of Foreign Studies)
Sandra Thompson (University of California at Santa Barbara)
Teun A. Van Dijk (University of Amsterdam); Richard Watts (University of Bern)
63
Wolfram Bublitz, Uta Lenk and Eija Ventola (eds)
Coherence in Spoken and Written Discourse.
How to create it and how to describe it
COHERENCE IN
SPOKEN AND
WRITTEN DISCOURSE
HOW TO CREATE IT AND
HOW TO DESCRIBE IT
SELECTED PAPERS FROM THE
INTERNATIONAL WORKSHOP ON COHERENCE,
AUGSBURG, 24-27 APRIL 1997
WOLFRAM BUBLITZ
Universität Augsburg
UTA LENK
EIJA VENTOLA
Martin-Luther Universität
JOHN BENJAMINS PUBLISHING COMPANY

AMSTERDAM/PHILADELPHIA
The paper used in this publication meets the minimum requirements of
American National Standard for Information Sciences — Permanence of
Paper for Printed Library Materials, ANSI Z39.48-1984.
Library of Congress Cataloging-in-Publication Data

International Workshop on Coherence (1997 : Augsburg, Germany) Coherence in spoken
and written discourse : how to create it and how to describe it : selected papers from the
International Workshop on Coherence, Augsburg, 24-27 April 1997 / edited by Wolfram
Bublitz, Uta Lenk, Eija Ventola.
p. cm. -- (Pragmatics & beyond, ISSN 0922-842X ; new ser. 63)
Includes bibliographical references and indexes.
1. Cohesion (Linguistics) Congresses. 2. Discourse analysis Congresses. I. Bublitz,
Wolfram. IL Lenk, Uta. III. Ventola, Eija. IV. Title. V. Series.
P302.2.I58 1999
401'.41--dc21 99-28183
ISBN 90 272 5077 4 (Eur.) / 1 55619 941 4 (US) (alk. paper) CIP
© 1999 - John Benjamins B.V.
No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any
other means, without written permission from the publisher.
John Benjamins Publishing Co. • P.O.Box 75577 • 1070 AN Amsterdam • The Netherlands
John Benjamins North America • P.O.Box 27519 • Philadelphia PA 19118-0519 • USA
In memoriam
Maria Elisabeth Conte
(1935-1997).
Table of Contents
Acknowledgements ix
About the authors xi
Introduction: Views of Coherence 1

Wolfram Bublitz
Part I: How to (Re-)Create Coherence: Means of Coherence
Coherent Voicing: On Prosody in Conversational Reported Speech 11

Elizabeth Couper-Kuhlen
It Takes Two to Cohere: The Collaborative Dimension 35

of Topical Coherence in Conversation
Ronald Geluykens
Learning to Cohere: Causal Links in Native vs. 55

Non-Native Argumentative Writing
Gunter Lorenz
Coherence through Understanding through Discourse Patterns: 77

Focus on News Reports
Jan-Ola Östman
Semiotic Spanning at Conferences: Cohesion and Coherence 101

in and across Conference Papers and their Discussions
Eija Ventola
Coherent Keying in Conversational Humour: 125

Contextualising Joint Fictionalisation
Helga Kotthoff
Viii
Part II: How to Negotiate Coherence: Degrees of Coherence

Disturbed Coherence: 'Fill me in' 153
Wolfram Bublitz and Uta Lenk
Coherence and Misunderstanding in Everyday Conversations 175

Carla Bazzanella and Rossana Damiano
The Effect of Context in the Definition and Negotiation of Coherence 189

Anna Ciliberti
Coherence in Summary: The Contexts of Appropriate Discourse 205

Barbara Seidlhofer and Henry Widdowson
Coherence in Hypertext 221

Gerd Fritz
Part III: How to Describe Coherence: Views of Coherence

Communicative Intentions and Coherence Relations 235
Ted Sanders and Wilbert Spooren
If Coherence Is Achieved, Then Where Doth Meaning Lie? 251

Willis J. Edmondson
A Bibliography of Coherence and Cohesion 267

Uta Lenk, Sarah Gietl and Wolfram Bublitz
Index 297
Acknowledgements
A volume such as this stands or falls by the papers it presents. We would
therefore like to thank the authors of the articles here collected, not only for the
hard work they put in for the completion of this volume, but also for their lively
and interesting contributions to the discussions during the workshop. We also
wish to thank those colleagues who participated in the workshop but whose
papers for various reasons do not appear in this reader: Andrea Gerbig, Gisela
Redeker, Sorin Stati, Edda Weigand, and the late Maria-Elisabeth Conte
(University of Pavia), whose premature death a few months after our workshop
prevented her from writing up her contribution on Semantic integration devices
in texts, which had received much acclaim and positive comments at
presentation.
Our special thanks go to Carla Bazzanella, who was involved in developing the
first plan for this workshop, and to Gudrun Nelle, Sarah Gietl and Nils Engel,
the always reliable and eager groundcrew who were the efficient organizers
behind the scenes. Without them, the workshop would not have materialized.
In editing this reader we have received valuable help from Carol LeRoux, who
checked the non-native speakers' English, Gudrun Nelle, a reliable proof-reader,
Sarah Gietl, our desk-top publisher-in-chief, and, in the final stages, Thomas
Henrichs, our computer wizard.
Last, but not least, we want to thank the Gesellschaft der Freunde der
Universität Augsburg and the Deutsche Forschungsgemeinschaft for their
financial support which made the workshop possible, a gathering which was
experienced by all who came as most delightful and interesting.
About the authors
Carla Bazzanella is Professor of the Philosophy of Language at the University
of Torino, Italy. She has published on pragmatics, morphosyntax, socio-
linguistics, and applied linguistics. Her most recent monograph is Le Facce del
Parlare (1994) and she has edited Repetition in Dialogue (1996).
Wolfram Bublitz is Professor of English Linguistics at the University of

Augsburg, Germany. Semantics, pragmatics and discourse analysis are his major
research areas, in which he has published on modal particles, negation, modality,
listener responses, discourse topic and repetition. His most recent publications
are on collocation and coherence and include "The rather mystical notion of
coherence" (1994), "Semantic prosody and cohesive company" (1996) and
"Copying semantic features in collocations with up-scaling intensifies" (1998).
Anna Ciliberti is Professor of Applied Linguistics at the University for

Foreigners of Perugia, Italy. Her current research interests are professional
discourse, language ideology and the pedagogy of cross-cultural
communication. She has published papers in a number of Italian and
international journals such as Lingua e Stile, Journal of Pragmatics. Her most
recent books are Grammatica, Pedagogia, Discorso (1991) and Manuale di
Glottodidatica (1995).
Elizabeth Couper-Kuhlen is Professor of English Linguistics at the University

of Konstanz, Germany. Her research interests are prosody and grammar in
verbal interaction, reported speech, clause combining, tense and aspect. Her
major publications include An Introduction to English Prosody (1986), English
Speech Rhythm: Form and Function in Everyday Verbal Interaction (1993),
Prosody in Conversation. Interactional Studies (co-edited with Margret
Setting) (1996), and Language in Time: The Rhythm and Tempo of Spoken
Interaction (co-authored with Peter Auer and Frank Mueller) (1999).
Xii
Rossana Damiano is currently a Ph.D. student in Computer Science at the

University of Torino, Italy. She graduated in Modern Literature (Technics of
Communication) from the University of Torino with a thesis on
"Misunderstanding in man-machine interaction". Her interests are in linguistics,
computational linguistics, man-machine interaction and natural language
processing.
Willis Edmondson is Professor of Foreign Language Learning Research at the

University of Hamburg, Germany. His work on coherence began with his book
Spoken Discourse (1981). More recent publications include Einführung in die
Sprachlehrforschung (1993) and Twelve Lectures on Second Language
Acquisition (1998).
Gerd Fritz is Professor of German Linguistics at the University of Gießen,

Germany. His main areas of research are the foundations of discourse analysis,
comprehensibility of texts and historical semantics. Book-length publications
include Kohärenz. Grundfragen der linguistischen Kommunikationsanalyse
(1982) and Kommunikation und Grammatik (1984, with M. Muckenhaupt). He
is co-editor of the Handbuch der Dialoganalyse (1994) and a volume on
Historical Dialogue Analysis (to appear, with A. Jucker).
Ronald Geluykens is Professor of English Linguistics at the University of

Münster, Germany. His major research interests are all in the area of discourse
analysis and pragmatics, and include: conversational discourse, the discourse-
grammar interface, prosody in spoken discourse, and second language
pragmatics. Apart from about 40 research papers on these topics, he has
published two books: From Discourse Process to Grammatical Construction
(1992) and The Pragmatics of Discourse Anaphora in English (1994). He is
currently working on an introductory textbook on discourse pragmatics.
Helga Kotthoff is full-time researcher in Linguistics at the University of

Konstanz, Germany. Her main research areas are discourse analysis, pragmatics,
intercultural communication, linguistic gender studies, and anthropological
linguistics. Her recent publications include Communicating Gender in Context
(ed. with R. Wodak, 1997), and Spass Verstehen. Zur Pragmatik von
konversationellem Humor (1998).
Xiii
Uta Lenk teaches Linguistics in the English Department of the University of

Augsburg, Germany. Her interests include pragmatics and discourse analysis,
sociolinguistics and language change. She has published several articles and a
book on coherence and discourse markers in conversation, Marking Discourse
Coherence: The Role of Discourse Markers in Spoken English (1998); she has
also written on the role of language in Utopian fiction.
Gunter Lorenz teaches Linguistics and Foreign Language Methodology in the

English Department of the University of Augsburg, Germany. He has worked in
the fields of intercultural communication, foreign language acquisition and
learner corpus linguistics. He is the author of Adjective Intensification -
Learners versus Native Speakers: A Corpus Study of Argumentative Writing
(1999).
Jan-Ola Östman is Professor of Linguistics at the English Department and at

the Department of General Linguistics at the University of Helsinki, Finland. His
main interests are in pragmatics (pragmatic particles; ideology; contact
phenomena), text and discourse analysis, minority languages and cultures
(dialects, American Indian languages), and syntax (Construction Grammar). He
is co-editor of the Handbook of Pragmatics and the author of You Know: A
Discourse-Functional Approach (1981) and Pragmatics as Implicitness (1986).
Ted Sanders is Assistant Professor at the Utrecht Institute of Linguistics and

the Department of Dutch at the University of Utrecht, The Netherlands. His
main research interests are the coherence and structure of discourse, and
readers' and writers' cognitive discourse representation in text processing. His
publications include several articles in this field, among others on coherence
relations (with Wilbert Spooren and Leo Noordman in Discourse Processes and
Cognitive Linguistics), on coherence relations and connectives (with Alistair
Knott in Journal of Pragmatics), on hierarchical text structure (with Carel van
Wijk in Text) and on the role of hierarchical text structure in on-line text
production (with Joost Schilperoord).
Barbara Seidlhofer is Associate Professor at the English Department of the

University of Wien, Austria. Her main area of research is the application of
discourse analysis, pragmatics, phonetics, and sociolinguistics to second
language education. Her publications include Pronunciation (with C. Dalton,
XiV
1994) and Approaches to Summarization. Discourse Analysis and Language

Education (1995).
Wilbert Spooren is Assistant Professor at the Discourse Studies Group,

Department of Language and Literature at the University of Tilburg, The
Netherlands. His main research interests are issues of text structure and text
coherence, and their relevance for the efficacy of texts. He has published articles
on psychological implications of coherence relations (together with Ted Sanders
and Leo Noordman, in Discourse Processes and Cognitive Linguistics) and on
epistemic modification and perspective (together with José Sanders, in
Cognitive Linguistics).
Eija Ventola is Professor of English Linguistics at the Martin-Luther-

University, Halle-Wittenberg, Germany. She has published extensively on
discourse analysis, academic discourse and scientific writing considering texts
from functional and intercultural perspectives, and on Australian literature and
its reception and on its translation. Her current interests include the language of
conferencing and 'internetting', and the language of business and tourism. Her
major publications include The Structure of Social Interaction (1987),
Australian Prose in Finnish: A Bibliography 1890-1989 and its Interpretation
(1992; in Finnish); Writing Scientific English: A Textlinguistic Approach to its
Teaching (with A. Mauranen, 1992; in Finnish); Researchers and Writing in
English (with A. Mauranen, 1990; in Finnish). In addition, she has edited
several books on academic writing, functional and systemic linguistics, and
literary analysis.
Henry Widdowson holds professorial appointments in Language Education

and Applied Linguistics at the University of London and the University of
Essex, England. His work on coherence began with his research in Edinburgh in
1969-72, and has been developed through a number of publications, including
Explorations in Applied Linguistics I & II (1979/1984), Learning Purpose and
Language Use (1984), and Aspects of Language Teaching (1990). His most
recent work is a critical appraisal of current ideas on text and discourse analysis.
Introduction: Views of Coherence
Wolfram Bublitz
Our aim in putting together this collection of papers is to help relieve

coherence of its mystifying aura. Coherence is a concept which in its
complexity is still not fully understood and a matter of continuing debate.
While after the publication of Halliday and Hasan's book Cohesion in English
in 1976 the notion of cohesion was widely welcomed and accepted as a well-
defined and useful category for the analysis of text beyond the sentence,
coherence was regarded or even dismissed as a vague, fuzzy and "rather
mystical notion" (Sinclair 1991: 102) with little practical value for the analyst.
This view was held by large parts of the linguistic community with only some
notable exceptions (prominent among them Fritz 1982). However, the past two
decades have seen a considerable shift in orientation and, in particular, a
fundamental rethinking of the concept of coherence. The amazing number of
well over four hundred books and papers listed in the general bibliography,
which concludes this volume, bears witness to this intriguing development and
the rapidly changing scene in coherence research over the last twenty years.
Evidently, coherence has found its place as a key concept, perhaps even the
key concept, in discourse and text analysis.
And yet, when preparing this reader we have been very conscious of the
fact that linguists have still to agree both on a generally accepted definition of
coherence as well as on an adequate theory of coherence. (It is, therefore, not
surprising that so far there is no widely used introductory textbook on
coherence, which could also serve as both a state of the art review and an
impetus for further research.) But while it is true that, in the past, very different
descriptive conceptions have dominated much of the research on coherence
(for a brief overview cf. e.g. Hellman 1995), it is also apparent that a growing
number of new studies share various theoretical and methodological
assumptions. We are witnessing a trend in the field of discourse and text
analysis which is moving away from reducing coherence to a product of
(formally represented) cohesion and/or (semantically established) connectivity.
Instead, there is an attempt to reach a more user- and context-oriented
interpretive understanding which is more interactively negotiated and is less
2 Wolfram Bublitz
dependent on the language of the text itself. This view of coherence underlies
an increasing number of recent analyses of authentic data which relate micro-
linguistic to macro-linguistic issues and in which cohesion often plays merely a
minor role.
The tendency to observe a basic stock of fundamental descriptive
assumptions while refraining from accepting one canonic definition of
coherence is also reflected in this reader, which consists solely of papers
presented at the 'Workshop on Coherence in Discourse', held at the University
of Augsburg in April 1997. In the invitation to the workshop, we spelt out our
essentially hermeneutic understanding of coherence, which is outlined in the
following snapshot account.
In accordance with a fairly long tradition on the European continent,
though not in Anglo-American work, we use coherence as a context-dependent,
hearer- (or reader-) oriented and comprehension-based, interpretive notion.
Coherence is intrinsically indeterminate because it is relative to the way in
which language users ascribe their understanding to what-they-hear (or what-
they-read). Accordingly, coherence is not a text-inherent property at all (as are
cohesion and connectivity). It is not given in the text invariantly and
independently of an interpretation, but rather 'comes out' of the text in the
sense that it is based on the language of the text, in the same way as it is based
on additional information provided inter alia by the linguistic context, the
socio-cultural environment, the valid communicative principles and maxims
and the interpreter's encyclopedic knowledge. Since it is not texts but rather
people that cohere when understanding texts (be it at the producing, receiving
or, indeed, analysing side), it can be said that for one and the same text there
exist a speaker's (or writer's), a hearer's (or reader's) and an analyst's
coherence, which may or may not match. However, normally speakers (or
writers) are set to help create coherence by (more or less subtly) guiding their
hearers (or readers) to a suggested line of understanding which comes close to
or, ideally, even matches their own. Conversely, hearers (or readers) use these
guiding signals as instructions to re-create coherence and to re-align their
interpretation with what they take to be the speakers' (or writers') intentions.
To put it differently, they assemble and subsequently test a view of coherence
which they assume comes closest to that of the speaker (or writer). Hence,
coherence is not a state but a process, helped along by a host of interacting
factors situated on all levels of communication (from prosodic variation to
textual organisation, from topic progression to knowledge alignment). As a
process, coherence is not taken for granted but, depending on situation, genre
or text type, rather viewed as being more or less tentative and temporary,
continually in need of being checked against new information which may make
adaption and updating necessary. Eventually, coherence can, especially in
retrospect, lose some of its processural and temporary character and acquire a
higher degree of permanence. Some of the following papers address a further
aspect of coherence: that it is also a cooperative achievement because it
depends on both the speaker's (or writer's) and the hearer's (or reader's)
willingness to negotiate coherence (in the same way as they negotiate
ideational meaning and illocutionary force). Mutual understanding not only
rests on the participants sharing the same socio-cultural background, the same
range of knowledge and communicative assumptions, but also on their ability
to figure out unshared experience, i.e. to adjust their own world-view to that of
their interlocutor's. Adopting Reddy's (1979) toolmakers metaphor we can say
that hearers (or readers) are constantly engaged in trying to re-create coherence
as a 'replica' of the speaker's (or writer's) coherence, but despite their efforts
they can never succeed in coming up with an exact replica. Thus, coherence is
only approximate and a matter of degree and best described as a scalar notion.
Any interpretation of the coherence of a text is restricted and, accordingly,
partial in different degrees. Nevertheless, partial coherence rarely turns into
incoherence because, as a rule, participants operate on a generally shared
default assumption of coherence (cf. Bublitz and Lenk, this volume).
To somewhat differing degrees, the concept of coherence proposed
above is shared by the contributors to this collection. Furthermore, lying
behind most essays are several other shared assumptions: that studies on
coherence must be based on authentic, non-fabricated data (using corpora of
spoken or written discourse); that when describing coherence, it is imperative
to sharply distinguish between the producer's, the recipient's and the analysing
observer's point of view; that coherence is medium- and genre-specific, i.e.
that the strategies and means used by speakers or writers to suggest coherence
can (and often do) vary from spoken to written language, from genre to genre,
from text-type to text-type. All papers are original and unpublished works
reflecting contemporary, state of the art trends in coherence research. They
critically re-examine coherence from a variety of different angles, propose new
ways of thinking about it and point out methodological and descriptive
desiderata and caveats that an adequate theory of coherence should eventually
account for.
Included in this volume are 13 of the 18 papers read at the workshop by
contributors whose work over the years has proved to be a major stimulus for
coherence research, among them many of the leading figures in discourse
analysis and pragmatics. The contributions have been deliberately organised
into sections in order to create topical coherence for the reader.
The six papers of the first section {How to (re-)create coherence: means
of coherence) contribute answers to the question of how exactly various
linguistic strategies and means (among them inter alia prosodic variation, topic
4 Wolfram Bublitz
management, meaning and other kinds of semiotic relations and discourse

pattern) can be instrumental in establishing coherence both from the producing
and the receiving point of view. Elizabeth Couper-Kuhlen describes how in
natural conversation interlocutors contextualise shifts of footing, which
involves different reception or production roles and is occasioned by reported
speech when explicit non-prosodic markers are missing. She shows that in
order to overcome such absences the reporting speaker regularly voices the
interlocutor's reported message by employing various prosodic and
paralinguistic cues as highly effective signals. Combining the analytic methods
of conversation analysis and prosodic analysis, Couper-Kuhlen demonstrates
how both the interactants and the observing analysts rely on these prosodie
cues to come to a coherent understanding of the multi-voiced sequences.
Ronald Geluykens, who demands a more empirically grounded approach to
coherence and consequently adopts conversation analysis as a methodological
framework, establishes as a powerful means of coherence the organisation of
discourse topic, which is collaboratively negotiated by the participants. More
specifically, he focuses on question-answer adjacency pairs, which regulate the
introduction or elicitation of new topics. Gunter Lorenz chooses data from
written language to illustrate the coherence establishing function of semantic
relations. He is concerned with the construction of coherence in learners'
versus native speakers' written argument. Particularly, he seeks to explore the
various ways in which relationships of CAUSE and EFFECT are expressed.
Starting from the premise that the differences between native and non-native
reasoning strategies will at least partly be reflected in distinctive causal
patterns, his study uses a learner/native speaker corpus of written English to
determine, quantify and explain the respective stylistic preferences. Jan-Ola
Östman modifies the traditional approach to coherence by stressing that we
arrive at coherence through socio-cognitive understanding which involves
different kinds of pre-existing and easily recognisable patterns. He argues that
speakers normally have a conceptualisation of a text that relies primarily on
what he calls discourse patterns and only secondarily on genre and text type.
Within a speech community, language users share linguistic, textual as well as
non-linguistic, visual expectations about a text which need to be met in order to
make, for instance, a recipe, a contact ad or a death notice a coherent and thus
'proper' text. Discourse patterns can differ within a culture and, of course,
regularly do differ between cultures. In an exemplary analysis of news texts he
demonstrates the usefulness of his concept for coherence analysis in particular,
and discourse analysis in general. Eija Ventola's analysis also involves pre
existing and easily recognisable patterns, though of a different kind. Examining
conference data, she proposes a new term, semiotic spanning, to capture the
various kinds of semiotic relations that exist in such data. She claims that the
traditional concepts of cohesion and coherence are bound to individual texts

and to the immediate contexts of production and interpretation. The notions of
intertextuality and genre, although they do consider text relations, generally
prove to be too limited. Semiotic spanning involves a more extensive view of
how texts link up with other texts, visuals etc. semiotically. The process of
spanning is dynamic in nature and creates a new semiosis not only with the
existing and experienced texts, but also with future texts, be they of the same or
a different generic form. Helga Kotthoff looks at humorous episodes in
everyday conversations, whose coherence appears to be governed by principles
that are not found elsewhere in ordinary discourse. Typically, humorous
utterances display not only unexpected content but also unexpected structure,
patterning, voicing, etc. Specifically, Kotthoff picks out keying and the
collective construction of fictionalisations by several participants as special
dimensions of oral humour. Fictionalisations, fantasies etc, which are
collectively developed by relying on various framing procedures, are often not
worked out in detail and do not have to be. Nonetheless, interlocutors easily
manage to do the necessary coherence work by jumping from scene to scene,
matching each against a set of stored cognitive patterns.
The five papers of the second section {How to negotiate coherence:
degrees of coherence) address the scalar nature of coherence, concentrating on
several different aspects. The first three contributions focus on various kinds of
disturbed coherence or, indeed, incoherence and discuss their causes,
characteristics and effects. Mainly using data from authentic conversations,
Wolfram Bublitz and Uta Lenk describe disturbed coherence as a common
phenomenon of communication. Disturbed coherence must be kept distinct
from incoherent speech of persons whose command of language is impaired,
on the one hand, and should not be treated as simply the opposite of coherence,
on the other. Though supporting the view that coherence is a scalar notion and
that, accordingly, the hearer's understanding of coherence is always only
approximate or partial, they argue that partial coherence is not necessarily
disturbed coherence. It turns into disturbed coherence only when the extent to
which the text is only partly understood is no longer tolerated by the hearer.
They distinguish between different types of disturbed coherence, deliberately
disturbed and accidentally disturbed coherence, and give a detailed account of
some of their sources. Unnegotiated topic drifts and topic changes, unclear
reference, frame breaks and register breaks may prevent the hearer, who
normally operates on a default assumption of coherence, from reaching a
coherent interpretation. The ensuing breakdown in the flow of conversation
regularly prompts a process of renegotiation before the intended line of
conversation is resumed by the speaker. Carla Bazzanella and Rossana
Damiano, who take their data from several corpora of spoken discourse, look
6 Wolfram Bublitz
at one frequent cause of disturbed coherence: conversational mis

understandings. They can manifest themselves in two distinct ways, either by a
mismatch between speaker-intention and hearer-interpretation or by a non-
awareness of the (acoustic) misunderstanding. Both types of misunderstanding
eventually lead to a point in the ongoing conversations where coherence is no
longer successfully negotiated and consequently collapses. Detection of such
non-alignment of interpretation ultimately leads to different types of repair
behaviour aiming at a reestablishment of shared coherence. The authors, who
also subscribe to the interpretive view of coherence as sketched above, point
out that misunderstanding which leads to coherence disruption, is participant
dependent and, as such, not a clear cut notion but open to negotiation. Using
data collected from oral examinations at an Italian university, Anna Ciliberti
also touches upon degrees of coherence. She discusses asymmetrical relations
between the examiner and the examinee and the influence these relations have
on the creation of coherence in their discourse. She distinguishes two major
types of oral examination. In the first type, the examiner makes a diagnosis of
the knowledge displayed by the student. This type is characterised by a
frequent change of topics (not acceptable in everyday conversations) and by
the use of disconnected question-answer pairs (which may appear to be linked
when the student uses cohesive devices). In the second type, the discussion is
carried out on a more equal basis. Questions can be asked by either side and
topics are connected and build up a coherent line of argumentation. In general,
unlike their examinees, examiners are allowed to be 'incoherent', shifting and
changing topics throughout this specific type of discourse. Barbara Seidlhofer
and Henry Widdowson, who emphasise that coherence can only be
adequately defined by making a clear distinction between text and discourse,
ask in which way language pedagogy can draw upon findings from applied
linguistics. On the grounds that coherence is best described as a discourse
function achieved by both the speaker (or writer) and the hearer (or reader),
they also see the possibility of interpretive mismatches, which support the
variability of coherence interpretation. They illustrate their view of coherence,
and to reach implications for teaching, the authors examine students' elicited
reactions to a summarisation task as evidence of different degrees of
coherence. Gerd Fritz discusses a very special way of 'negotiating' coherence
in the very young text genre hypertext. At first sight, the traditional concept of
coherence does not apply because hypertext is non-linear text. However, the
user follows a path by selecting from a set of given textual elements which will
then turn into a linear meaningful string. But despite their linearity, hypertext
sequences require special modes of coherence establishment because reliance
on explicit indicators and standard text sequences is reduced. Users are thus
required to permanently perform the difficult task of monitoring where they are
in the hierarchical structure of the text. Construing coherence across the

sequential elements of the chosen hypertext path prompts a number of
problems which the author discusses in his paper (among them the role of
opening nodes, the multifunctionality of text segments, the use of explicit
markers of function).
The last two papers in the section titled How to describe coherence:
views of coherence tackle methodological aspects of coherence description in
different ways. Ted Sanders and Wilbert Spooren juxtapose the linguists5
view of coherence as a relational concept (relying on relations such as those
between cause and effect) and the cognitivists' view of coherence as a
realisation of participant intentions manifest in each section of a discourse.
While agreeing that both views are necessary for an adequate theory of
discourse coherence, the authors suggest that the coherence relations be
regarded as realisations of communicative intentions. Accordingly, there would
then be no need to propose a separate discourse level for participants'
intentions. Willis Edmondson introduces a rather more general topic into this
reader. He is concerned with questions of validation in discourse analysis. Is
the analyst's perception of coherence of a transcription or a videorecording
independent of or the same as the participants' achieved coherence? And what
validates the analyst's view of coherence? Are meanings to be found in
discourse analyst's representations or are they elsewhere? He points out that
when doing discourse and coherence analyses, we need an explicit theory of
discourse which enables us to show how discourse varies in regards to
structural, interactional and pragmatic realisation types. Analysts should not be
satisfied with a relativistic analytical position where 'anything goes' and all
interpretations are equally valid. Neither should they accept the resultative
stance where the analysis is merely descriptive and for which it is the results of
the analyses that validate the analytical procedure.
Sinclair, J. 1991. Corpus, Concordance, Collocation. Oxford: Oxford University

Press.
Reddy, M. 1979. "The conduit metaphor: a case of frame conflict in our language
about language". In A. Ortony (ed), Metaphor and Thought. Cambridge:
Cambridge University Press, 284-324.
Augsburg August 1998

Part I
How to (Re-)Create Coherence: Means of Coherence
Coherent Voicing: On Prosody in Conversational
Reported Speech
Elizabeth Couper-Kuhlen
Universität Konstanz
Coherence should be findable for everything that is a demonstrably relevant aspect

of the talk for the parties, or there should be evidence of trouble or of its
suppression. (Schegloff 1990)
Goffman has pointed out that interlocutors in the course of any natural
conversation are constantly changing the footing of their talk. In Goffman's
usage, this term refers to the alignment which speakers take up to themselves
and to others as evidenced by the way they handle the production and reception
of utterances (1981: 128). Changes in footing may involve different reception
roles or different production roles or both (Goffman 1981: 226ff; also Levinson
1988), and they are commonly understood to be signalled inter alia by prosodic
cues and code-switching, which contextualise the particular footing or
participant framework currently relevant (Gumperz 1982; Tannen (ed) 1993).
Yet precisely how this contextualisation is accomplished and what specific
contribution prosody makes to the 'management' of footing has not yet been
fully spelled out1 - at least not for all types of shift.
The present paper addresses one of the most frequent shifts of footing,
namely that occasioned by the use of reported speech in conversation. What
happens with reported speech is that the unity within a single speaker of the
three production roles which Goffman identifies - animator, author and principal
- dissolves, leaving the role of animator separate from, and independent of,
those of author and/or principal (cf. also Seidlhofer and Widdowson, this
volume). The 'reporting' speaker animates or voices a 'reported' figure without
necessarily composing the words which this figure is made to utter or espousing
the beliefs which the figure's words will be heard as attesting to. 2 The question
which the 'voicing' of figures raises for a prosodist is whether and to what
extent the speaker's phonatory voice is instrumental in the process.3 Using a
methodology developed by crossing prosodic analysis with conversation
analysis (Couper-Kuhlen and Selting 1996), this paper attempts to pin down
12 Elizabeth Couper-Kuhlen
exactly which tasks the 'voicing' of reported speech confronts

conversationalists with and how speakers' prosodic and paralinguistic voice
resources contribute to the accomplishment of these tasks.
1 Coherence as a conversationalist's practice and an analyst's

object
ScheglofF(1990: 55) has suggested that in conversation the issue of coherence
can be subsumed under the general question Why that now? In other words,
participants in interaction are constantly trying to make sense of talk as
recipient-designed, situated action. When they are unable to infer plausible
answers to the question Why that now?, they have sets of methods which allow
them to remedy the situation, one of these sets involving the initiation and
execution of 'repair'. Remedial procedures help clarify the misunderstood or the
misunderstandable, on occasion they make explicit the unexplicit (see also
Schegloff 1996). But remedial procedures also provide analysts with an
invaluable instrument of analysis. It is via conversationalists' pursuit of
coherence that analysts can learn more about the object from an insider
perspective.
Observations such as these on coherence in interaction suggest a way to
approach the relation between prosody and reported speech. Coherence in
reported speech sequences,4 it can be argued, will be manifestly lacking where
participants in interaction find repair to be necessary. When 'troubles' in
coherence can be plausibly reconstructed as involving some prosodic or
paralinguistic factor, insight will be gained into the specific nature of prosody's
contribution to reported speech. A subsequent comparison of repaired and
repairable reported speech sequences with non-repaired and non-repairable ones
will suggest some of the methods which participants employ for the prosodie
animation of voices,
2 Three types of 'trouble' in reported speech sequences

That the use of reported speech places special demands upon conversationalists
is evident from the 'trouble' it sometimes occasions for interaction. One type of
trouble appears to stem from unclarity as to whether or not a speaker is
reporting speech in the first place. These are cases in which there is no clear
answer to the question Is this current speaker's 'voice ' or someone else's? The
reason why there might be doubt about this is because - contrary to what
grammar books propagate about reported speech - speakers in conversational
Coherent Voicing 13
interaction do not always explicitly introduce different 'voices' with reporting

verbs or quotative constructions. Instead figures are often 'brought on stage'
for the first time merely by being animated, without, for instance, a prefatory he
said or she said. In order for this device to succeed, however, the figure's
'voice' must be reconstructibly different from the current speaker's own 'voice'.
The following sequence demonstrates that participants are aware of this
distinction and that prosody may be a means for marking it:5
(1)
Toxic chocolate (38/22.16)
Ann and a girlfriend are spending the weekend as guests at Joy's house. The
following exchange takes place over lunch.
(The small f signals that a reportedfigureis being voiced somewhere within the
line, different indices representing different figures. The hyphen signals a return to
the reporting situation.)
1 J: oh and on Sunday,
you can open the first door of the advent calendar.
All: aahh oohh
J: «p>\ never had one before.>
5 A: did you buy it?
J: mhm,
A: you didn't buy the kind with chocolates in it?
J: no.
All: huh huh huh huh
10 J: f1 I didn't think I needed any « / > chocolate (thing).>
fl it's a « / > to:xin you know.>
A: - who're you talking to (.)
talking about.
J: there-
15 f1 « / > chocolate is toxic.>
-> A: - yeah but you said that like somebody says that.
J: no that's the (.) cancer pre
cancer pre- [ventative type.
A: [aaah.
20 J: I honestly cannot fathom; (.)
following that diet;
just to prevent cancer.
Joy presents opening the first door of the advent calendar as a special
treat for her guests (lines 1-2). In fact, even the calendar itself takes on special
status when Joy claims that it is her first (line 4). Ann expands on this topic by
asking whether Joy bought the calendar. (Buying an advent calendar is to be
understood as contrasted with making one, a custom common in Germany,
where the conversation takes place.) When Joy acknowledges that she did buy
it, Ann expands the sequence again by asking you didn't buy the kind with
chocolates in it (line 7), subtly implying that this kind of calendar might be more
desirable. Thus when Joy now states I didn't think I needed any chocolate
(thing) and it' 's a toxin you know, she is heard as justifying her purchase, an
activity prompted by Ann's treatment of buying a calendar without chocolates
as remarkable and therefore 'accountable'.
Joy's accounts are couched as a warning implying that chocolate is a
health risk (chocolate (thing). it's a toxin you know, lines 10-11). This warning,
however, is presented in someone else's voice, as Ann herself observes: you
said that like somebody says that (line 16). Yet Joy's 'voicing' in lines l0ff
manifestly creates a problem for Ann, because she initiates repair immediately
thereafter: who're you talking to () talking about (lines 12-13). Note that it is
not the content of the utterance which is at issue here. It is true that Joy initially
treats the problem for which repair is initiated as referential in nature. She
responds with a partial repeat, making the reference of it explicit: chocolate is
toxic (line 15). However, Ann's yeah but in next turn (line 16) shows that what
was said is not the point. The issue is the way the utterance was said. As Ann
puts it, the utterance 'sounded like' someone else. What is her interpretation
based on?
Deconstructing Ann's observation leads to a consideration of the nature
of vocal deixis. Prosodic and paralinguistic effects are in fact deictic to a certain
extent: they involve speaking within a given range of relative loudness, pitch
and tempo (Laver 1994) and with a given voice quality (Laver 1990). In the
default case, loudness, pitch and tempo ranges together with voice quality are
anchored to the prosodic/paralinguistic habitus of the speaker. That is, speakers
are accustomed to deploy, and their interlocutors are accustomed to expect,
certain prosodic and paralinguistic 'reference values'. A noticeable shift of these
values - using a pitch, loudness or tempo range or a voice quality which departs
from the speaker's habitus - will be heard as shifted deixis and can evoke the
presence of a second deictic centre. It is precisely this kind of shift which is
hearable in Joy's chocolate (thing) (line 10), toxin you know (line 11) and
chocolate is toxic (line 15): in each phrase she drops into low pitch register
shading off into a final 'vocal fry' or glottal creak. Fragment (1) thus provides
demonstrable evidence of the fact that a figure can be 'voiced' by the way in
which an utterance is configured prosodically and paralinguistically. At the same
time it suggests that the question underlying coherence must be expanded to
why that now and in that way?
Coherent Voicing 15
Given the deictic nature of prosodic and paralinguistic phenomena,

'troubles' in reported speech sequences are to be expected when prosodic and
paralinguistic cues are ambiguous as to whose vocal deictic centre they index.
This is arguably what creates the necessity for repair initiation in the following
sequence:
(2)
Political contradiction
Bill is telling his girlfriend Gina about a 'real political ' discussion which he had
on the beach with someone he had just met
1 B: You know and it's so funny cause he's-
he's a Catholic?
()
and::
you know
like I nailed him on the contradiction;
he's like pro-capital punishment,
but- and pro life,
hhh
10 G: uh huh,
B: fl I said try to explain that to me.
fl I don't underst(h)a(h)nd(h).
fl heh heh
G: - (wait) you said he's:
15 pro:: capital punishment -
B: and pro life,
G: and pro life.
ri:::ght.
B: that's like
20 to me that's like a hu::ge contradiction?
(.)
and he said
?f2 wellhe
he- justified it as (.)
see
25 ?f2 th- those little inf-
?f2 the little (.) uhm
?f2 embryos?
0
?f2 they don't have a (.) decis-
30 ?f2 they can't make a decision.
?f2 you know
?f2 ('n) they're gonna die.

- this is what he was saying.
yeah
and he said that you know
f2 people who (.) commit crimes:
?- you know
f2 that (.) call(s) for the death penalty;
f2 they brought it upon themselves.
- oh wait
just
can you hang on just a sec?
yeas
In recounting how he 'nailed' his interlocutor on the political

contradiction between supporting capital punishment but being against abortion,
Bill re-enacts the incident as a dialogic exchange, animating himself as a figure:
try to explain that to me. I don't underst(h)a(h)nd(h). heh heh (lines 11-13).
What his interlocutor's response to this request is, however, is not immediately
clear. A response is projected with the reporting verb and he said (line 22) and
then with another quotative construction hehe justified it as (line 23). Both
constructions announce immediately upcoming talk as reported speech.6 But
Bill's next utterances have no prosodic or paralinguistic shift of voice.
Moreover, they deploy the discourse markers well see (line 24) and a high rising
intonation pattern following embryos? (line 27), which characteristically elicits
an in situ response from one's interlocutor.7 Thus, one interpretation this talk
can be given is that it is part of the reporting situation, addressed to Gina as
background information or as an aside. But this interpretation stands in direct
conflict with the original framing of upcoming talk as reported speech (lines 22-
23).
It is arguably the conflicting signals of quotative introduction but lack of
vocal deictic shift (or any other marker clearly framing the talk as
reconstructed) which prompt Gina's repair initiation this is what he was saying
(line 32). The fact that he is stressed here suggests that it is to be interpreted as
standing in contrast with other possible members of a set, in this case with you.
Thus the format of Gina's repair initiation supports the interpretation that her
problem is knowing whether this is some background commentary by Bill the
narrator (part of the reporting situation) or his animation of a figure in the story
(part of the reported situation).
'Trouble' in reported speech sequences occurs not only when there is no
straightforward answer to Is this current speaker's 'voice ' or someone else's?.
Coherent Voicing 17
It may also occur if the answer to the question Whose 'other voice ' is this? is
unclear, as we remember from example (1).
Joy's it's a toxin you know (line 11) is followed in the next turn by a
repair initiator from Ann: who're you talking to (line 12). Following a short
pause, talking to is self-repaired to talking about (line 13). Ann's problem thus
is now displayed as being that she does not know who Joy is impersonating
(although she does know that Joy is impersonating someone). It is only once
Joy has specified what figure she has in mind that's the cancer pre- cancer
preventative type (lines 17-18) that Ann acknowledges the repair, her aaah in
line 21 attesting to a changed state of knowledge (Heritage 1984).
A third type of 'trouble' occurs in reported speech sequences when the
answer to the question How is this 'other voice ' being done?, or What is the
speaker doing with this 'other voice '? , is unclear. Figures are always animated
for a particular purpose in situated interaction. It is often the way the voices are
formatted prosodically and paralinguistically which contextualises what they are
doing, or rather what current speaker is doing with them. Where this is unclear,
participants may find it necessary to initiate repair. Please, consider example (2)
once again, where, in re-enacting the exchange he has had with his new
acquaintance, Bill first introduces himself as a figure in the story and animates
this 'voice': try to explain that to me. I don't underst(h)a(h)nd(h) heh heh (lines
12-14). The animation is noticeable first by shifts in personal and temporal
deixis: Bill's interlocutor becomes you, to whom the request try to explain that
to me is directed, and the speaker's account for this request I don't understand
is anchored to the moment of speaking in the reported situation. But there is
also a sign of shifted vocal deixis in the breathiness superimposed on the
figure's account I don't underst(h)a(h)nd(h) (line 12).8 The laughter particles in
line 13 are compatible with such a shift.
The parahnguistic formatting of Bill-the-figure's voice is interpretable as
Bill-the-narrator 'doing' something. But Gina manifestly has a problem in
determining what exactly he is doing with Bill-the-figure's voice. Her problem
becomes clear when she initiates repair in next turn: (wait) you said he's: (line
14). Her initiation is accompanied by a candidate repair pro:: capital
punishment (line 15), which Bill acknowledges by completing with and pro life
(line 16). Gina ratifies this completion and pro life (line 17), signalling with a
follow-up ri:::ght (line 18) that she has got the point. Moreover, Bill now
makes explicit in the next turn what his point is: that's like - to me that's like a
hu::ge contradiction (lines 19-20). It is thus an understanding of the
contradiction between the two political positions referred to which is treated as
being necessary in order for Gina to make sense of Bill-the-figure's turn. Once it
is clear that there is a contradiction and what it is, the parahnguistic overlay of
breathiness and laughter fits in as a contextualisation cue to how Bill-the-figure
evaluates the contradiction (as well as the person who espouses it) and to how
Bill-the-narrator wishes his addressee to evaluate it.
To summarise the argument so far: by examining reported speech
sequences in which there is 'trouble', as evidenced by participants finding it
necessary to initiate repair, we have shown that the pursuit of coherence in
conversational reported speech involves finding plausible answers to at least
three questions: Is this current speaker's 'voice ' or someone else's?; If so,
whose 'other voice' is this? and How is this 'other voice' being done? (or
What is the speaker doing with this 'other voice '?). 'Trouble' may occur when
there are no clear answers or only conflicting answers to these questions.
Moreover, finding appropriate answers sometimes depends crucially on
prosodic and paralinguistic framing devices. Where prosodic and paralinguistic
signals are inappropriate or ambiguous (and verbal content does not fully
disambiguate), participants may find that repair is required in order to establish
coherence in reported speech sequences.
The question which now arises is what counts as appropriate prosodie and
paralinguistic framing for conversational reported speech. To address this
question we shall examine a selection of'successful' reported speech sequences
and compare them to the repairable ones above.
3 Prosodie and paralinguistic framing in 'successful' reported

speech sequences
'Successful' reported speech sequences are recognised not only negatively by
the absence of repair. There are often other clues to the fact that participants are
making sense of ongoing talk. For instance, participants may show through the
recipient design of their uptake that they are orienting to a bit of talk as the
reported speech of a figure rather than as current speaker's own words. One
special case of this involves the phenomenon of 'chiming in', when recipients
participate in the voicing of a particular figure. On other occasions
conversationalists may produce metalinguistic talk about the 'other voice'
and/or the way it is being done. These phenomena can be thought of as positive
signs of coherence in conversational reported speech sequences. They will all be
documented in one way or another in the examples we shall now examine of
successful ways of handling the three questions of reported speech.
Coherent Voicing 19
3.1 Vocalframing of an 'other voice '
If 'trouble' in reported speech sequences involves problematic answers to the

question Is this current speaker's 'voice ' or someone else's?, then 'successful'
sequences should display the use of clear and unambiguous signals that some
figure is being animated. Where prosody and paralinguistics are crucially
involved, this means using features which are not likely to be taken as indexing
current speaker's own voice. Use of a marked voice quality, intonation or
rhythmic pattern which departs from the local, momentary 'norm' of talk will
thus often suffice to cue some 'other voice'. Such a departure from local
prosodic norms is demonstrated, for instance, in the following episode:
(3)
Rainbows (12/739)
Two high school friends Janet and Ann, reunited after several years, are recalling
their college experiences. Janet's husband Steve is present
1 J: we had a professor from Carleton;

who was an atheist.
A: uh huh ((off-stage))
J: and there was this one guy in class.
5 and I can't remember what his name was;
but he insisted upon::
the Bible being thee truth;
and thee word of Go- [d.
A: [.of course.
10 J: and argued everything.
A: of course.
J: fl </There /were /no /rain=/bo:ws.> <rhythmic>
fl before the flood.
fl [because
-> A: - [<<h> WHAT?>
16 S: ha ha ha
-> J: fl </GOD /MADE THE /RAINBOW!> <rhythmic>
J: - and he's like
f2 <I'm sorry:::> <stylised>
20 - you know
f2 <if you really look at s:ci:entific evide:nce;>
<stepping down>
f2 You will see that
f2 <whenever these things happen:;> <stepping down>
f2 a rainbow occu:rs.
25 f2 and as long as those things happened

f2 <<l> before the flood
f2 there were rainbows.>
and this (.)
kid just could not (.)
30 accept [it.
A: [see you have to say kid too now
[hhh( )
J: [yeah he was.
he was really a kid.
35 (.)
and so- yeah
it would be like
Every lecture
he would bring up another: (.)
40 Bible story; (.)
and (.)
defend it because (.)
-> f1 « / > /that's what it /said in the /Bible.> <rhythmic>
()
45 - it was very frustrating.
A: wo::w.
[(you were)
J: [Amusing; (.)
but frustrating.
Janet brings two figures on stage here, the 'professor from Carleton' (line
1) and a 'guy in class' (line 4). In line 12, with no further quotative introduction,
she animates the figure of the student. Moreover, her interlocutor Ann orients
to the presence of this 'other voice' with her high WHAT? in line 15.9 Although
the token what functions as a next-turn repair initiator with respect to prior talk
in some contexts, Janet does not treat it this way here. She does not respond by
recycling her turn but carries on instead with (because) God made the rainbow:
In fact, Ann's what token is cued with loud volume and high pitch - a display of
astonishment (Selting 1996), in this case strongly projective of disagreement.
Such strong disagreement would hardly be appropriate, were Janet's talk to be
understood as being 'in her own voice'. But it is fully appropriate if there were
no rainbows before the flood is animated as belonging to someone else.
How is it that Ann recognises lines 12ff as being 'in another voice'? In
this case there is no significant change in the voice quality, pitch or loudness of
Janet's speech. What does distinguish these lines from surrounding talk,
however, is their pronounced rhythmic quality. The accented syllables are timed
Coherent Voicing 21
so as to come at approximately equal intervals in time, i.e. they are perceptually

isochronous (Couper-Kuhlen 1993). Moreover, in line 12, every word is
'scanned' with a separate accent (the compound rainbows being rendered
prosodically as two independent words rain and bows).10 It is thus the rhythmic
declamation of these lines which contributes to marking them as different from
prior talk and serves as an iconic cue to the shift in footing.
Yet the rhythm of lines 12ff is not only a cue to the presence of different
'voice'; in its declamatory style it is also indexical of some specific 'other
voice', a 'voice' which could be described in this context as school-boy-like.
(See Janet's characterisation of him as a kid in lines 29 and 34.) In fact, the
student's voice is consistently given this rhythmic configuration in Janet's story.
Not only is GOD MADE THE RAINBOW (line 17) also rhythmically marked,
but later in the story, when the student reappears, his voice is once again
rhythmically declamatory: that's what it said in the Bible (line 43).11 The choice
of prosodic device for cueing the voice of a figure is thus a motivated one. In
conversation analytic terms, it is part of the recipient design of a speaker's talk,
talk which is geared to accomplish specific goals in specific situations. In
reported speech sequences these goals often include presenting figures with a
particular 'take', a point we shall return to shortly.
Acknowledging the fact that prosodic voicing in reported speech
sequences is recipient-designed allows us (as analysts) to appreciate the
significance of'chiming in'. Conversationalists on occasion display orientation
to an 'other voice' by participating in the reporter's animation:
(4)
Galileo (12/370)
Same speaker constellation as in (3). Talk here is about the Catholic church and
the fact that its doctrines change very slowly
1 J: They just decided what.

like last year,
that uhm (.) tsk
who was it now.
5 « / > I'm forgetting the name of the guy.>
S: Oh yeah
Copernicus?
J: C-
no not Copernicus;
10 [( )
S: [GALILEO.
J: Galileo.
A: .haah
J: fl that [<<ƒ> /tGALI/LEO:: <scanded,

15 A: - [yes
J: f1 [/WAS: stepping down>
A: - [yes
J: f1 (.) /RIGHT!>
THEY JUST DE-[CIDED THAT.
20 A: [uhhhehheh
J: f(h)ive hu(h)ndred years later!
A: That's right.
that's right,
that was uh-
25 J: f1 <that /↑ma:ybe he /ha:d a /good i/dea.>
<rhythmic, stepping down >
S: - that's the speed at which they uhm (.)
A: hnn.
S: yeah.
J: fl <they /tprobably /shouldn't have /tortured him
30 fl as /much.> <rhythmic, stepping down >
All: - HA HA HA
-> A: next year they'll decide;
-> f1 <↑probably the ro- > <stepping down >
-> f1 the world tis round.
35 - heh heh
J: - Right.
right.
Janet's anecdote is occasioned by Steve's prior remark even the Vatican

changes its policy - over the millennia. As part of the anecdote she animates the
voice of the Catholic church in a recent proclamation concerning Galileo.12 This
voice is given a distinctive prosodic configuration: the statement Galileo was
right (lines 14, 16 and 18), for instance, is configured as one intonation phrase
with high pitch on its first accented syllable and all subsequent accents
descending in a stylised fashion throughout the phrase. Every possible lexical
stress (including secondary stresses and monosyllabic function words) is given
an accent and these accents are timed so as to occur regularly in time. The same
distinctive isochronous pattern, with exactly the same high onset and stylised
accents which step down, recurs on maybe he had a good idea (line 25) and
they probably shouldn 't have tortured him as much (lines 29-30).
But this pattern is not only distinctive for analysts. Janet's interlocutor
Ann joins in with the animation of the Vatican's voice in lines 32ff and thus
displays her orientation to it. And she attempts the same prosodic configuration:
Coherent Voicing 23
her accent on probably (line 33) hits exactly the pitch level that Janet's prior
animated onsets had and the next accent on ro- is targeted to step down from
there. The pitch replication now breaks down with what is manifestly a slip of
the tongue: ro(und) instead of world (cf. line 33) but the rhythmic replication
persists with the accumulation of accents in the world is round (line 34). With
her prosodic 'chiming in', Ann thus implicitly signals her understanding of the
presence of some 'other voice' and indeed of some particular 'other voice'.
Moreover, she can also be heard to co-align with the 'take' which current
speaker has on this voice (see below).
3.2 Vocal framing of several 'other voices '
When more than one figure is 'on stage' in a reported speech sequence, the task
which conversationalists face is all the more complex. If the figures' voices are
prosodically and paralinguistically animated, they must be done in such a way
that they are not only recognisably distinct from the reporter's own voice but
also distinct from one another. Reconsider fragment (3), which demonstrates
one way conversationalists have of accomplishing this task.
Following her animation of the student in lines 12, 14 and 16, Janet begins
to animate the professor in line 19 with I 'm sorry. Significantly, the fact that this
line is the professor's voice and not the student's is inferable only from its
distinctly different prosodic delivery. (The anaphoric expression in Janet's
quotative and he's like (line 18) could refer to either the student or the
professor.) Instead of the rhythmic and loud configuration which characterises
the student's there were no rainbows (line 12) and God made the rainbow (line
17), line 19 - I'm sorry - has lower volume, stylised pitch and syllable
stretching. It is because of this markedly different prosody that we infer that it is
not the student who is 'apologising' but some other figure.
Just as with the prosodie framing of a single figure, the voicing of two or
more figures also displays individualised recipient design. That is, different
figures receive not only distinct voicing but also specific different voicing.
Consider now another case in which this is successfully accomplished:
(5)
Rollerblades (12a/40)
Bret and Wanda are brother and sister. Ricky is Wanda's young son. The topic
here is rollerb lading.
1 B: well I mean
Here you see
you know five-year-old kids doing -
W: well Ricky
5 some of Ricky's friends
and they do wonderful -
but I mean
°gosh°
They're dangerous too
10 and kids don't real-
His one friend
uhm
gosh he-
I said
->15 fl </wha:t /↑h:appened. < slow>
fl did a /tr:u:ck hit him?
?fl <the kid ('s got) his faxe, <listing>
?fl his who:le l:e:g,»
-> f2 «h+p> ↑well I was rollerblading;> <lax, breathy
20 - <and I'm going> <whispered>
fl <wooh!> <whispered>
A: - <yeah <whispered>
I mean>
B: [WELL ACTUALLY YOU SHOULD HAVE-
25 A: [you can't stop!
B: YOU SHOULD HAVE ELBOW PADS
KNEE PADS [AND A
W: [ « h > oh yeah!>
B: AND A HELMET ON.
30 A: you can't stop!
Wanda's anecdote involves three figures, two of which are animated:

herself and the kid, a friend of her son Ricky. The first animation begins in line
15, where Wanda the figure asks what happened, the type of question which -
as an opener and accompanied by the appropriate prosody - tends to be
occasioned by some remarkable feature in the addressee's external appearance.
As the anaphora in line 16 makes clear, however, this question is not addressed
to the kid himself but to a third party, who is thereby presented as
knowledgeable about the kid and his affairs; this is presumably Ricky, the kid's
friend. The next line did a truck hit him (line 16) proffers a candidate answer to
Wanda the figure's own question, couched itself as a question presumably to the
same third party. Both of these lines are animated with distinct prosody: the
accented syllables are given a slow, rhythmic delivery, with marked lengthening
of syllables and initial consonants. The pitch movements have a relatively wide
range. This distinct delivery pattern is continued in the next two lines: the kid('s
Coherent Voicing 25
got) his face, his whole leg, with noticeable lengthenings and slow tempo.
These lines (17-18) are thus interpretable on one reading as Wanda the figure
making explicit to her interlocutor what the visual signs were which occasioned
the concern she expressed in lines 15-16.13
In line 19 a new figure enters.14 Despite the lack of any reportative
introduction, the presence of a different figure is clearly marked by a shift in
prosody: the line well I was rollerblading is configured with overall high pitch
and low volume. The voice quality is breathy and the articulation lax. It is the
prosodic contrast between this line and prior talk which cues the new figure.15
Yet not only the contrast between this line and prior talk is noteworthy,
but also the way the contrast is constructed. The contrastive figure being
enacted is a particular kid, with particular characteristics, ones which the
recipient-designed prosody of his voice cues: rather than the slow, weighty
delivery of lines 15-18 - cued as belonging to Wanda the figure - line 19 comes
across as light and airy. The kid is thus not only distinguished from Wanda, his
voice is also designed to suggest an opposing stance. Both sets of prosodic
features can be heard as cueing stances which are hinted at elsewhere in talk:
Wanda the figure's 'weighty' prosody becomes an index of they're dangerous
too (line 9), while the kid's 'light and airy' prosody indexes kids don't real(ise)
(line 10).
Notice now that the next two lines enact Wanda the figure's reaction to
the kid's response and its stance. Her wooh (line 21) is delivered in a whispered
voice, which 'leaks' into the prior reporting construction and I'm going (line
20). The whispered quality of Wanda the figure's turn here is significant in two
ways. First, it forces us to refine the statement made earlier that figures' voices
are animated consistently in conversation. If this were true here, Wanda the
figure should say w:ooh: and use normal volume just as she does in lines 15-18.
Instead, the whispery prosody of this line is designed to cue Wanda the figure's
reaction to the kid's response, and this stance is not - for want of a better label -
'serious concern' but something closer to 'speechless (or voiceless) amazement'
(cf. her gosh in lines 8 and 13, the first time also done with whispery prosody).
Thus, to have used the slow, rhythmic prosody of lines 15-18 would have cued
the wrong message. Yet although Wanda's voice is not consistently done, she is
constructed as a consistent figure, i.e. someone who is concerned by young
rollerbladers' accidents and amazed by their carefree attitude.16
3.3 Vocal characterisation of 'other voices '
Figures, we have stressed, are voiced as specific 'other voices': not only have
their 'lines' been constructed for them to 'say', their prosody hints at stances
which they are constructed as assuming - and/or which their constructor, the
current speaker, wishes to take towards them. The speaker, in other words, is
not only 'doing' voices but also doing something with those voices which it is
incumbent upon recipients to infer. Every detail of the way figures in reported
speech sequences are constructed is thus inspectable for some clue as to what
the speaker is doing with these 'voices'. In particular, the way a given 'voice' is
configured prosodically and paralinguistically will serve as a hint to the 'take'
the speaker has on that figure, suggesting how it, or the talk of which it is part,
is to be received and evaluated by interlocutors.
How do interlocutors make sense of the prosodic and paralinguistic cues
of reported speech? Or better: how do we as analysts know what sense they
have made of them? What signs of participant 'coherence-making' are visible in
the interaction itself? One type of sign is verbal in nature: recipients may engage
in meta-talk which displays how they have interpreted the prosodic cueing of a
particular figure:
(6)
Breathing in smoke (12/106)
Same speaker constellation as in (3). The topic of conversation here is the ban on
smoking in public buildings which has been instated since Ann left the country to
study abroad.
1 A: it's funny (.)

cause I haven't- (.)
I mean I- (.)
I remember being kinda for it?
5 when- (.)
when I left
it was just kinda like (.)
fl ↑Actually you know - (.)
fl ↑I don't like to smoke all this- er
10 fl « / > /breathe in all this /smoke all the /time,>
Now I come back
and I'm just kinda like (..)
f2 «p> /they should be able to /sm(h)oke,>
f2 I mean /what are you guys DOING! heh
15 S: yeah=
J: =right
-> it's a little overboard.
-> A: yeah
Coherent Voicing 27
Ann animates two figures here, both of which are herself at different
moments in time. The 'early' Ann figure is made to say lines which are hearable
as supportive of the anti-smoking ban actually you know I don't like to (..)
breathe in all this smoke all the time (lines 8-10), whereas the 'later' Ann figure
claims that those who are prevented from smoking should be able to do so (line
13), thereby implying that she no longer supports the ban. Both 'voices' are set
off prosodically from Ann's framing talk {it was just kinda like, line 7; and I'm
just kinda like, line 12) by pauses, pitch shifts and volume shifts. But the way
the 'later' Ann figure is voiced hints at something more: smoke in line 13 can be
heard as having the beginning of an overlaid 'snort' and the line what are you
guys doing, delivered with raised larynx, is followed by a laughter particle.
These delivery features are signs that Ann the narrator is not just contrasting an
earlier state of mind with a later state of mind on smoking bans but that she is
also evaluating these positions, aligning herself with the latter as opposed to the
former.17 Ann's interlocutors respond in ways which are hearable as aligning
with this critical stance. In fact Janet's next turn actually formulates verbally
what Ann was alluding to with her 'snort' and laugh particle: it's a little
overboard (line 17).18 Moreover, Ann ratifies this understanding with her
follow-up yeah (line 18). Thus, Janet's turn amounts to 'putting into so many
words' the effect of the work which prior speaker's vocal animation was
designed to do. It is metapragmatic discourse (Silverstein 1993) par excellence.
The Galileo fragment (4) also provides an example of metapragmatic
reference to the work which prosody is doing in a reported speech sequence.
In line 26 Steve responds to Janet's enactment of the Vatican
proclamation with that's the speed at which they uhm. On one level this
utterance is hearable as an explicit reference to - or formulation of - the point
which Janet is making with they just decided what, like last year (line 1f), they
just decided that (line 19) and five hundred years later (line 21). On another
level, however, Steve's remark can be heard as putting into so many words
what Janet's (slow and laborious) prosody is cueing in her animation of the
Vatican's voice. On this reading, it 'verbalises' the coherence which Steve is
attributing to the way the reported speech is done.
On other occasions, rather than verbalising the effect of the vocal framing
of a figure or figures, recipients will instead show their understanding of what
the speaker is doing by making responses tailored specifically to the reported
speech sequence. The Rollerblades fragment (5) also provides an example of
this.
Recall that Wanda the figure is animated with 'weighty' prosody, indexing
a stance which Wanda the narrator has articulated verbally with but I mean gosh
they 're dangerous too (lines 7-9), while the kid is animated with a breathy, light
voice, indexical of the stance implicit in Wanda the narrator's kids don't real-
(ize) in line 10. In a sense then Wanda the narrator has 'primed' the vocal
characterisations of her figures and in doing so, set up two models for co-
alignment, one roughly associated with the danger motif, the other with not
realising the danger. Therefore, it is not wholly coincidental that her
interlocutors orient to these models in subsequent talk. Ann picks up the danger
motif by specifying why rollerblades are dangerous - you can't stop (line 2),
while Bret addresses the kid's non-realisation by detailing what protective
equipment rollerbladers should wear: actually you should have elbow pads,
knee pads and a helmet on (lines 24, 26f and 29). 19 Thus here participants show
through responses specifically designed with respect to the perceived stances
what understanding they have of the way the figures' voices are being done. In
this case, the inferencing job is made easier by the verbal hints which current
speaker has provided in prior talk.
Yet what if a current speaker's 'priming' talk is ambiguous? A final
fragment demonstrates that this can and does happen in conversational reported
speech sequences:
(7)
Rented cars with phones (12a/200)
Same speaker constellation as (5). Reference has just been made to cellular
phones.
1 B: Oh I got a kick outa- (.)

well you had a rented car;
and Ricky was so worried;
cause you were gone somewhere.
5 (it's just)
fl «l+p> what if she has a flat ↑tire. >
fl «cresc> she /hasn't /got a /↑phone;
fl in her /tcar!> <rhythmic>
-> W: heh heh heh heh (.) heh
10 B: fl <CAN YOU IMAGINE /RENTING A /CAR
fl THAT /DOESN'T HAVE A /PHONE?>
<rhythmic>
-> All: ha ha
W: I didn't know that they rented cars with phones.
15 although I'm sure they do;
-> isn't that fun(ny)?
B: he was so worried cuz-
W: «h> when was this;>
B: when you were up here,
Coherent Voicing 29
20 sometime,
W: «p> hn>
last time,
W: «p> isn't that funny?>
OHHH
25 at- at Christmastime.
B: yeah,
W: right after Christmas.
yeah
B: he was so:: worried;
30 fl that his /mom was out in the /car,
fl and she had /rented one;
fl <that didn't have a /phone;> <rhythmic,
stylised melody>
C: - ha ha ha
B: fl <I hope she's all right; <rhythmic,
35 fl she can't /call us; same stylised
fl she /hasn't got a /phone;> melody>
W: heh heh
(he) takes after his father.
I don't worry about stuff like that.
40 they do though.
Bret's priming for the figure of Ricky, whom he animates in line 6ff,
becomes apparent for the first time in line 3: Ricky was so worried. This verbal
characterisation is repeated in line 17: he was so worried cuz- and once again in
line 29: he was so:: worried. Yet the way Bret 'does' Ricky's voice is not
indexical of worry throughout the animation. Line 6, what if she has a flat tire,
is configured with somewhat softer volume and lower pitch than prior talk, a
style of delivery which is indeed suggestive of worry. Yet in lines 7-8, although
Bret continues to animate Ricky's voice, the prosody changes gradually to
increasingly louder volume and marked rhythm; a high point is reached in lines
10-12, where the volume is very loud and the rhythm heavily marked. This
prosodic configuration suggests something more than worry, and indeed
recipients do not respond with, say, co-aligning expressions of concern and
sympathy but rather with giggling (line 9) and outright laughter (line 12).
Wanda even verbalises her understanding of Bret's voicing twice with isn 7 that
funny? (lines 16 and 23). Bret now animates Ricky once again, this time
employing, in addition to marked rhythm, a stylised melody ending with a call
contour (Ladd 1978) on each of the animated lines. This animation too is
receipted with laughter from recipients in lines 33 and 37. Thus, Bret has
verbally cued Ricky as being worried, while his voicing of Ricky is suggestive of
something which provokes giggling and laughter in recipients.20
This example is telling because it reminds us that recipients do not rely
blindly on verbal 'priming' by the speaker in trying to make sense of the
prosodic and paralinguistic cueing of an animated figure in a reported speech
sequence. Instead, in cases of non-congruence, the verbal hints must be weighed
up against the prosodic hints and a more or less warranted 'guess' must be
hazarded as to what the speaker is doing. This is of course risky business, but
example (7) shows that it need not lead to a break-down of coherence.
4 Conclusion
In conclusion, the above discussion has shown that not only when explicit cues
to reported speech are lacking but also when explicit cues about reported
speech are misleading, participants make sense of conversational reported
speech in part by relying on the prosodie and paralinguistic details of 'voicing'.
As Schegloff reminds us, "talk is laced through and through with inexplicitness
and indexicality" and this inexplicitness is constantly being "solved" by hearers.
Moreover, "its results are displayed (even when not formulated) in the ensuing
talk and action and are subject to repair there if found problematic" (1996:
219f). This paper has attempted to spell out exactly what might be problematic
about indexically (or prosodically) cued reported speech, thereby making it
repairable, and how in more 'successful' instances, recipients display their
'solutions' of the inexplicitness involved to each other and to analysts.
Notes
1. The collection of articles in Auer and di Luzio (1992), takes one step in this direction.
2. This description is intended to be general enough to apply to all forms of reported
speech, including canonical direct as well as canonical indirect speech.
3. To keep the two notions apart, I shall use scare quotes around voice when reference it to
the animation of a figure; voice without scare quotes refers to phonatory voice.
4. I use the term reported speech sequence for any part of a conversational exchange in
which the presence of reported speech can be identified.
5. Transcription conventions:
One line One intonation phrase
First word capitalized High onset (=declination reset)
[Line
[Line Overlapped utterances
Coherent Voicing 31
Line=
=Line Latched utterances
Line. Final pitch falling to low
Line! Final pitch falling to low, emphatic
Line; Final pitch falling slightly
Line - Final level pitch
Line, Final pitch rising slightly
Line? Final pitch rising to high
«p> Line> Piano
<<ƒ>Line> Forte
« / > Line> Low register
«h> Line> High register
«cresc> Line> Crescendo
«decresc> Line> Decrescendo
ÎWord Noticeable step-up in pitch
Wo::rd Lengthened sound or syllable
Word- Cut-off sound or syllable
WORD Loud volume
°word° Soft volume
word (Extra) stress
/word /word /word Rhythmic delivery
(h) Breathiness
.hhh Inbreath
hhh Outbreath
(word) Unsure transcription
(.) Brief pause
(1.0) Measured pause
Which canonical form is being projected is unclear. The quotative he said is compatible
both with the presence of expressive elements (canonical direct reported speech) and
with their absence (canonical indirect reported speech), whereas the quotative he justified
it as, canonically speaking, projects upcoming talk in which expressive elements are
absent and would thus be considered indirect reported speech.
The noticeable pause following embryos? (line 28) is thus attributable to Gina. Her
silence here may be a first indication of the problem which Bill's conflicting signals are
creating for her.
Alternatively, the breathiness and laughter particles could be interpreted as indexing the
reporting situation, signalling the narrator's commentary on the figure's action (see
below).
Steve's laughter in line 16 can also be thought of as a sign of orientation to Janet's
'voicing' of the student.
10. The phenomenon of rhythmic scansion in everyday discourse is discussed at length in
Auer, Couper-Kuhlen & Mueller (forthcoming).
11. In fact, it is partly due to the distinctive rhythmic cueing that we recognise the voice in
line 43 as the student's and not, for instance, the professor's.
12. Interestingly, the grammatical framing is that of indirect reported speech, not canonically
associated with 'expressivity' or vocal animation: they just decided (line 1) <...> that
(line 3) <...> that Galileo was right (lines 14ff), that maybe he had a good idea (line 25).
See Günthner (1997) for further discussion of this point.
13. Alternatively lines 17-18 could be attributed to Wanda the narrator, with their marked
prosody contextualising her 'take' on the events she is recounting (see below). In this
case the switch to Conversational Historical Present (Wolfson 1979, 1982; Schiffrin
1981) would begin here rather than in line 20.
14. Although we might expect Ricky to enter the scene at this point, the wording of the line
makes clear that Ricky's friend, the kid, is answering Wanda's question.
15. The fact that prior talk contained a first pair-part (what happened?) which makes a
second from some other (addressed) party conditionally relevant may also contribute to
the fact that we hear line 19 as belonging to a different figure.
16. In other words, as Tom Luckmann and Susanne Günthner (p.c.) have pointed out to me,
these characteristics do not contradict each other but are quite compatible.
17. The overlaid 'snort' and the laugh particle could be thought of as cueing the talk of
either the 'later' Ann figure or the narrator Ann. Since Ann the narrator 's stance is
congruent with that of Ann the 'later' figure, the net outcome is the same as far as the
inferencing here is concerned. See, however, fragment (7) below and its analysis.
18. In this sense it is the same practice as that documented in Schegloff's (1996) collection,
namely formulating a candidate observation, interpretation, or understanding of
something which a prior speaker has conveyed without saying: "[...] some telling may be
constructed by its teller, and/or be taken by its recipient, to embody and/or to reveal a
tack that the teller is taking to the tale, some stance being taken up, or some action being
done. When a recipient makes that explicit in the uptake, the teller can confirm both the
particulars of the uptake, its 'propositional content' so to speak, and that he or she was
engaged in such a 'project'" (1996: 188).
19. Note too that Ann 'chimes in' with Janet's vocal framing when she whispers her uptake
yeah I mean (line 221).
20. Note that the broken off I got a kick outa (line 1) and the expressive lengthening on so::
worried (line 29) both serve as cues to Bret's overall framing.
References
Auer, P. and Luzio, A. di. (eds). 1992. The Contextualization of Language.
Amsterdam: Benjamins.
Auer, P., Couper-Kuhlen, E. and Mueller, F. 1999. Language in Time: The Rhythm
and Tempo of Spoken Interaction. New York: Oxford University Press.
Couper-Kuhlen, E. 1993. English Speech Rhythm: Form and Function in Everyday
Verbal Interaction. Amsterdam: Benjamins.
Coherent Voicing 33
Couper-Kuhlen, E. and Selting, M. 1996. "Towards an interactional perspective on

prosody and a prosodic perspective on interaction". In E. Couper-Kuhlen and M.
Selting (eds), Prosody in Conversation. Cambridge: Cambridge University
Press, 11-56.
Goffinan, E. 1981. Forms of Talk. Oxford: Blackwell.
Giinthner, S. 1997. "Stilisierungs verfahren in der Redewiedergabe. Die 'Überlagerung
von Stimmen' als Mittel der moralischen Verurteilung in Vorwurfs
rekonstruktionen". In B. Sandig and M. Selting (eds), Stil und Stilisierung.
Berlin: de Gruyter, 94-122.
Gumperz, J.J. 1982. Discourse Strategies. Cambridge: Cambridge University Press.
Heritage, J. 1984. "A change-of-state token and aspects of its sequential placement". In
M.J. Atkinson and J. Heritage (eds), Structures of Social Action. Studies in
Conversation Analysis. Cambridge: Cambridge University Press, 299-345.
Ladd, D.R. 1978. "Stylized intonation". Language 54: 517-539.
Laver, J. 1980. The Phonetic Description of Voice Quality. Cambridge: Cambridge
University Press.
Laver, J. 1994. Principles of Phonetics. Cambridge: Cambridge University Press.
Levinson, S. 1988. "Putting linguistics on a proper footing: Explorations in Goffman's
concepts of participation". In P. Drew and A.J. Wootton (eds), Erving Goffinan:
Exploring the Interaction Order. Cambridge: Polity Press, 161-227.
Schegloff, E.A. 1990. "On the organization of sequences as a source of 'coherence' in
talk-in-interaction". In B. Dorval (ed), Conversational Organization and its
Development. Norwood: Ablex, 51-77.
Schegloff, E.A. 1996. "Confirming allusions: Towards an empirical account of
actions". American Journal of Sociology 102: 161-216.
Schiffrin, D. 1981. "Tense variation in narrative". Language 57: 45-62.
Selting, M. 1996. "Prosody as an activity-type distinctive cue in conversation: The case
of so-called 'astonished' questions in repair initiation". In E. Couper-Kuhlen and
M. Selting (eds), Prosody in Conversation. Cambridge: Cambridge University
Press, 231-270.
Silverstein, M. 1993. "Metapragmatic discourse and metapragmatic function". In J.A.
Lucy (ed), Reflexive Language. Reported Speech and Metapragmatics.
Cambridge: Cambridge University Press, 33-58.
Tarmen, D. (ed). 1993. Framing in Discourse. Oxford: Oxford University Press.
Wolfson, N. 1979. "The conversational historical present alternation". Language 55:
168-182.
Wolfson, N. 1982. The Conversational Historical Present in American English
Narrative. Dordrecht: Foris.
It Takes Two to Cohere:
The Collaborative Dimension of Topical Coherence in
Conversation1
Ronald Geluykens
Westfälische Wilhelms- Universität Münster
1 Introduction
This paper approaches the notion of coherence from a perhaps somewhat
unusual perspective, viz. that of conversation-analysis (CA). I will argue here in
favour of a more empirically grounded approach to coherence. In fact, although
the notion of coherence is perhaps a useftil cover term for the kind of
phenomena which interest us here, giving a precise characterisation of the term
is not an objective of this paper. Interestingly, CA is defined in Crystal (1997:
92) as "a method of studying the sequential STRUCTURE and COHERENCE
in conversations" [my underscoring]. In the CA literature, however, the term
coherence is conspicuous by its absence. It is striking, for instance, that the
keyword coherence is not even listed in Harvey Sacks' lecture notes (1992).
However, in the local organisation of conversation, interlocutors appear to
reach agreement interactively on what they are talking about, and in doing so
they create coherence as they go along. What I will be mainly dealing with here,
then, is with topic organisation as a reflection of conversational coherence. We
will look at stretches of conversation in which (at least) two participants
attempt to come to some agreement on topical coherence by negotiating about
it.
It is important to point out that this methodology, and that of CA in
general, differs crucially from approaches to coherence in discourse analysis
(DA) (as developed in e.g. Brown and Yule 1983; Stubbs 1983). In the DA
view, coherence is present in the text, and the analyst is able to attempt to find it
by developing some theoretical framework combined with some suitable
heuristics. In other words, this is a top-down approach to coherence. In such
approaches, topic organisation is connected to the notion of aboutness, and the
36 Ronald Geluykens
emphasis is on cognitive properties which tie together chunks of discourse. One

could, for instance, focus on the referential chains in a piece of discourse, and
assume that these somehow reflect topical organisation. This is the approach
taken, for instance, by Givón (1983). As Levinson points out, however:
The point is simply that topical coherence cannot be thought of as residing in some
independently calculable procedure for ascertaining (for example) shared reference
across utterances. Rather, topical coherence is something constructed across turns
by the collaboration of participants. What needs then to be studied is how potential
topics are introduced and collaboratively ratified, how they are marked as 'new',
'touched off, 'misplaced' and so on, how they are avoided or competed over and how
they are collaboratively closed down (Levinson 1983: 315).
In the CA view, on the other hand, it is observed that people appear to be

interacting coherently, i.e. they are aware of local organisation in conversation;
when trouble-spots occur, they are usually quickly resolved. As analysts, we can
study how people interact in certain sensitive contexts (e.g. in transition zones
between one topic and the next) to determine to what extent topical coherence
depends on collaboration. This is thus a bottom-up approach: we study
interlocutors' verbal behaviour and try to draw conclusions about how they deal
interactively with potential coherence problems.
We are making the assumption, then, that topic organisation is relevant to
local coherence; somehow, interlocutors have to agree on what they are talking
about. In fact, it could be argued that topical coherence is a prerequisite if
conversational participants are to interact successfully. Or, to quote Bublitz:
"Topic and the attached topical actions (...) are among the most effective
coherence-forming devices in everyday conversations" (Bublitz 1988: 139). The
point here is that topic organisation is an aspect of the local sequential
organisation of conversation.
The main thesis of this paper is very simple: topical coherence is achieved
collaboratively. This has two different aspects: (a) when a new referential topic
is introduced, the speaker has to prepare the ground by signalling that a topic
change might take place; (b) once a new topic has been introduced, or rather
proposed for introduction, by a participant, then it needs to be negotiated and
acknowledged by the other participant(s) in order to become integrated into the
conversation. In other words, in terms of sequential organisation, the
collaborative dimension of conversational coherence may show up in two
locations:
Topical Coherence and Collaboration 37
(1)
- step 1 : topic-preparing sequence (with possible elicitor)
- step 2: speaker-turn containing first mention of topic
- step 3 : topic-negotiating sequence
I will concentrate on one particular topic-introducing strategy here, viz.

questions. Questions are, by their very nature, interactive, since they form the
first part of a question-answer adjacency pair (Sacks et al. 1974). When a
speaker asks a question which attempts to trigger a new topic, chances of
success are quite high, since the hearer will be expected to respond
appropriately with the second part of the adjacency pair (cf. also Geluykens
1993: 199-204). One has to keep in mind here, of course, that we are dealing
exclusively with spontaneous conversational interaction. Other spoken genres
(e.g. classroom interaction) may well yield other topic-triggering (question)
formats.
Speakers can exploit questions in two different ways in the topic-
introducing process. First of all, the current speaker can ask a question about an
element which they want to propose as a new topic, after which the next
speaker can either accept or reject this element. Topical coherence is thus
achieved through the question-answer sequence (cf. section 2), or through a
more extended negotiation (cf. section 4). I will call such questions topic-
proposing, since the proposed topic is mentioned in the actual question (step 2
in (1) above). Regularly, speakers will pre-signal such questions in the question-
turn or in an earlier turn, to facilitate the negotiation process (cf. section 3; step
1 above).
Secondly, current speakers can attempt to elicit a topic from the next
speaker by asking a relatively open-ended question which explicitly asks for a
new topic on the part of the latter. Such questions will be labelled topic-eliciting
(cf. section 2), and are similar to the topic initial elicitors discussed by Button
and Casey (1984) (step 1 above).
In each case, the outcome, depending as it does on negotiation between
participants, is not guaranteed to be successful (step 3). It will be shown (cf.
section 5) that topical coherence is not achieved when this negotiation process
fails. Such unsuccessful topic-introductions are, in fact, the best evidence that
coherence is indeed an interactive phenomenon, and that it indeed 'takes (at
least) two to cohere' in terms of topic flow.
Four additional points need to be made here. First of all, in step one, the
way transition is achieved from a previous topic can vary. Topic shifting can
take place in stepwise movements, or the topic shift can be boundaried. In both
cases, collaboration is a crucial component. I will only deal here with the so-
38 Ronald Geluykens
called boundaried topic shift (Button and Casey 1984), but it is important to
realise that stepwise transition (Jefferson 1984) can also occur.
Secondly, the main impetus for the introduction of a new topic can come
from the speaker (i.e. the participant who first mentions the new topic) or from
the hearer (i.e. some other participant). In the latter case, the hearer will
typically use a topic initial elicitor (cf. Button and Casey 1984). The difference
between these two types of topic-introduction is reflected in my distinction
between topic-proposing on the one hand, and topic-eliciting on the other hand.
Thirdly, there is probably some correlation between the way a topic is
prepared on the one hand (step 1), and the way it is negotiated on the other
hand (step 3). While an investigation into this potential correlation lies outside
the scope of this paper, it seems a logical assumption to make. A topic which is
explicitly elicited by the hearer, for instance, may have no need for elaborate
negotiation; conversely, a topic which is introduced 'out of the blue', without
any link to prior topic(s), may require extensive negotiation.
Finally, the fact that topics need to be negotiated opens the possibility that
attempts to introduce a new topic may be unsuccessful. This is in fact a
relatively rare occurrence, since participants in informal conversations tend to
be co-operative and supportive in this respect (cf. Bublitz 1988 for an extensive
discussion).
A final aspect of the topic-introductions considered here that needs to be
pointed out is the following. In the CA tradition, little or no attention is paid to
the social roles of speaker and hearer. Such social positions may, in fact, have
some influence on the way participants use questions as topic-introducing
strategies. Factors such as relative power, solidarity, and even gender, may have
an influence on who does the topic-eliciting and/or -proposing in a given
context. My CA-based analysis disregards such potential variables.
Data employed here are from the conversational files of the Survey of
English Usage, based at University College London.2
2 Simple negotiations
2.1 Topic-proposing by current speaker
As a first example of how a new topic can be introduced in a straightforward

manner, with minimal negotiation, consider (2):
(2)
A: oh yes it is a household god of some sort isn't it I should think or is it a dancer I
don't know it's got an enormous belly —
B: ((m))-
how do you get on with Thorpe

((oh) - ((we)) get on quite well I think . rare occasions I see him (. giggles)
(. giggles) —
no one does see him very much -
(S.1.6.72.1)
In this exchange, the first turn by speaker A is obviously unconnected to

what follows, and there is a long pause in the second turn (speaker B) which
shows that the current topic appears to be exhausted. Speaker A then asks a
question containing the proposed topic getting on with Thorpe, which is
immediately accepted by B, after which the new topic is developed further by
both participants. Minimal as this negotiation may be, this example shows that
topic shifts require the active participation of both participants.
To illustrate the fact that we are dealing with boundaried topic shifts here,
i.e. with the new topic being unrelated to what has gone on before, let us
examine (3) below:
(3)
1 A: (...) he'd probably have burst into tears *and* committed suicide in front
of me
2 B: *yeah*
3 A&B (-- laugh) -
4 A: (- laugh)
5 B: well. *how eventful*
6 A: *((m)) (- gasps)* yes . ((indeed)). so -
7 B: so what sort's thisflatyou're in at the moment
8 A: well it. [∂:m] do you remember Jackie .
9 B: yes . m
10 A: it's her flat.
(S.2.12.120.8)
All the material in turns 2 to 6 indicates a closing off of the initial topic of
the first turn, and in turn 7 speaker B proposes something entirely new. As we
will see later on (section 3), in quite a few cases the current speaker will pre-
signal the fact that he/she is about to propose such a new topic. This makes
sense, as it may facilitate the negotiation process, and avoid potential confusion
as regards topic flow.
40 Ronald Geluykens
2.2 Topic-eliciting by current speaker
Whereas the questions posed by the speakers during their conversation in the
preceding section constitute the second step in the negotiating process (i.e. the
actual first mention of the proposed topic), the questions discussed in this
section merely attempt to elicit a topic on the part of the second speaker, and
thus represent the topic-eliciting first stage in the process. A clear example is (4)
below:
(4)
A: oh fine was there anything else [∂:m]
B: so I just [d] I'd left some some records in Smart's room last
night which I was all panic-stricken about 'cos they're not mine
A: m
B: but I told Ned
(S.7.1.e5.6)
What speaker A does here is explicitly prompt B to provide something

else, which B proceeds to do, introducing the topic some records in the process.
Note also that, after B's turn, there is some agreement signal from the hearer,
which shows that such topics, even though they have been elicited by the first
speaker, still need to be agreed upon by the other participant(s). We could
represent the interactional process taking place as follows:
(4')
Stage 1 (A): topic-eliciting question
Stage 2 (B): topic-introduction (response)
Stage 3 (A): acceptance of proposed topic
Instance (4) discussed above could be labelled an open-ended topic-

eliciting question, in that it leaves the hearer completely free in his or her choice
of a new topic. There are, however, some topic-eliciting questions which do
narrow down the choice of topic to some extent. Instance (5) is a good example
of this strategy:
(5)
A: good . [∂:m]. any more that we want to talk about the library —
CF: I *I ((mean some of the some of the books aren't there at all))*
A: *((I)) mean this is this is this is* . absolutely at the heart
of your teaching (...)
(S.3.3.40.5)
In this exchange, speaker A seeks to elicit a topic, but at the same time he
determines that the topic should be related to the library; the hearer (CF)
responds co-operatively by uttering the semantically related referent some of the
books.
Whereas truly open-ended wh-questions leave it more or less entirely up
to the hearer to pick a topic, a wh-question like (5) above restricts the number
of options the hearer can select from as a possible topic. Let us have a brief look
at another example:
(6)
A: but what was your conference about
B: oh it's about hydrology . ((and that sort of thing)) -
C: is that water
B: *water*
A: *be more* explicit that's just water to us -
B: [dm]
C: the science of water
B: hydrology is the science of- science of the quantity of water
(S.2.8.b10.7)
In this instance, speaker A is probably trying to elicit a specific topic (i.e.

hydrology) from participant B; he does not, therefore, give him complete
freedom, as the referent is already specified by the speaker as being semantically
related to your conference. These non-open-ended questions are, as it were,
halfway between open-ended topic-eliciting questions (in which the hearer is
free as to what topic to choose) on the one hand, and topic-proposing questions
(in which the speaker already provides a topic, cf. supra) on the other hand.
3 Pre-signalling a proposed topic

The rather abrupt topic shift examples presented in this paper show that it
makes sense for the current speaker to pre-signal to the next speaker that a new
topic is about to be proposed. A variety of such pre-signalling cues can be
found in the data. I would simply like to provide some exemplification here,
without going into details (pre-signalling cues are highlighted in boldface):
(7)
B: aren't you going to tell me what it is -
A: no -
42 Ronald Geluykens
B: anyway [∂:]what was I going to say . oh yes so what time are you coming
this afternoon
A: [dm] . [∂w∂] as we said about four o'clock
(S.7.2.b5.4)
(8)
A: (...) oh I must do that sometime - oh yes one thing too . [d:m] . are you at all
interested in coming to the B Minor Mass -
B: ((when is it)) the fifteenth of April
(S.7.3.f40.1)
(9)
C: well now Elsie follow it up this way do you visit your own mother -
B: well I do and I feel very sorry for her (...)

(S.1.12.102.8)
(10)
B: right. *I'll . ((think)) about that*
A: *if you do let me know* . yeah - m ~ what else . haven't been up to Wales
again have you . or
B: [∂:m]I went up at Easter -
(S.7.3.f48.1)
In all four instances, the hearer provides some signal within the topic-
proposing turn that such a topic proposal is coming up. Since the next speaker
is forewarned, this may facilitate subsequent negotiation of the pre-announced
topic.
Whereas we find a high frequency of cues which signal the fact that the
current speaker is about to propose a new topic, this is not the only way in
which a topic shift can be announced. Rather than announcing the new topic,
the current speaker can also provide some cues which signal closing of previous
topic. This is the case in (11) and (12) below (closing cues highlighted in
boldface):
(11)
C: so you exposed it
A: I exposed it yes ((but)) [i?] but they can't get it back - . well I mustn't go on
boringly talking about me what are you doing Geoffrey -
B: [∂m] [∂:]much the same old things .
(S.1.9.99.7)
(12)
A: well that's Eileen's four . and we covered Margaret
what's her son doing now - ((Dan))
*Dan -
Dan*
*oh well Dan's doing quite* well (...)
(S.1.13.74.1)
In (11), the speaker explicitly signals the topic change coming up, by
indicating that the current topic (me) is finished; in (12), the speaker does so
more indirectly, by summarising the previously discussed topics.
Needless to say, both types of cue (pre-signalling of next topic and closing
of current topic) can be combined, resulting in explicit boundaries. Perhaps the
most elaborate instance of pre-signalling in the database may be found in
instance (13) below, since it combines a closing cue regarding the previous
topic with a pre-introductory cue regarding the proposed topic:
(13)
A: (...)" [∂:m] you're very kind old Sam — bless you well that finishes that. [∂:m]
. ((now)) what was the other thing I wanted to ask you . [ïi] is . is it this year
that \d:] Nightingale goes -
B: [∂:]no next year --
(S. 1.1.15.10)
In this instance, the speaker not only explicitly closes the previous topic
(cf. that finishes that), but also announces his new topic (cf. what was the other
thing...).
It is worth mentioning one example in which the topic-introducing wh-
question is, as it were, 'announced' by means of some explicit verbal material in
a previous turn rather than in the actual topic-proposing turn; consider (14):
(14)
C: I thought of a . new word that I. marked to discuss with you . at some time
B: yes
C: but I I'd better look up to see if it's in the dictionary first what about opt out
B: *yes*
C: *now* that must have a Latin root
B: m
C: I guess it must be something to do with the army (...)
(S.2.5.a61.3)
44 Ronald Geluykens
Two things are worth mentioning about the exchange in (14). First of all,
note the explicit introducing material in C' s first turn; this paves the way for his
subsequent wh-question, as it prepares the hearer (who acknowledges the
introductory clause!) that something is about to come up. Secondly, (14) is also
interesting in that it represents an interactional variant on exchanges such as (2)-
(3) above. Note that in this exchange, it is the speaker himself rather than the
hearer who establishes the referent; the hearer merely prompts the speaker to go
on about his newly proposed topic (the word opt out). We thus get the
following underlying process:
(14')
Stage 1 (C) introductory material
Stage 2 (B) acknowledgement of introductory material
Stage 3 (C) topic-proposing (Wh-phrase)
Stage 4 (B) acknowledgement of Wh-phrase
Stage 5 (C) elaboration on proposed topic
Stage 6 (B) acknowledgement signal
Stage 7 (C) further elaboration on proposed topic
What is especially interesting here are stages 3 to 5, which represent an

untypical interactional development. It is interesting to note that B in stage 4
does not really give an unmarked response to a wh-question; he merely
acknowledges the relevance of the question. However, this does not mean, in
this case, that the hearer is uncooperative: he merely gives a marked response,
enabling the speaker to go on about the new referent. In stage 5, it is thus the
speaker rather than the hearer who follows up on the proposed topic; it is
important to realise, however, that he can only do so by virtue of the
encouraging marked response from the hearer in stage 4. The way in which the
topic becomes integrated thus depends to a large extent on the way the speaker
chooses to introduce it.
4 Complex negotiations
In all the instances of topic-proposing mentioned so far, the topic-introducing
process is very smooth: negotiation of the new proposed topic takes place
through the adjacency pair sequence, after which speaker and hearer appear to
agree on the new topic. The same can be said about the topic-eliciting
exchanges: after the current speaker elicits a new topic (through the question),
the next speaker provides one (through the response), after which this
suggested topic is developed further. Such smooth negotiations, however, are
by no means self-evident. Often, in order to achieve topical coherence

collaboratively, the proposed topic has to be negotiated over a number of turns
before being accepted by the hearer; in other words, step 3 in (1) above is far
more complex.
One context in which such extensive negotiation takes places is when the
identity of the proposed topic is unclear to the next speaker. In that case, the
latter will prompt the current speaker to provide some additional information.
The result is that the whole process can take several turns until topical
coherence is achieved.
(15)
A: m - do you remember Terry Greenbridge .
B: Terry Greenbridge
A: he was
B: *yeah who was he*
A: *he used to be* master of Hereford College -
B: was he a theologian **((or someth)) yes**
A: **yes that's right that's** right. yes . well he he wrote and told me all about
this .(...)
(S.5.9.7.9)
(16)
C: have you tried the bookcase -
B: bookcase
C: in our room —
B: eh.
C: you know the big glass fronted bookcase -
B; *yeah*
C: *the* new one
B: oh that [∂.m] no as a matter of fact (...)
(S.7.1.a57.7)
In both of these instances, the hearer has difficulties in identifying the

intended topic {Terry Greenbridge and the bookcase, respectively); we thus get
a negotiating process similar to the one already encountered in previous
chapters, and which can be summarised in the following manner:
(15')
Stage 1 (A) topic-introduction (Q)
Stage 2 (B) request for clarification
Stage 3 (A) clarification of referent
46 Ronald Geluykens
Stage 4 (A/B): topic-establishment
If one allows for the fact that stages 2 and 3 can be repeated (as is the
case in the examples above), this schema represents the majority of more
complex instances. Note that the topic in such cases can be eventually
developed by the current speaker, as in (15), or by the next speaker, as in (16).
Instance (17) below is slightly different, in that the next speaker
contributes actively to the identification process:
(17)
A: yes but ((Power)) can ask him if he sees him every day -
B: yeah but you need to know who he is 'cos he is much the more important man -
(noise)
A: true --
B; I just don't remember at all.
A: what about the others I ought to ring up the others ((didn't)) I . there ((was))
another one I sent it to . Wills —
B: oh the people we met first.
A: yes
B: the quite attractive young man and the older man .
A: yes ~
B: yes we ought to I think they might be quite reasonable - they looked a bit more
sort of ((rooking)) -
(S.4.2.10.2)
In this exchange, what appears to happen is the following: A first

introduces a new topical referent (the others), which he then reformulates; B
then asks a counter-question to ensure correct identification of the introduced
referent (this, in fact, happens twice); it is only after this repeated identification
exchange that B answers A's original topic-introducing question, thereby
creating topical coherence.
Another type of trouble spot in coherence can occur if there is confusion
about the intended reference of some element, giving rise to a repair sequence.
This appears to be case in (18):
(18)
B what's that weird creature over there ~ *m*
C *in* the corner
B mhm
C [d] it's just a [dm] fern plant
B: no the one to the left of it

C: that's [thi:] television aerial a the aerial - it pulls out
(S.2.4.al.l)
In exchange (18), the identity of the NP that weird creature is not self-
evident from the extralinguistic context, which leads speaker C to assume
incorrectly that B means the fern plant rather than the television aerial that was
apparently intended. Once again, the conversation cannot proceed in a coherent
manner until this trouble spot is resolved.
Another instance of a complex interaction caused by the need for repair is
(19) below:
(19)
A: hm but they do disapprove of his [= McCarthy's] methods .
B: yeah most of them - yeah he has about twenty five per cent ((of)) support as
far as that's concerned and much more - not - and many don't know -
A: yes — what about in the university circles at Michigan —
B: at Michigan - *you mean Wisconsin*
A: *oh no no I mean* Wisconsin .
B: well I have never found anybody admitting to a liking for McCarthy or his
methods - *-*
A: *no*
B: . but nevertheless Gallup says . that in . university circles McCarthy is as
popular as he is in any other circles - ((2 or 3 sylls))
(S.2.1.b8.2)
In this particular instance, speaker A introduces a noun phrase {university

circles at Michigan), part of which is queried by the hearer (the latter obviously
being able to deduce from his background knowledge that A in fact meant to
refer to Wisconsin rather than Michigan); this query is confirmed by the
speaker, after which B answers A's original wh-question, thereby allowing the
new topic to be followed up.
While participants are generally able to deal with these trouble spots and
resolve them relatively swiftly through negotiation, this is not always the case.
In some cases, negotiation can fail, resulting in a topic which does not get
introduced coherently, as I will demonstrate in the next section.
48 Ronald Geluykens
5 Unsuccessful negotiations
In an earlier paper (Geluykens 1995), I have already devoted some attention to
what I called short-circuited topic-introductions. Since topic flow depends on
local negotiation through the turn taking system, successful introduction of a
new topic is by no means self-evident. In Geluykens (1995: 235-239), I
identified three underlying reasons which may cause short-circuiting. First of all,
the speaker may decide to abort a new topic in the middle of the negotiating
process; needless to say, this is a rare occurrence. Secondly, the hearer may
decide to interfere, and stop a topic from being developed (cf. below). Thirdly,
short-circuiting may be the result of competition for the floor, as appears to be
the case in (20):
(20)
C: m — and yet you feel terribly antisocial if you . ((you)) do just stay in the
kitchen anyway
a: yes - *what film*
C: *oh god those* stairs oh oh
a: - sorry
C: those stairs — **((you could . you could))**
a: __ **yes they're bad** aren't they - good exercise though
C: oh ((yes)) (- laughs) -
a: what film have you just been to see .
C: film.
a: I thought you went. you were going to the National - Theatre - National Film
Theatre
C: no no . [∂:m] . that was at the weekend (...)
(S.2.7.10.5)
In this instance, speaker a has started producing the wh-question, but

speaker C is still involved in a side sequence (which he started simultaneously);
speaker a thus makes a second attempt after having dealt with the side
sequence; schematically, this looks like (20'):
(20')
Stage 1 (A): topic-introduction (first attempt)
Stage 2 (B): non-acknowledgement (side sequence)
Stage 3 (A): topic-introduction (second attempt)
Stage 4 (B): successful topical coherence
The outcome, in this particular case, is eventually successful, since the

topic does get taken up after the second attempt. A similar case is (21) below:
(21)
C: but are books more *expensive in Australia*
A: *tell them with my compliments*
all: (— laugh)
A: artificially cheap my arse (- laughs)
C: I say are they more expensive in *Australia*
A: *oh* god yes-(...)
(S. 1.10.97.5)
In this exchange, participant A has simply ignored the question the first
time, in other words he has not made any attempt at being co-operative; note in
this respect the overlap between the first two turns. This leads to the speaker to
repeat his question, resulting in a four-stage process similar to (20').
However, even without such floor-competition, topic-introductions may
be unsuccessful. In (22) below, for instance, the topic introduced by speaker B
is not responded to:
(22)
A: *m* [...] have a glass of sherry .
B: oh that's nice ((of you as)) I'm not driving . thank you
A: bloody hell —
B: but what about you \d] Crispin . [d] what's what's [thi] - how far were you
[∂:m]
((ye gods))
[∂:m] banking on *this*
*((...)) this k e y -
locked yourself out .
yes
(. laughs)
no the trouble is . oh for god's sake the key ((won't go in the lock)) (...)
(S. 1.2.52.4)
In this particular instance, speaker B tries to turn a reference to the hearer

himself (cf. you) into a topic for the subsequent discourse, but fails to do so, as
the hearer is busy trying to open a door with his key, and ignores B's question;
the attempt to introduce the topic is subsequently abandoned by B. Since the
hearer can normally be assumed to be cooperative, the fact that such short-
circuited topic-introductions are rare is not very surprising, for it is only under
very restricted conditions that the hearer can 'get away with' ignoring the
50 Ronald Geluykens
speaker's question. Cases such as this do show, however, that without

collaboration there is no topical coherence.
While we have few illustrations of totally uncooperative behaviour on the
part of the next speaker, there is one specific situation which can arise after a
topic-eliciting question. In such cases, it is not unthinkable that the next speaker
does not really have an idea about which topic to produce after being prompted
by the current speaker. (23) is such an instance:
(23)
B: I'm afraid [thi:] fifth of February isn't all right for [thi:] - for [thi:] [dm] lunch
with Mr Parrot.
A: not
B: no cos he had something that he [?] . that I didn't realise .
A: [∂h∂ . ∂:m] well have you any other suggestions .
B: well he didn't give me any (- laughs)
A: ch
B: [∂:m] -
A: can I make . have you got his book there I mean
B: yes I have yes
(S.8.2.f3.2)
In this instance, one could not exactly label B 'uncooperative', because he

does supply an answer; the answer, however, fails to respond to the topic-
eliciting aspect of A's question; as a result, A himself ends up introducing a new
topic (cf. have you got his books) into the discourse.
Given the nature of topic-eliciting questions, which give the hearer an
ideal opportunity to introduce a new topic, and given the co-operative principle,
responses such as the one in (23) can be considered non-prototypical. Exchange
(24) presents a similar case:
(24)
A: (...) I could go on being a research assistant going up and up and up until I
((was at)) . eighty five or whatever
C: m -
A: so what's new Ann --.
C: well I don't know if anything's terribly new at all . really . ((or is it)) all much
the same (. laughs) - *((just been))*
B: *((you still)) living* with Deb -
C: no no(...)
(S.2.7.48.8)
In this exchange, the next speaker again (C) explicitly states that he is not
really in a position to introduce a new topic; another participant then decides to
suggest a new topic (the living with Deb in the fourth turn), which the original
addressee replies to, thereby developing this new topic.
One final exchange containing a topic-eliciting question which is not
appropriately responded to is worth mentioning here, viz. (25):
(25)
A: (...)- you will certainly be involved I should think .
D: ah
A: because I was talking to them about their language policy over the next few
years
D: yes .
A: [∂:m] . yeah what [el] what other news have you got -
D: [∂:]
A: [so] I [wo] [d] I should just add that II meant to send you [∂:] a postcard
signed by Julius and myself but you know the way it is
D: yes
A: [w∂] you get chased around like a scalded cat (...)
(S.9.2.111.1)
In this instance, the original speaker (A) ends up responding to his own
topic-eliciting question. This appears to be due to the fact that next speaker D,
when given the floor, fails to come up with a timely response (cf. D's hesitation
in the sixth turn); speaker A then decides to resume the floor. Although this
exchange cannot really be labelled unsuccessful, it once again shows that, when
participants do not do the expected thing in terms of local topic organisation,
some re-organising is called for in order to preserve or create topical coherence.
6 Conclusion
I have tried to show in this paper how conversational coherence is
collaboratively achieved between the interlocutors. In particular, topic flow,
which is arguably one of the most important measures of a stretch of
conversation, depends for its regulation on the negotiation of new topics. It was
shown that the conversational activity of asking questions can be employed for
these purposes. Question-answer adjacency pairs can provide the framework
through which topical coherence is achieved, and they can do this in two
different ways. First of all, the current speaker can propose a new topic in
52 Ronald Geluykens
his/her question, giving the next speaker the opportunity to accept (or decline)
this proposed topic in the second part of the adjacency pair, the response to the
question. Alternatively, the current speaker can ask a question which elicits a
potential new topic from the next speaker. In both cases, we appear to be
dealing with boundaried topic shifts, i.e. some signal is given that the topic-
introducing question which is about to follow is unrelated to the previous
discourse, and/or some cues are provided that suggest that a previous topic has
been rounded off.
When such negotiation initially fails, participants strive to resolve this
potential trouble spot in subsequent turns. In that case, the potential outcome is
one of two possibilities. If the next speaker is unable to accept the proposed
new topic for some reason (e.g. because some element in the topic-proposing
question is unclear), then some further negotiation will take place until this
trouble spot is resolved. If, on the other hand, the current speaker's proposal of
a new topic is not accepted by the next speaker (e.g. if he/she has an alternative
topic they want to pursue), then the negotiation process will be suspended,
sometimes temporarily, sometimes permanently. In this case, too, participants
will collaborate until a satisfactory point is reached where coherence is again
apparent.
It would be unwise to think that this paper presents a complete picture of
conversational coherence. For one thing, one could question whether topical
coherence - which in itself depends to a substantial extent on referential
coherence - is a sufficient prerequisite for achieving conversational coherence.
For another, this paper is limited to one particular conversational device, viz.
questioning, as an illustration of the collaborative nature of coherence. Further
studies will hopefully provide us with a more complete picture of the variety of
strategies interlocutors use for creating topical coherence through the turn-
taking system (cf. Geluykens 1988, 1991, 1992, in prep.).
Notes
1. Thanks are due to Wolfram Bublitz, Uta Lenk, and Eija Ventola for their comments on
an earlier version of this paper.
2. Transcription conventions for the current purposes are the following:
—speaker identity: A, B, C,...
—overlapping speech :*...* or **...**
—dubious transcription or intranscribable speech: ((...))
—occurrence of pauses (from short to long):., -, - , —
—phonetic transcription: [...] (e.g. [d] = schwa)
—topic-introducing question: underlined.
References
Atkinson, J.M. and Heritage, J. (eds). 1984. Structures of Social Action: Studies in
Conversation Analysis. Cambridge: Cambridge University Press.
Bublitz, W. 1988. Supportive Fellow-Speakers and Cooperative Conversations.
Button, G. and Casey, N. 1984. "Generating topic: the use of topic initial elicitors". In
J.M. Atkinson and J. Heritage (eds), Structures of Social Action: Studies in
Conversation Analysis. Cambridge: Cambridge University Press, 167-190.
Crystal, D. 21997. A Dictionary of Linguistics and Phonetics. Oxford: Blackwell.
Geluykens, R. 1988. "The interactional nature of referent-introduction". Papers from
the 24th Regional Meeting of the Chicago Linguistic Society: 141-154.
Geluykens, R. 1991. "Topic management in conversational discourse: the collaborative
dimension". Papers from the 27th Regional Meeting of the Chicago Linguistic
Society: 182-195.
Geluykens, R. 1992. From Discourse Process to Grammatical Construction: On Left-
Dislocation in English. Amsterdam: Benjamins.
Geluykens, R. 1993. "Topic-introduction in English conversation". Transactions of the
Philological Society 91:189-214.
Geluykens, R. 1995. "On establishing reference in conversation". In R.A. Geiger (ed),
Reference in Multidisciplinary Perspective. Hildesheim: Olms, 230-240.
Geluykens, R. (in prep). "Information flow through interaction: Topic organization in
English conversation". Ms., University of Münster.
Jefferson, G. 1984. "On stepwise transition from talk about a trouble to inappropriately
next-positioned matters". In J.M. Atkinson and J. Heritage (eds), Structures of
Social Action: Studies in Conversation Analysis. Cambridge: Cambridge
University Press, 191-222.
Levinson, S. 1983. Pragmatics. Cambridge: Cambridge University Press.
Sacks, H. 1992. Lectures on conversation. (2 vols.) Oxford: Blackwell.
Sacks, H., Schegloff, E.A. and Jefferson, G. 1974. "A simplest systematics for the
organization of turn-taking in conversation". Language 50: 696-735.
Learning to Cohere: Causal Links in Native vs.
Non-Native Argumentative Writing
Gunter Lorenz
1 Review of the issue: cohesive ties in learners' writing

Among other characteristics, writing differs from face-to-face interaction in the
way coherence is constructed. In written communication - mostly monologous
and without direct contact between writer and reader - coherence cannot be
explicitly negotiated. Instead, there is an implicit co-construction of meaning,
and writers therefore have every reason for trying to be unequivocal and to
make their ideas, intentions and arguments unmistakably clear. One way of
doing this is to carefully signal logical relations and thereby 'signpost' the path
to coherence for the reader. Consequently, when looking at the fabrication of
coherence in written discourse, we need to pay special attention to those
explicit signposts of coherence, i.e. the text's cohesive ties.
Despite the fact that logical relations are often overtly signposted, their
construction and re-construction remains an intricate to-ing and fro-ing between
the writer and the reader, an interplay of given and new information, of mutually
shared knowledge and the message to be imparted. It is hardly surprising, then,
that this complex process is easily disrupted when the shared knowledge of
reader and writer becomes a less than reliable source of information. This is
particularly evident in native/non-native speaker interaction. It has repeatedly
been noted that language learners, even at advanced stages of acquisition, have
considerable deficits in creating a coherent text.
There has been a fair amount of speculation as to the roots of this
phenomenon. On the surface, of course, it has already been said that the shaping
of coherence is a complex task, and there is plenty of reason why learners
should find it difficult. Yet on the other hand, experienced assessors of non-
native texts have always found it difficult to see why otherwise perfectly
intelligent human beings, who have mastered the biggest part of English syntax
and morphology, should so often slip up in the realm of coherent argument.
56 Gunter Lorenz
It is tempting to postulate that deviant L2 production arises from a

conflict of overall discourse organising principles between L1 and L2. This
hypothesis has become known as 'contrastive rhetoric', most famously through
the work of Robert Kaplan and his associates. Regardless of whether or not the
contrastive rhetoric paradigm deserves the high esteem it has been held in, it can
hardly provide a comprehensive explanation for all non-native coherence
problems. On the contrary, many patterns that are felt to be deviant do not seem
to be Ll-motivated at all; if they were, i.e. if learners were merely following the
textual conventions of their native language, one would expect non-native
markers who share the writers' L1 background to be far less perturbed by the
deviancies in L2 discourse. By all professional accounts, however, this is not the
case. Secondly, it has been widely acknowledged in recent years that advanced
learners' deficits are most resilient in the area of lexico-grammar, where lexical
items are employed to signal grammatical and textual relations. Since cohesive
ties clearly fall into this field, a lack of coherence in advanced learners' writing
must at least partly be attributable to lexico-grammatical deficits.
According to this latter hypothesis, the number and (mis-)use of cohesive
devices in learners' discourse should somehow function as indicators of the
texts' coherence. If such a relationship could be substantiated, this would go a
long way towards explaining - and ultimately even teaching - an otherwise
elusive textual quality.
Naturally, this issue is not being addressed for the first time; previous
research into EFL connector usage has produced a number of rather
idiosyncratic results. Hong Kong learners, for example, have been reported to
overuse logical connectors in general (Field and Yip 1992; Milton and Tsang
1993) - possibly due to the fact that Hong Kong examination guidelines award
points "for the mere presence of connectors, often regardless of their
appropriacy" (Crewe 1990: 325). For Finnish writers, Ventola and Mauranen
(1991) note an underuse of connectors and a positive correlation of frequency
and EFL proficiency. There was little variation in the learners' usage, who
seemed to rely on a few all-purpose favourites. Lack of variety has also been
found in Norwegian EFL connector usage (Lintermann-Rygh 1985), and while
here the more proficient learners tended to use more connectors, the better
native English writers actually used fewer. It is also worth noting that
Norwegians seem to use more connectors in EFL discourse than in their native
language (Evenson and Lintermann-Rygh 1988) - a finding which, incidentally,
casts further doubt on the merits of the contrastive rhetoric paradigm.
According to Granger and Tyson (1996), lastly, native French texts tend
to be more explicitly structured by connectors than native English ones, and one
might therefore expect French speakers to overuse connectors in EFL writing.
Yet this did not turn out to be the case: there was hardly a difference in native
Learning to Cohere 57
and non-native connector frequency, but rather in stylistic aspects such as

position and register.
Despite this overall diversity of results, connector usage is evidently an
area which is sensitive to markedly non-native style. Whichever way a given
group of learners may lean in terms of connector frequency, and whether or not
the number of connectives correlates with proficiency, the way in which logical
relations are signalled does have significant bearing on the subjective coherence
within learners' writing. The present paper builds on this assumption, with an
investigation of logical linkage in advanced German EFL writing. And among
the various logical relations in a text, it will concentrate on those of CAUSE and
EFFECT. Starting out from contrastive corpus searches, it seeks to determine the
various causal connectives employed, and to glean meaningful insights from the
learners' strategies and preferences.
2 A contrastive developmental corpus: rationale and retrieval

When dealing with preferences of usage, one is inevitably concerned with
questions of frequency. In large enough samples of naturally occurring
language, systematic stylistic preferences become manifest as numerical
patterns. This is also true for non-native style, when contrasted with samples of
native usage. The natural testbed for this purpose is a contrastive learner/native
speaker corpus which provides a sufficient amount of lexico-grammatical data.
This study is based on data gained from a contrastive corpus of
argumentative essays, with a four-partite structure:
BWF 142, 131 words from 16-18-year-old German learners

UNI 71, 881 words from German undergraduates
GCE 106,730 words from 15-18-year-old British writers
LOC 94,962 words from British undergraduates (LOCNESS)
In addition to reflecting German versus British English usage, the corpus

affords a quasi-developmental comparison. It takes account of the fact that it is
not only EFL usage which is subject to change and - hopefully - refinement, but
that the same is also true for juvenile native discourse. Corpus searches can
therefore not only bring out the differences between native and non-native style,
but also show whether or not the two groups are on the same route towards
linguistic maturation.
For the present purpose, argumentative writing promises to be a rich
resource: written argument entails an interplay of thesis and antithesis, cause
and effect, as well as super- and subordination. The extent to which this is the
58 Gunter Lorenz
case may differ across cultures and educational contexts, and there may likewise
be differences as to an argument's structural properties. Yet all four corpora are
meant to serve the same - argumentative - purpose.
Similarly, the connective signals under scrutiny are not defined formally,
but functionally: the corpus searches not only include the well-established causal
connectors, but also attempt to explore the more ad hoc type of causal linking
'strategies'. In this manner, the investigation tries to acknowledge that logical
relations are often lexically paraphrased, rather than being expressed as one
single conjunction, adverb etc. As paraphrasing strategies are known to be a
prominent feature of non-native discourse, a search which merely considered
the standard connectors would be severely invalidated.
Such a functional selection of lexico-grammatical data can, of course, not
be carried out in a purely mechanical way. Each potential item has to be viewed
and evaluated in context. Not all causal markers are causal in all their usages
(compare, most notably, since and for) and the numbers would be skewed if all
that was done were to count word-forms. Moreover, frequency profiles are not
the goal of this investigation, but rather its starting-point. Differences in number
may point to differences in style, which can only be properly gauged manually.
3 Causal conjunctions I: dealing with numerical data

As a starting point, the well-known, grammaticalised causal connectors can be
found listed in the standard grammars and learners' usage guides. And among
them, the conjunctions are probably the prototypical subset; they combine
clauses depicting CAUSE with those depicting EFFECT. The corpus counts are
reproduced in table 1 below.
Note that, despite the differences in corpus size, the numbers can be
directly compared. Each one represents a standardised frequency (SF),
statistically normalised as "number of occurrences per 100,000 words"; hence
such oddly decimal frequencies as "0.9" or "11.7". Admittedly, standardised
figures are hardly reliable in the lower frequency band. As will become
apparent, however, even low frequencies gain in significance when falling into a
larger numerical pattern. The present view of corpus statistics is a cumulative
one, concerned with trends, tendencies and developments. Overall tendencies
are only pointed out when they are statistically significant on the .05 level at
least (as determined by 2 ). For the individual small counts, no claim of
statistical significance is being made, and yet it cannot be ignored that, more
often than not, they do bear out the general tendency.
Table 1: standardised frequencies of causal conjunctions
Conjunctions BWF UNI GCE LOC [
Because 262.4 192.7 157.4 151.6

As 87.2 57.8 137.7 168.5
Since 43.6 10.3 9.4 33.7
For 30.3 10.3 3.8 20.0
Seeing as — — 0.9 —
in that ... ... 1.9 20.0
in asfar as 0.7 — — —
that is why 13.4 11.7 5.6 LI
this is why 4.9 ... 3.8 2.1
which is why 0.7 ... 0.9 1.1
Σ (SF) 443.2 282.8 321.4 398.1
In the case of causal conjunctions, a simple NS/NNS comparison is not very

revealing: in summing up the bottom-line totals for learners (BWF + UNI=
NNS 726.0 SF) and native speakers (GCE + LOC = NS 719.5 SF), there is a
very slight non-native overuse (<1%). Not least for its statistical insignificance
(2 =0.025), however, any interpretations from this would be ill-founded and
haphazard. Yet this case shows how a simple two-way contrast may fail to
account for a significant difference in the two populations' stylistic
development: in the process of linguistic maturation, i.e. from sixth form to
undergraduate level, the use of causal conjunctions undergoes a substantial
decrease on the learners' part (from 443.2 to 282.8 SF), and increase on the
part of the native speakers (from 321.4 to 398.1 SF).
The conjunctions fall into three different groups: the standard simplex
ones (because, as, since, for), well-established composite ones (seeing as, in
that, in as far as ) , as well as resultative phrasals (that is why, this is why, which
is why). The corpus counts for the first group give rise to several observations:
the figures for because, for example, the most frequent causal conjunction, are
almost level between GCE and LOC, the two native speaker corpora; there
does not seem to be much stylistic development underway. The learner corpora,
conversely, are much less stable.
First of all, there is a substantial non-native overuse of because, by far the
most frequent item on the list. This is a well-known phenomenon: learner usage
tends to amplify the high frequencies and diminish the low ones. Learners are
naturally much more exposed to high-frequency words and patterns, and are
hence more likely to acquire and use them; the reverse, of course, applies to
60 Gunter Lorenz
low-frequency items. The figures for because point towards a 'wild-card use'.
Secondly, there is a distinct decrease of because from BWF to UNI, i.e. from
the less to the more mature learners. At face value, this development is a very
welcome one, and one which appears to be set to rectify the statistical learners'
overuse. On further inspection, however, the learners' decreasing use of
because is more plausibly interpreted as part of an overall trend: as, since and
for are also declining from BWF to UNI, while undergoing a marked rise from
GCE to LOC. Without pre-empting a more thorough evaluation, there should
be a stylistic explanation to this numerical pattern.
The second group, on the other hand, does not really offer many insights;
the items are comparatively infrequent - with one small exception: explicatory in
that is a markedly native usage, mostly occurring in LOC, the more mature
corpus.
Group three conjunctions, while being only marginally more frequent, are
nonetheless slightly more revealing: the learners employed more than twice as
many paraphrases of the "pron is why" type than the native writers (30.7 vs.
14.6 SF). This finding confirms the suspicion that learners are more prone to
using explicit paraphrasing strategies. (Note also how in the native speakers'
data, too, such constructions occur much more frequently with the younger
population, indicating a negative correlation with linguistic maturity.)
It can already be seen how the developmental corpus structure enables a
kind of reasoning that a simple NS/NNS comparison could not achieve. In order
to substantiate the findings, the most frequent causal conjunctions will now be
subjected to a more qualitative analysis.
4 Causal conjunctions II: pointers to non-native patterns

As the numerical data has shown, the four subcorpora follow a distinct
developmental pattern: an increase of causal conjunctions on the native
speakers', and a decrease on the learners' part. Yet such numerical patterns can
only be seen as meaningful if the features examined are indicative of a certain
type of communicative behaviour. If there was a connection between the figures
and factors such as age, gender or context of situation, for example, then the
respective linguistic features could be seen as stylistic variables.
When a whole functional class is concerned, as in the present case,
stylistic interpretations become more stringent, the more plausibly any
exceptions to the trend can be explained. Such an explanation was found for
because, which has been termed a 'wild-card use'. If a given linguistic element
is used as an all-purpose wild card, that usage is bound to include a number of
instances of over-extension. In other words, it can be expected that learners
may disregard target-language restrictions which are not that obvious, or even
accounted for in the standard grammars, but which are nevertheless observed by
the native speakers. Such 'simplification' is one of the most frequently cited
features of learner language.
In an attempt to uncover instances of simplification in the present learner
data, causal because was carefully re-evaluated. This analysis revealed an
interesting pattern in the conjunction's position: it is only in the learner corpora
that because (excluding prepositional because of) appears in sentence-initial
position (BWF 9.2, UNI 10.3 SF). The NNS usages with sentence-initial
because fall into two categories; here is a sample of the first one:
01 Because I'm very interested in that sort of job, I'd like to apply for it. BWF
02 Because I couldn't get the permission to visit my aunt at that late point of time
I had to drive home again which took me another six hours. UNI
03 Because they can't find a solution they form two smaller groups. BWF
At first glance, these sentences appear to be perfectly ordinary

constructions, connecting CAUSE and EFFECT - in this order, and in one
syntagma. All the more surprising that the native speakers did not use that
construction once, or rather: they never used it with sentence-initial because.
For the second exclusively non-native pattern, compare:
04 There are millions of possibilities to follow a new trend and therein lies the
difficulty. Because everybody has to have a trend to be possessed with,
otherwise one doesn't fit into our up-to-date society. UNI
05 If you keep this in your minds you don't have to be ashamed if compared with
European students. Because you can use your reason to see the things all men
have in common. BWF
06 All countries on the whole world have only one environment and it's up to us,
to everybody of us, to help it to survive. Because if the environment doesn't
function any longer the exitus of men is just a matter of time. UNI
Unlike the first set, these examples do follow the unmarked sequence of
"EFFECT→CAUSE", but they do so in a fashion unusual for this type of
discourse. By breaking the sequence up into two sentences, the concept of
CAUSE appears to be added as if in an afterthought - which would more likely
occur in spontaneous speech than in formal writing. This interpretation, albeit
informed by merely a small section of learner usage, adds another dimension to
the data: NNS style, as evident in the use of because, is not only governed by
wild-card use, but also by register-mixing, i.e. by the interference of other, less
formal text-types.
62 Gunter Lorenz
This latter interpretation is corroborated by another set of the above

conjunctions which breaks the general trend, namely the "pron is why" type. It
has been remarked earlier that their higher number might be due to a
paraphrasing strategy on the learners' part. Yet in terms of register, an
alternative, or complementary, explanation arises: among all instances of that is
why - the only one which permits is-contraction - almost three quarters of NNS
uses were realised as that's why (20 out of 28 actual corpus occurrences). In the
younger native speakers' data, the contraction merely occurred twice, and was
even dropped entirely by the older native speakers (LOC). Consider:
07 Many travellers are really interested in learning new things and understanding
foreign cultures. That's why travel also helps to connect people from different
backgrounds. BWF
08 It is not great fun to watch a movie that is interrupted by commercials every 15
minutes. That's why I think television commercials should be banned. UNI
09 First of all, I want to say that politics, as it is something made by man, can't be
perfect! That's why it is no use to tell you that this guy and that person have
done something wrong. BWF
These extracts give an impression of the learners' style of argument. Note

how in 07 the conclusion introduced by that's why does not strictly follow from
the premise, and how in 08, the argument is purely impressionistic. Sample 09,
moreover, contains several explicit signals of spoken, rather than written
argument.
To sum up, a more qualitative look at the use of causal conjunctions
shows that where the general numerical tendency is broken, plausible
explanations can be found. These concern a) the learners' inclination towards
wild-card use, b) their tendency to paraphrase, and c) a markedly NNS
treatment of register.
5 Causal adverbs and prepositions

In probing into other classes of causal markers, the main questions will be firstly
whether or not the overall numerical trend continues to hold, and secondly
whether the qualitative explanations can be further substantiated.
The sum total of causal adverbs, again, show an increase in native and a
decrease in non-native use. The former is even more pronounced than was the
case for causal conjunctions, but the NNS drop is only a very slight one (from
BWF 248.3 to UNI 240.3 SF).
Table 2: standardised frequencies of causal adverbs
Adverbs BWF UNI GCE ' LOC f
So 130.9 167.0 186.5 102.2

Therefore 64.0 45.0 40.3 149.5
Then 1.4 3.8 17.9
Thus 40.8 24.4 15.0 74.8
Hence 4.2 1.3 0.9 17.9
Consequently 6.3 2.6 0.9 10.5
Accordingly 0.7 1.1
Σ (SF) 248.3 240.3 247.4 373.9 ||
The alleviation of non-native decrease, however, is easily explained; it is

due to the increasing resultative use of so, which counteracts the overall trend in
both the learners' and the native speakers' data. Its sharp decline in the native
speakers' data provides a likely clue: causal so is not only the most frequent, but
also the least formal of all the items on the list, and it is clearly a dispreferred
feature in more mature argumentative writing. Consider:
10 Today many people have to live with haste and excitement, they have a hard
job, they do not earn much money, so it is understandable that they do not like
it even to work at home. BWF
11 But I think not only history is important but also the present time. So in my
opinion novels of modern writers should also be included in the curriculum.
BWF
12 However, no war is the same, as they have all different causes and objectives.
So, one should ask oneself the question, whether a war is justified or whether
peace is a good thing at any price. UNI
Leaving aside the linguistic and argumentative weaknesses, these

examples of causal so illustrate the learners' usage: the item is not only
exclusively used in clause-initial position, but as many as 61.9% of all instances
are placed at the beginning of the sentence (NS: 22.5%). In most cases, these
are very reminiscent of spoken-language discourse markers (see 11, and
particularly 12). This use hardly ever occurs in the native speaker corpora,
especially not in LOC.
Compare, by way of contrast, the corpus counts for therefore, thus or
hence in table 2; as these causal adverbs are generally characteristic of a rather
formal style of writing, their rising figures from GCE to LOC is best interpreted
as indicating an increasing text-type differentiation in native British English
between 16+ and university level. There is therefore mounting evidence that
64 Gunter Lorenz
text-type sensitivity does indeed lie at the heart of the NS/NNS numerical
contrast.
Table 3: standardised frequencies of causal prepositions
Prepositions BWF UNI GCE LOC

because of 31.0 48.8 38.4 49.5
due to 12.0 18.0 28.1 37.9
out of 0.7 1.3 0.9 9.5
owing to 2.1
Σ (SF) 43.7 68.1 67.4 99.0
Finally in the exploration of grammaticalised causal signals, these

prepositions introduce causal adverbial phrases typical of the nominal style in
formal written text. Unsurprisingly, therefore, there is another developmental
increase in the native speaker corpora. But for the first time, there is NS/NNS
agreement in the two developments; the learners, too, show a marked increase
in causal prepositional phrases.
Unfortunately, however, this seemingly desirable finding cannot be viewed
in isolation. The rising NNS use of because of for instance, is characterised by
markedly non-native patterns. Only the learners used sentence-initial Because of
this and Because of all this, which are more like ad hoc paraphrases of
Therefore than like central native prepositional usage. Furthermore, the
prepositional figures are not high enough to balance out the overall trend,
The prevailing tendency remains the same: there is a marked increase in
causal ties in the writing development of native speakers, which is,
unfortunately not emulated by the learners in the same way. On the contrary, all
grammaticalised causal markers add up to a steep NS rise from GCE 636.2 to
LOC 871.0 (SF), and to a sharp NNS drop from BWF 735.2 to UNI 591.2
(SF).
6 Extending the search: lexical causal patterns

The corpus searches have so far concentrated on well-established,
grammaticalised causal markers. Yet there are ways of marking CAUSE or
EFFECT which this procedure does not account for. It would be plausible to
assume, for instance, that the learners, for want of conventional lexico-
grammatical means, occasionally revert to more open-choice, maybe even
somewhat roundabout, strategies. Where this is the case, i.e. where reason or
consequence are signalled by means other than lexico-grammatical, they must be

expressed lexically. Most notably, this can be achieved by conjoining CAUSE and
EFFECT in a verbal construction (13 and 14), or by explicitly 'labelling' CAUSE
or EFFECT with an appropriate noun (15 and 16):9
13 Furthermore, let me add the fact that mass tourism leads to more understanding
for strangers on one's own country. BWF
14 A very high percentage of all accidents is caused by really unreasonable
drivers, who feel much too good in their cars. UNI
15 A really visible reason for the emancipated woman being alive is the high rate
of unmarried or divorced women. UNI
16 By watching frequently such discussions the citizens get to know the principle
aims and the guidelines of each party and as a consequence they can take part
in elections in a reasonable way. BWF
This small sample already shows that the learners use lexical causal
strategies in various ways. And despite the fact that quotes 13 to 16 contain
several instances of non-nativeness, the causal markers themselves hardly
appear to be chosen as mere substitutes for grammaticalised connectors. In
many ways, patterns such as "A leads to B" or "C is caused by D" connect
CAUSE and EFFECT in a far more direct and literal manner than an adverb or
conjunction would, for example. The same is true for explicit labellings such as
reason or consequence. It would therefore be imprudent to expect that lexical
markers should turn out to be an exclusively non-native domain. On the
contrary, as such literal reasoning puts more emphasis on the causal relationship
itself, native writers may well make copious use of it in the context of
persuasive argument. Compare:
17 The British tend to be very proud and possessive of the law, and changes may
well cause illfeeling and discontent. LOC
18 Three have died since 1986 in Britain alone but these deaths may have been
caused by "violation of the rules". GCE
19 The reason for such a large sixth form is our "open entry policy, which gives
anyone with the right attitude a chance to do A-levels". GCE
20 The consequence for business is that it has had to adapt its production away
from baby and children's products, to those required by the older generations,
in particular medical supplies. LOC
These samples from native writing are, of course, stylistically smoother

than the non-native ones. But there is no way of predicting which of the two
populations is more prone to lexical causal marking.
66 Gunter Lorenz
The corpus searches for the next section draw on the hypothesis that
lexical causal patterns will inevitably contain a variant of the lexical field of
CAUSE or EFFECT. Retrieval was therefore effected with the help of dictionaries
and thesauruses. As in the case of grammaticalised items, all elements have been
viewed and evaluated in context.
7 Lexical strategies, semantic prosody and more non-native

patterns
On searching the corpus, the figures immediately show that verbal causal
marking can not compensate for the numerical differences noted earlier;
whichever way they were to lean, they are far too low to account for an
alternative learners' strategy.
Table 4: verbal realisations of causal relations
Causal Verbs BWF UNI GCE LOC

cause (active) 9.2 28.3 32.8 25.3
lead to 2.8 2.6 6.6 12.6
result in 2.1 1.3 2.8 14.7
evoke (active) 3.5 1.3 1.9 13.7
cause (passive) 5.6 18.0 4.7 2.1

result from 3.5 — 0.9 4.2
evoke (passive) 1.4 — — 4,2
come from 0.7 — 0.9 3.2
derive from 0.7 — 0.9 2.1
stem from 0.7 1.3 — 2.1
\ Σ (SF) 30.2 52.8 51.5 84.2
The frequency profile resembles that of causal prepositions above (cf.

table 3): the rise from GCE to LOC corroborates the tendency of a native
increase in causal marking, while the non-native development from BWF to
UNI follows suit, if lower in number. In this case, the NNS rise is entirely due
to the older learners' pronounced preference for high-frequency items (cf. cause
active + passive). The causal verbs fall into two categories according to their
respective transitivity, i.e. whether they follow the pattern of "CAUSE leads to
EFFECT", or that of "EFFECT results from CAUSE". In each set, the most frequent
item contravenes the general trend. In other words, if the UNI learners did not
make as much use of cause (active & passive), the causal verbs would perfectly
comply with the general tendency. Unlike causal so above, this is not likely to
be a question of text-type or register, but rather of another wild-card overuse on
the older learners' part.
As we have seen in the case of because, non-native wild-card use may
well lead to the violation of target-language regularities. It is therefore not very
surprising that the same was true for the learners' use of cause. To understand
the nature of the NNS irregularity in question, we must take recourse to a
concept which has recently become associated with the term 'semantic
prosody'. It has repeatedly been noted that cause - despite its seemingly
neutral denotation - has a certain negative undertone and a strong association
with undesirable events or states of affairs. Native speakers are much more
likely to say that X caused a riot than to say that ?Y caused peace and quiet. In
cases where apparently neutral words systematically cooccur with markedly
positive or negative concepts, we speak of a positive or negative semantic
prosody.
The negative semantic prosody of cause has been validated by a 120
million word corpus (Stubbs 1995). Yet even the present, comparatively small
corpora produce unequivocal results: the most frequent noun cause collocates,
for example, are damage [10 occurrences], problem(s) [9], accident(s) [7], and
death(s) [7] - doubtless all unpleasant concepts. In the few cases where the
collocates carry positive connotations (joy, gain) or neutral ones (reaction,
society), the context makes it clear that the negative bias still holds:
21 I in particular feel that members of Parliament are unable to relate to the

problems of society, such as homelessness and unemployment (...). Until such
times the normal human being will feel inadequate and this will cause a two-tier
society of Chief and Indians. GCE
22 This essay has illustrated how fashion is not simply an ally or just a tyrant, but
that it is both. Fashion can cause both joy and suffering, gain and loss. GCE
23 The occasional infringement and violation of the constitution causes little
reaction from the French. The French have never had much respect for the
prevailing constitution. LOC
In 21, society is further modified so as to indicate an undesirable

development; in 22, the positive concepts of joy and gain are juxtaposed with
suffering and loss; and in 23, a desirable reaction is almost negated by little -
resulting in fierce criticism.
The learners mostly comply with the negative semantic prosody of cause;
occasionally, however, it is disregarded:
68 Gunter Lorenz
24 What does the American Dream mean for you? Is it only the cause of your
comfortable existence, the reason why you can feel pride in your country?
BWF
25 I will cause the rise of human power and dignity by giving man a help. An
inferior. No human intelligence, feeling or creativity will ever be replaced by
any kind of circuitry. BWF
26 Nowadays women without a husband very often cause respect, especially if
they have a good position in their job. UNI
Regardless of how serious or slight a violation each of these uses may

appear - with quote 24 implying an element of sarcasm (see Louw 1993), 25
lending a certain insincerity to the pathos of the argument, and 26 probably
being the least acceptable -, all three are instances of NNS usage violating a
target-language regularity. It is such stylistic side-effects of 'lexical teddy-bears'
which cast doubt on their wide and easy applicability. In a final retrieval, it
remains to be seen whether this also applies to nouns as causal markers.
Table 5: nominal realisations of causal relations
Causal Nouns BWF UNI GCE LOC |
Reason 52.8 72.0 39.4 36.9

cause 3.5 5.1 2.8 7.4
motive 2.8 3.9 0.9 12.6
motivation 2.1 2.1
effect 21.1 23.1 21.6 34.8

result 19.7 10.3 11.2 25.3
consequence 32.4 23.1 1.9 19.0
outcome 5.3 J
Σ (SF) 134.4 137.5 77.8 143.4 f
As with causal verbs, the noun figures confirm that lexical causal patterns
cannot compensate for the NNS/NS differences in grammaticalised items - at
least not in terms of frequency. As the NS increase shows, causal labelling is not
an indicator of linguistic immaturity; on the contrary, the counts for the native
speakers continue to rise with age, and the increase is even stronger than before.
The learners' frequency counts show a slight rise from BWF (134.4 SF)
to UNI (137.5 SF), which is, however, too low to offer any compelling insights.
The NNS total is, for once, higher than the NS one (271.9 vs. 221.2 SF), and
again this is due to a marked overuse of the most frequent element. Among the
eight nouns under scrutiny, reason not only has the highest counts, but is also
an exception to the overall trend: its numbers show a substantial non-native rise,
and a small drop from GCE to LOC. As the word itself is not marked for
register, the figures suggest another instance of NNS wild-card use. As such, it
is - naturally - far less needed on the native speakers' part, but increasingly
relied on in learners' writing.
Finally, and in keeping with the line of inquiry so far, reason is checked
for patterns of non-native usage. Two different ones emerge:
27 Another reason why I consider the job as a newscaster to be attractive is that I

enjoy working with the voice. BWF
28 One reason also is the fact that politics are often performed without any
context. BWF
29 Our government has decided that it should be the population's choice if the
project is to be accepted. For this reason, you will be able to vote in two
months time. BWF
30 The watcher has no conscious control over the effects TV-commercials exert on
him and he cannot escape them except by switching off the TV. For that reason
the spots should be banned. UNI
Extracts 27 and 28 are examples of sentence-initial A/Another/One

reason... is..., which occurs nowhere in the native speakers' writing. In
sentences such as 27, where the subject noun phrase is elaborately modified to
include the concept of EFFECT, the sentence's theme is made awkwardly heavy
or "packed".11 And in constructions like 28, merely adding another reason to a
chain of arguments, the pattern lends the learners' discourse a certain staccato
like, enumerative quality. Both stylistic impressions may occasionally also arise
from samples of native writing, but the point here is that the inducing patterns
did not occur in the NS corpora.
The pattern For this/that reason (29 and 30) is more of a phrasal
building-block than an open-choice use of reason. In this respect, it is similar to
sentence-initial Because of this above. Both may be literal translations of
German Aus diesem Grund, resulting in markedly non-native causal patterns.
In sum, the individual overuse of reason, the most frequent causal noun,
has once more pointed to patterns of non-native usage. This is not to say that
this kind of procedure will bring up all such non-native patterns, or that they are
necessarily to be treated as undesirable - or even unacceptable constructions.
The analysis so far has simply tried to take stock of the various means and
preferences in expressing causal relations, without tackling the issue of
acceptability and "preferred ways of putting things" (Kennedy 1992).
70 Gunter Lorenz
8 Summing up: idiomaticity and non-native deviancies

The analysis of causal markers has thrown up a number of interesting results. It
remains to be discussed a) what kinds of conclusions may be drawn from the
corpus totals, and b) what the implications are in terms of markedly non-native
constructions. But first the various types of markers can finally be added up to
enable an overall numerical comparison.
Table 6: sum total of causal markers
Causal BWF UNI GCE LOC

Connectors (SF) (SF) (SF) (SF)
Conjunctions 443.2 282.8 321.4 398.1
Adverbs 248.3 240.3 247.4 373.9
Prepositions 43.7 68.1 67.4 99.0
Σ1 735.2 591.2 636.2 871.0
Causal Verbs 30.2 52.8 51.5 84.2
Causal Nouns 134.4 137.4 77.8 143.4
Σ2 164.6 190.3 129.3 227.6
Σ3 899.8 781.2 765.5 1098.6
The general tendency that can be gleaned from the distribution of

grammaticalised causal markers (Σ1) is as follows: there is a significant
developmental increase from the younger to the older native speakers
(x2 =36.371). Furthermore, as both groups of native speakers taken together
used more such items than the learners did (NS 1507.2 vs. NNS 1326.4 SF;
x2 =11.483), there seems to be a positive correlation between causal marking
and stylistic maturity in argumentative writing. Unfortunately, however, the
learners' development does not point in the same direction; the figures decrease
from 16+ to university level(x2=15.535).
The same trend does not hold true for the lexical subset (S2); there is
hardly any NS/NNS contrast here (NS 356.9 vs. NNS 354.9 SF; x2=0.006!),
and the two populations' developments go in the same direction, albeit much
more steeply on the native speakers' part. Numerically, therefore, the lexical
patterns are not very indicative. Yet their corpus counts are not high enough to
influence the sum totals (S3) significantly: the overall tendency remains the same
as outlined for Σ1.
If we can indeed conclude from the figures that causal marking positively
correlates with mature argumentative style, as hinted above, then this has
serious ramifications for the interpretation of the learners' developmental
decline. In simple terms, it would imply that the older learners' writing must
either be considered less mature, or "less argumentative" than that of their
younger counterparts. The former of these conclusions is not very likely; if it
were true, this would mean that the learning and teaching efforts between the
two stages had been counter-productive - at least for the small section of causal
marking. Fortunately, the second surmise appears to be far more plausible. The
high UNI figures for causal so and that's why indicate a manifest change in the
learners' style of writing: when asked to argue a case, the older learners adopt a
more casual, conversational register, shifting away from the more formal
school-type written argument. This shift of style, however, is not consistent; as
some features of formal writing are still prominent - see the high counts for
prepositional markers - it makes for a striking mix of registers.
In trying to explain the differences in native and non-native stylistic
development, we should consider the different points of departure. Whereas
adolescent native writers naturally have no problems with casual English
idiomaticity, it is their written register which is still developing. Advanced
German learners of English, on the other hand, are used to text-based argument,
modelled on the style of quality weeklies. In attempting to emulate such
journalistic prose, they resort to the structural signals typical of that type of
discourse - without, of course, having the corresponding lexical means and
argumentative skills. If the younger learners therefore exhibit a high number of
causal connectives, higher even than their native-speaking counterparts, this
may first of all be due to the educational fallacy of trying to get adolescents to
write like professional journalists - in a foreign language.
This puts the older learners' data in perspective: university students of
English are, of course, far more prone to register-mixing than school pupils,
simply because they have gradually become exposed to a wide variety of written
and spoken styles. The seeming deterioration in the learners' development is
hence most plausibly interpreted as the deconstruction of a didactic
misconception. The older learners' interlanguage contains both elements of
casual speech and formal writing, but the learners lack the intuitions which
guide a native speaker's usage.
The pedagogical consequence of this study therefore cannot be a mere
quantitative one: teachers would be ill-advised to simply try and increase the
number of connective devices in their students' writing. There is, regrettably, no
quick access to linguistic intuitions, particularly so in the finer points of usage,
72 Gunter Lorenz
such as register and style. A potential, if laborious, way of approaching the

problem lies in another finding of this study: where a single high-frequency
lexical item contravened the numerical trend, its use contained one or more
distinctive patterns of non-native usage. These include sentence-initial Because,
as well as Because of this,..., For this reason,..., One reason...is..., and the
violation of the negative semantic prosody of cause. It can be hoped that by
carefully exploring such patterns of non-nativeness, and in a long process from
language description to teaching, learners' productions will gradually get to
resemble more those of competent native speakers.
In the discussion which followed this paper at the Augsburg workshop, it
was remarked that it may not necessarily be desirable for language learners to
emulate the verbal behaviour of native speakers. It was pointed out that, in the
context of English as an international language, native speaker norms will and
should lose their monopoly of English usage. The present study, however, is not
at odds with the rationale of international English. It is interlanguage English
which almost world-wide acts as working means of communication. And it may
even be advisable, in very restricted functional domains, to lower and broaden
the teaching objectives to suit the respective real-world communicative
demands. Yet for tertiary pedagogical purposes, we cannot resort to teaching a
variety of English that is in any way 'reduced' or rudimentary. For future
teachers of English - and more than 80% of our students study English in order
to teach it - the primary goal must be to close the gap between their own
command of English and what it takes to speak and write English idiomatically.
Idiomaticity is a probabilistic concept; it hinges on what kinds of
structures are most likely to occur. If a given utterance complies with the basic
rules of English as codified, it may still be seen as unidiomatic - or "un-English"
- by native speakers. Such patterns of non-nativeness systematically violate a
reader's and hearer's expectations, and the effect of such violation is a
cumulative one which may not only lead to estrangement, but may ultimately
also endanger mutual understanding.
This study makes no claims as to the acceptability of a given structure. It
does not mean to suggest that a pattern which is exclusive to learner language is
necessarily not part of the target language, or even altogether faulty.
Acceptability judgements will always remain a native speakers' prerogative. But
in many cases the systematic character of idiomaticity is not even realised by
native speakers. In the present case, non-native causal markers may perform a
dual role in the construction of coherence: on top of their cohesive function,
they may help create a sub-discourse of non-nativeness. And where non-
nativeness is - even subconsciously - perceived as clumsiness, it is counter
productive to the demands of coherent argument. If we want to help advanced
learners of English in their task, we must provide them with a thorough

description of how their own usage differs from that of native speakers.
Notes
1. The adjective causal is used here to refer to items which mark either CAUSE or EFFECT; in
either case, the logical relationship is clearly a causal one.
2. The four sub-corpora are part of an ongoing learner corpus project conducted by the
author; for more details see also Lorenz (to appear). While the corpus is still growing in
size, the word counts given here represent the stage of corpus search for this
investigation (early 1997).
3. No instance of in so far as or insofar as found in the corpus.
4. Note how this corresponds with the research findings quoted above. Within the
International Corpus of Learner English project, highly overused high-frequency items
have come to be known as 'lexical teddy-bears' - comfortably familiar and easy to cling
to. The term was coined by Angela Hasselgren, who has worked with Norwegian
learners' data at Bergen.
5. There was one exclusively native pattern which resembles that of quotes 04 to 06: where
the native speakers felt the need to add a reason in a single clause, they introduced it by
this is because [15 occurrences], thereby creating a full-fledged main clause, as opposed
to the learners' stand-alone adverbial one. Once more, it is surprising that the learners
did not resort to that pattern once.
6. Compare Schleppegrell (1996a, 1996b), who asserts that the use of because as such is far
more common in spoken than in written discourse.
7. For the purpose of this paper, such stylistic ratings are based on the information given in
monolingual dictionaries and usage-oriented handbooks (see references).
8. This term was coined by John Sinclair; see (1991:109ff) for 'open-choice' versus 'idiom'
principle.
9. The concept of textual 'label' has been borrowed from Francis (1986, 1994), the idea
being that a certain class of nouns can be used as identifying tags for preceding or
ensuing stretches of discourse. These tags can work within or across sentence boundaries.
In the first sentence of this footnote, for example, the noun idea can be seen as a label for
the whole ensuing that-clause.
10. See, e.g., Louw (1993), Stubbs (1995), Bublitz (1996).
11. Compare Ventola (1996), as well as Lorenz (1998).
12. It should be added here that in two cases the native speakers did employ a parallel
construction - namely It is for this reason that (both LOC). This is similar to what was
noted in footnote 5: two separate patterns, an NNS and an NS one. And as with This is
because, where the native speakers used a more elaborate focusing pattern, the learners
74 Gunter Lorenz
tried to make do with a blunter construction...
References
Bublitz, W. 1996. "Semantic prosody and cohesive company: somewhat predictable".
Leuvense Bijdragen 85: 1-32.
Connor, U. 1984. "A study of cohesion and coherence in English as a second language
students' writing". Papers in Linguistics 17: 301-316.
Crewe, W.J. 1990. 'The illogic of logical connectives". ELT Journal 44: 316-325.
Ehrlich, S. 1988. "Cohesive devices and discourse competence". World Englishes 7:
111-118.
Evenson, L.S. and Lintermann-Rygh, I. 1988. "Connecting L1 and FL in discourse-
level performance analysis". Papers and Studies in Contrasfive Linguistics 22:
133-178.
Field, Y. and Yip, L. 1992. "A comparison of internal cohesive conjunction in the
English essay writing of Cantonese speakers and native speakers of English".
RELC Journal 23: 15-28.
Francis, G. 1986. Anaphoric Nouns. Birmingham: University of Birmingham.
Francis, G. 1994. "Labelling discourse: an aspect of nominal group lexical cohesion".
In M. Coulthard (ed), Advances in Written Text Analysis. London: Routledge,
83-101.
Granger, S. and Tyson, S. 1996. "Connector usage in the English essay writing of
native and non-native EFL speakers of English". World Englishes 15: 17-27.
Halliday, M.A.K. and Hasan, R. 1976. Cohesion in English. London: Longman.
Kennedy, G. 1992. "Preferred ways of putting things with implications for language
teaching". In J. Svartvik (ed), Directions in Corpus Linguistics. Berlin: de
Gruyter, 335-378.
Lintermann-Rygh, I. 1985. "Connector density - an indicator of essay quality?" Text 5:
347-357.
Lorenz, G. 1998. "Overstatement in advanced learners' writing: stylistic aspects of
adjective intensification". In S. Granger (ed), Learner English on Computer.
London: Addison Wesley Longman, 53-66.
Lorenz, G. 1999. Adjective Intensification - Learners versus Native Speakers: A
Corpus Study of Argumentative Writing. Amsterdam: Rodopi.
Louw, B. 1993. "Irony in the text or insincerity in the writer? The diagnostic potential
of semantic prosodies". In M. Baker, G. Francis and E. Tognini-Bonelli (eds),
Text and Technology. In Honour of John Sinclair. Amsterdam: Benjamins, 157-
175.
Milton, J. and Tsang, E. 1993. "A corpus-based study of logical connectors in EFL
students' writing". In R. Pemberton and E. Tsang (eds), Studies in Lexis. Hong
Kong: HKUST, 215-246.
Redeker, G. 1990. "Ideational and pragmatic markers of discourse structure". Journal

of Pragmatics 14: 367-381.
Sanders, T., Spooren, W. and Noordman, L. 1992. 'Toward a taxonomy of coherence
relations". Discourse Processes 15: 1-35.
Schleppegrell, M. 1996a. "Conjunction in spoken English and ESL writing". Applied
Linguistics 17: 271-285.
Schleppegrell, M. 1996b. "Strategies for discourse cohesion. Because in ESL writing".
Functions of Language 3: 235-254.
Seidlhofer, B. 1986. "Cohesion in Austrian learners' English". In O. Rauchbauer (ed),
A Yearbook of Studies in English Language and Literature 1985/86. Festschrift
für Siegfried Korninger, 221-236.
Sinclair, J. 1991. Corpus, Concordance, Collocation. Oxford: Oxford University
Press.
Stubbs, M. 1995. "Collocations and semantic profiles. On the cause of the trouble with
quantitative studies". Functions ofLanguage 2: 23-55.
Ventola, E. 1996. "Packing and unpacking of information in academic texts". In E.
Ventola and A. Mauranen (eds), Academic Writing. Intercultural and Textual
Issues. Amsterdam: Benjamins, 153-194.
Ventola, E. and Mauranen, A. 1991. "Non-native writing and native revising of
scientific articles". In E. Ventola (ed), Functional and Systemic Linguistics.
Approaches and Uses. Berlin: de Gruyter, 457-492.
Dictionaries, Grammar Books, Usage Guides:

Cambridge International Dictionary of English (1995), ed. by P. Procter et al,
Cambridge: Cambridge University Press.
Collins COBUILD English Dictionary (21995), ed. by J. McH. Sinclair et al, London:
HarperCollins.
Collins COBUILD English Grammar (1990), ed. by John McH. Sinclair et al,
London: HarperCollins.
Collins COBUILD English Usage (1992), ed. by John McH. Sinclair et al, London:
HarperCollins.
A Comprehensive Grammar of the English Language (1985), ed. by R. Quirk, S.
Greenbaum, G. Leech and J. Svartvik. London: Longman.
Longman Dictionary of Contemporary English ( 1987), ed. by D. Summers et al,
Harlow: Longman.
Longman Language Activator (1993), ed. by D. Summers et al., Harlow: Longman.
Oxford Advanced Learner's Dictionary of Current English ( 1995), ed. by A.P. Cowie
and J. Crowther, Oxford: Oxford University Press.
The Oxford Thesaurus. An A-Z Dictionary of Synonyms (1991), ed. by L. Urdang,
Oxford: Clarendon.
Practical English Usage (1980), ed. by M. Swan, Oxford: Oxford University Press.
Coherence through Understanding through Discourse
Patterns:
Focus on News Reports
Jan-Ola Östman
University of Helsinki
1 Aim
The purpose of this paper is threefold. I will begin by introducing the notion
discourse pattern as a technical term, relating it to cognitive aspects of
language use. Next, I will show how coherence can be viewed in terms of this
notion, and de facto in terms of understanding in general. Finally, I want to
illustrate the usefulness of the notion discourse pattern in discourse analysis at
large, through an analysis of news reports in a number of US and UK
newspapers on the meeting between presidents Clinton and Yeltsin in Helsinki
in March, 1997.
2 Genres and text types - coherence and cohesion

The traditional distinction between coherence as a cover term "for all kinds of
'semantico-functionaP phenomena which collaborate" to constitute what is
definitional about text and discourse (Östman 1978: 102), and cohesion as
referring to textual tightness as manifested by structural means, has remained
useful as a heuristic, even though the give-and-take between form and function
has turned out to be much more complicated - and much more intricate - than
we could envisage twenty years ago.
If- for the moment - we leave aside the extreme (although exceedingly
relevant, but ultimately unhelpful) view that any text or discourse can be
coherent and/or incoherent depending on context, we note that one of the first
steps towards concretely tackling issues of cohesion and coherence at large1
was to specify them in relation to genre and text type. Discourse is coherent in
relation to context in terms of specific cultural conventions, and in terms of the
purpose for which that discourse is produced and understood.
78 Jan-Ola Östman
Both genre and text type are clearly important notions, as is the need to
differentiate between the two. Recent work both in linguistics and in literary
studies has indicated the complexity of the notions genre and text type; e.g.,
they have to be related to social practices, to intertextuality, and to the
processual construal of discourse. Nevertheless, genre and text type are very
often seen to form a dichotomy, as being two perspectives on discourse: genre
zooms in on the external relations that a text/discourse displays, and text type
focuses on its internal relations.
It is worth noting that when these two approaches are taken
simultaneously, they are, metaphorically speaking, supposed to meet half-way,
and in this manner circumscribe and define das Ding an sich - the coherence -
of the text/discourse. When the starting point for an analysis of a text/discourse
is a categorization in terms of a particular genre (machine-readable corpora like
the LOB and Brown corpora are typically organized in terms of genre), the
next step is to minutely describe the kinds of structures, phrases and words that
can be found in each genre - both very locally, within sentences and clauses,
and more globally, e.g., with respect to how texts of a particular genre start or
end. On the basis of such analyses we can then find linguistic similarities and
differences between the various genres that we had set up on an intuitive basis
(but which, indeed, have been codified gradually through socialization). Thus,
we get information about the manner, degree and range in accordance with
which discourse, and consequently coherence, can vary.
However, when the starting point - and initial focus of interest - is, say,
a narrative text type, linguistically defined as, among other things, having its
foregrounded units (sentences, utterances, information units) being
sequentially ordered so as to correspond to the order in which the events
described took place in 'real life,' the logical concomitant step is to specify the
differing characteristics of this particular text type as against other text types.
The typical next step is (cf. e.g. Werlich 1976) to generalize from here:
grouping together different kinds of narratives, one eventually ends up with
what Werlich called text form variants - which, for all intents and purposes,
are what others call genres; e.g., a fairy tale is a narrative text form variant.
Schematically, the two approaches can be displayed as in tables 1 and 2.
Genre and text type differences clearly need to be taken into account in
any linguistic analysis, e.g., analyses of language change is extremely sensible
to genre and text type differences. Still, despite the fact that genre studies and
text type studies have a very similar concern, it is also important to keep in
mind that genres and text types are not the same. A genre like 'novel' can be
written in the form of a narrative text type, but it can also be in the form of,
say, an argumentative or exploratory text type. A novel like Julian Barnes's
Coherence and Discourse Patterns 79
Table 1. 'Approach 1 ': Starting with genre.
Genre (most general level) Story
Sub-genres fairy tale
Special discourse Once upon 'in medias time &
Characteristics a time... res' place setting
initially
Table 2. 'Approach 2 ': Starting with text type.
Text type Narrative
Text forms reporting
Text form variants Story Joke Anecdote

Flaubert's Parrot contains several different text types embedded in the
global narrative text type which takes the story forward. Similarly, narrative
text types are not restricted to fairy tales and novels. For instance, a genre like
place descriptions can be done in a narrative manner, as can a horoscope text.
Genre and text type are clearly two approaches we can take on texts and
discourse.
However, these two perspectives, and concomitant research approaches,
to what is constitutive of a text, are alike not only in that they are mirror
images of each other, but also in that they presuppose their object of study. The
definitions they suggest are negative definitions. Discourse as a phenomenon is
not approached from the point of view of it having a value on its own, as
discourse, but as something that needs to be defined with the help of other,
supposedly better known concepts.
In the light of recent advances in the study of discourse (cf. e.g. Goodwin
and Duranti 1992; Auer 1995; Kuusisto and Östman 1998) of how discourse
80 Jan-Ola Östman
construes its own context, of how discourse not only reflects, but creates
situations, a definition of discourse and its coherence cannot be adequately
given in the traditional terms presented above.
3 Introducing the notion discourse pattern

The internal-external distinction on discourse level mirrors the form-function
distinction traditionally made on sentence level. Schematically:
form/internal >
< function/external
Figure 1. External and internal approaches as two perspectives.
The picture in figure 1 in fact implies that there are two toolboxes needed
for linguistic analysis. My view is, however, that these two complementary
approaches are not enough. We need a third toolbox. In very much the same
way as I think we need a 'meaning' filter 'in-between' form and function, we
also need a filter 'in-between' genre and text-type descriptions. This third
toolbox I will call a discourse pattern description.
There are several, even purely heuristic reasons for recognizing such a
level of analysis. For instance, I would claim that it is not enough to
characterize a newspaper report (= a genre description) as a narrative text type,
since it is doubtful whether readers and interactants understand a news report
primarily in such terms. What analyses in terms of genre and text type lack is
reference to cognitive aspects like understanding. Schematically, the view
advocated in the present paper can be presented as a three-way, rather than as a
two-way division where 'discourse pattern' mediates between 'text type' and
'genre' in a parallel fashion to the way semantics can be said to mediate
between Syntax and Pragmatics:
Table 3. The Morrisian (1938) three-way distinction extended to

discourse.
form meaning function
Phonology, Semantics Pragmatics

Morphology, Syntax
- the structural - the cognitive - the socio-cultural
text type discourse pattern genre
As we already saw in the preceding section, one major problem with

characterizations of text/discourse in terms of genres and text types is that they
constitute abstract frames of references in terms of which language use, and
language understanding, can be supposed to take place. This is both the
strength of these approaches and their major weakness, since there is usually a
very small set of, say, text types to choose from (genres and sub-genres one
can add at will, but here, too, researchers who have attempted to systematicize
them have usually ended up with a fairly small inventory). That is, they make
up a fairly neat system, but if you want to use them in an actual analysis, you
will have to talk in terms of how a particular piece of discourse does not fit the
ready-made frame of reference.
Defining behaviour negatively is not a waste of time, but it clearly cannot
give the whole truth. A positive characterization has to start from the way
participants and interactants conceptualize discourse. This is also what
coherence in its all-embracing discourse-constitutive sense is all about. My
claim is that conceptualisation on discourse level primarily takes place in terms
of discourse patterns rather than (or, at least, in addition to) in terms of genres
and text types.
4 Coherence in terms of understanding

Searching for a characterization of discourse in terms of how interactants
conceptualize it, in effect means that we are looking for what makes a piece of
discourse a coherent whole to the interactants. That is, textual and discoursal
coherence has to be seen in terms of the socio-cognitive understanding which
holds texts and discourses together for the discourse producer(s) and the
addressee and audience - i.e. for the members of a particular culture or
community.
Thus, when somebody talks about, say, a recipe, the first thing that
comes to mind (in our culture), in terms of which we most readily, most
immediately, and most efficiently understand the notion of a recipe is the
visual, graphic display that looks something like this:
12x fdjkfjfd
23nn lkfdlkjf sdjklfdsjkl
34fr kjdfsjklfdiop
asälkfj oksd jfdkfj dfjfj dsklfj dfjsd fjsdfjfjf
söä df kskflsdkfd kf kdsfkdsf kdfkd fkdfkdsfl
öls kdflösd kfsdl kfsfksfk s dlfk sdfksäd fsdl
82 Jan-Ola Östman
sö ldkfs dlkflösdkföl s fks dlfksd 1 fksdl kfsd

lsdk fsdk flsd kfsdfksd k fsdfk dslfk lfkds fk
Figure 2. A first approximation of the (visual) schematization of the

Recipe image.
We have an image of the shape2 a recipe prototypically takes: Certain

specifications about measures (cf. the numbers in figure 2) in a list-like
fashion, followed by a text depicting the different steps to be taken in a
particular, incremental order (i.e., a prototypical recipe is iconic, it goes
through the steps in the order they should be performed; a prototypical recipe
does not end with "Before all this, you should have kept the dough in the fridge
3 hours.")
Understanding and categorising a recipe as a recipe, I suggest, does not
primarily take place in terms of whether the text is narrative, descriptive or
instructive - i.e. in terms of the text type it represents; nor is the
conceptualisation and codification primarily in terms of the activities that are
taking place in a kitchen, like why one often needs a text/discourse when
preparing food. The point is not that these aspects play no role at all in
conceptualisation, but that neither of them by itself, nor taken together, give a
full and satisfactory account. The most noticeable feature of recipes is,
concretely, the graphical layout they conform to; but this layout, due to its
regular occurrence, has been crystallized as a code on.
The shape of recipes is very pervasive. Even on the Internet, where there
would be infinite possibilities to vary the presentation of cooking instructions,
the traditional way of displaying recipes is used. The recipe in text 1 is from
http://members.access.net/ ~shuler/recipe.htm:
Country Cookout Recipe of the Month
Jack Daniels Sweet Potatoes
Great with pork or venison roast
Ingredients
5 or 6 medium to large sweet potatoes

1 cup Jack Daniels sippin' whiskey
2 cups white sugar
1 cup brown sugar
Preparation
Peel the potatoes and cut into 1 inch thick slices

Place the potatoes in a large dutch oven
Pour the whiskey all over the potatoes
Add the white and brown sugar
Do not stir and do not add water
Simmer over medium heat, covered for 1 hour
Remove the lid, lower the heat to low and cook for another 2 hours *
*sometimes I occasionally baste the potatoes with the liquid using a turkey baster
Text 1.
The pattern associated with text/discourse - like that in figure 2 for

recipes - could for instance be explicated in terms of what Lakoff (1987) calls
a prototypical Idealized Cognitive Model (ICM). My contention is that such
patterns constitute an additional tool in understanding and using texts. It is the
aspect that ultimately gives the text its perceived and (variously) codified
coherence. The schematization given in figure 3 focuses entirely on the
prototypical physical shape of recipes, and will thus be termed the Recipe
pattern. This pattern is constitutive of the inherent coherence of recipes in
general. If cooking instructions are presented in some other manner - as they,
indeed, sometimes are - more processing work will be demanded by the reader
or addressee for him/her to understand and conceptualize such non-
prototypical cooking instructions as a recipe.
Figure 3. The Recipe pattern.
5 Coherence in terms of discourse patterns

So far I have suggested that discourse patterns are cognitive schemata in
accordance with which a text or discourse is characterized, in terms of which it
84 Jan-Ola Östman
receives coherence, and in terms of which texts and discourse are readily
interpretable and understandable (and regularly interpreted and understood) as
being of that particular kind.
If coherence is seen in relation to cognition and understanding, in terms
of how particular interactants grasp and deal with discourse, the status of the
universality of coherence becomes an issue. In the case of the Recipe pattern,
we have a pattern that is very frequent in very many literate cultures. However,
if we look at another frequently encountered pattern, for instance, that of Death
notices, we find at once that any attempt at a 'universal' schematization of
these will be extremely different. Even neighboring communities and cultures
with close contacts use very differently codified Death notice patterns.
For instance, whereas in Britain, deaths are recorded as text, in Finland
deaths are advertised much more graphically. Still, both are patterned. Fries
(1990) has shown that British death notices like those under (1) have a typical
structure with particular slots of information in a particular order, cf. figure 4,
where both the general and a more extensive structure is depicted.
Died a few days ago, at his house in Greenwich, Capt. Robert Walter, of the Royal
Navy.
In Dublin, the Honourable Miss Isabella Howard, second daughter to the Right
Hon. Lord Clonmore.
Text 2.
DATE PLACE NAME RELATION OCCUPATION
DATE PLACE AGE CIRCUMSTANCES NAME RELATION ORIGIN OCCUPATION Other

info
Figure 4. The structure of British death notices.
In Finnish, death notices have a very particular graphic display, with a

cross (or a symbol for some other religion) to the left, the dead person's full
name in bold face, the specific dates and places of birth and death, names of
people who mourn, very often a psalm and finally the specification of where
the funeral will take place. An example is given in text 3.
Rakkaamme, insinööri
Pasi Antero
Ailio
s. 9. 7. 1951 Kajaanissa
k. 7. 10. 1996 Naantalissa vaikean
sairauden murtamana
Kaipaamaan jäivät
Marja-Helena. Heini ja Henri
Lea-āitî
Muut sukulaiset ja lukuisat ystâvàt
Sinun suunnaton rakkautesi ja huolenpitosi.

Rakas Pasi auttaa minua jaksamaan ja
jatkamaan meidän yhdessā aloitettua
purjehdustamme nyt elāmān vesillä.
Yhteistssä lapsissamme Heinissä ja Hcnnssa
olet aina oleva rinnallani.
Odota minua siellä jossain
kiitan yhteisistä ihanista vuosista Rakas
Marrusi
Jumalalle kiitos poikani elämäslä1

Äiti
Siunaustilaisuus Naanlalin kirkossa lauantaina 26.10.1996 klo 1Ü.00.
Sen jälkeen muistotilaisuus Naantalin Merisalissa. jonne lämpimas-
ti tervetuloa.
Text 3. A Finnish death notice.
A Finn might not off-hand recognize, nor even understand or believe that
a British death notice is a proper death notice. For a death notice to be coherent
as a death notice to him or her it would need to have the structure, the pattern,
of a Finnish death notice.
The Recipe pattern and the Death notice patterns are strictly codified
patterns - relative to particular cultures. I want to suggest, though, that the
usefulness of the notion of discourse pattern for understanding coherence is not
restricted to formulaic texts.
One very pertinent question that needs to be raised is what linguistic
evidence there is for suggesting that we need the concept discourse pattern. I
will show this by providing minimal pairs in a tabular form (cf. table 4), and
indicating that the notions genre and text type are not enough.
The examples in table 4 indicate, first of all, that there is no one-to-one
relationship between genres and text types: both fairy tales and novels often
utilize a narrative text type; a novel can be in narrative form, or it can, e.g., be
an epistolary novel in an instructive format.
86 Jan-Ola Östman
Table 4. Minimal pairs.
GENRE TEXT TYPE3 DISCOURSE

PATTERN
fairy tale narrative Aesthetic

novel narrative Aesthetic
novel instructive Aesthetic
advertisement argumentative Ad
contact ads descriptive Human interest
contact ads descriptive Plot
contact ads (FIN) descriptive Ad
The three first text examples - rows - in table 4 are all specified as
'Aesthetic' under discourse pattern. Without going into details about this
pattern here4, suffice it to say that it is built on the notion of the Golden
Section, which is a very prevalent structure in (traditional) fiction.
The last three examples in table 4 all utilize the descriptive text type, and
are all contact ads as to their genre. Patternwise, however, contact ads in
Finland (FIN) are very different from both British and American contact ads.
Finnish contact ads prototypically have the same structure and form as
ordinary advertisements for cars, washing powder or sausages: they all utilize
something to attract the eye, something that stands out from the background
and shouts to the potential reader: READ ME!. This pattern is depicted in
figure 5a as a contrast between (the salient) black and (the backgrounded)
white. Finnish contact ads typically contain phrases like HEY YOU! or
SPORTY BRUNETTE - utilizing, precisely, capital letters, italics, bold face,
etc. - and in addition, of course, they contain a description of who is writing,
and what kind of relationship and person he or she is interested in.
English contact ads are of two types, but very seldom of the Ad pattern
type of figure 5a. One way is to write them in the form of a story with a plot, a
clear beginning and end, very much like an ordinary description of a landscape,
but with a name, pseudonym, or telephone number at the end as the climax.
This pattern is depicted in figure 5b. It is the general pattern for an ordinary
story, with the internal boxes referring to the obligatory (more than one)
foregrounded parts of a story. Thus, as we see, when a visual picture is to be
taken as a graphic representation of how text/discourse is conceptualized, this
representation is also a visual abbreviation for what it means linguistically for a
text to be of this kind: The representation in figure 5b is a short-hand for all the
linguistic manifestations that partake in defining grounding in discourse.
Another English type of contact ad is the Human interest pattern

depicted in figure 5c. This pattern (and the contact ad presented in this pattern)
has a number of golden nuggets (or at least gold dust) - points that will draw
the reader's attention - dispersed all through it, usually with a biggish nugget
at the end of the ad - e.g., of the form Gays only.
n/x
Figure 5a. Figure 5b. Figure 5c.

The Ad pattern. The Plot pattern. The Human interest
pattern.
All three (cf. figures 5a, 5b, 5c) are contact ads, all are descriptive (of
who is writing and who is wanted), but they are presented - and conceptualized
- in the form of different discourse patterns.
The use in different cultures of different discourse patterns can also tell
us something about how that culture relates to particular activities. For
instance, with respect to the suggested discourse patterns for contact ads, we
might want to propose that the way English contact ads (in both patterns) are
written suggests that readers are prototypically expected to read the contact ads
section as a whole, as a smörgåsbord, from which they then choose their
potential partner. Finnish contact ads, however, are like small news stories, like
ads, which prototypical readers might (or might not) browse through, and they
will not stop to read a particular (contact) ad unless drawn to it by a particular
READ ME! sign.
The usefulness of the notion discourse pattern for understanding
coherence should by now be obvious. But what kind of phenomenon is a
discourse pattern?
6 On the ontologicai status of discourse patterns

Detailed knowledge about the extent to which interactants make use of
discourse patterns, and the manner in which this takes place, will have to wait
for the results of empirical - and preferably experimental - research. At this
stage two points need to be made.
88 Jan-Ola Östman
It might seem unscientific to illustrate what a discourse pattern is with

the use of boxes and pictures. Admittedly, to a certain extent, the visual figures
constitute an abbreviation for the purpose of illustration - and as any
abbreviation, they hide certain aspects and highlight others. But the graphical
display itself is no more arbitrary than the use of, say, formulae in
propositional logic. For one thing, vision - a cognitive domain - is an
appropriate means through which discourse conceptualisation can be
simulated. Secondly, it has almost become a tradition in text and discourse
linguistics to illustrate how different cultures use different strategies in
communication with visual representations of the kind I have been using here.
The best known pictures are no doubt those made famous by Kaplan (1972) to
indicate the differences in organization of the contents of paragraphs in
different cultures (examples in figure 6a), and those used by Galtung (1979) to
depict different academic styles (examples in figure 6b). There is no reason
why communication as depicted in terms of images could not just as well be
applied to different strategies of communication used within a culture - as
could, indeed, both Kaplan's and Galtung's figures be used as 'global'
discourse patterns within and across cultures.
Figure 6a. Kaplan's strategies of paragraph progression. From left to

right: English, Semitic, Oriental, Romance, and Slavic strategies.
It is important to note that all such figures have a readily built-in

perspective in them. For instance, if Kaplan had been Russian, he would most
likely have presented the Slavic strategy as a straight line, and the Anglo-
American strategy would probably have been one that suggests too much
explicitness in presentation. Thus, figures like these are primarily to be used as
mnemonic devices, and the important thing is not what they look like per se,
but what they look like in relation to other figures. This holds true for
discourse patterns, too.
AAAA
®
Figure 6b. Galtung 's academic styles.
Top left: Saxonic (non-dialectic, thesis oriented); down left: Teutonic (non-
dialectic, theory oriented); top right: Nipponic (dialectic, thesis oriented);
down right: Gallic (dialectic, theory oriented).
An equally important question related to the ontology of discourse

patterns is: How many discourse patterns are there? Again, this is a question
that cannot be answered appropriately without detailed empirical analyses. I
am inclined to think, though, that as in other spheres of human sense-making
and codification, there is indeterminacy, variability, and negotiability built into
the system of discourse patterns. In addition to the discourse patterns that I
have already given above, I will deal with a very basic one, the News pattern,
in the next section.5
The question of how many discourse patterns there are is a question of
the same magnitude as that of how many genres or text types there are. Sub-
questions here include: Which discourse patterns can be characterized as
variants of other patterns?; Is there a hierarchical order among them (e.g., Are
Kaplan's and Galtung's 'patterns' meta- - or mega- - patterns?); Dia-
chronically, when changes take place, is a change in genre, discourse pattern,
or text type independent of changes in the other two? Even larger questions
include: How far are we willing to take this notion? On what level of
abstraction is the notion still useful and feasible? For instance, can it usefully
be applied to traditional higher-level distinctions like those of speech vs.
90 Jan-Ola Östman
writing, formal vs. informal, planned vs. unplanned, fictive vs. non-fictive,
prose vs. poetry vs. drama? It is obvious that the more all-embracing definition
a discourse pattern gets, the more abstract and emptier it becomes: soon
everything is discourse patterns - but thereby the concept also loses its strength
as an analytic device. My view is that like other images, discourse patterns,
too, have to be placed among the basic-level terms of cognition. This is the
level they are most useful for; they are needed as tools for investigating how
interactants understand coherence in discourse. It is precisely at the level of
discourse patterning that an audience recognizes that a text is, say, a death
notice. This is thus also the level at which a text has the most direct effect on a
reader or audience.
As I have argued, discourse patterns pertain to the holistic perception of
text/discourse; discourse has a particular structure which is associated with a
particular function and contents. That is, they are not simply shape, but they
function as frames for understanding. They are not only linked to perception,
but to cognition: they activate scripts, and thus function as guides for
understanding. Discourse patterns are directly associated with coherence in
terms of understanding. But if that is the case, it also suggests that discourse
pattern similarity implies similarity in manner of understanding, and similarity
in how we perceive and process texts.6 For instance, procedural types of
discourse like the recipe, the guide-book, and direction-giving as interaction,
have a very similar structure: first a presentation of the ingredients (Recipe),
the (history of the) places worth seeing (Guide-book), and the joint
establishment of mutually known landmarks and means of transportation
(Direction giving); then a step-by-step account of the process by which one
gets from ingredients to the finished product, or from point A to point B.
It is unclear at this point to what extent the Recipe, Guide-book, and
Direction-giving patterns should be conflated into one pattern, i.e., to what
extent the Recipe pattern of figure 3 could as such be applied to them all. There
are both good and not so good sides to such a decision: on the one hand,
assigning them the same discourse pattern would stress their similarity; on the
other hand, if a particular discourse pattern gets to be too abstract, it is no
longer usable as a basic-level term. In the present case, if these three were
conflated, it could then be argued that the joint pattern needs to be renamed as
something like the Procedural, or Incremental pattern. But if so, we are very
close to talking about text types, rather than discourse patterns, and thus to
losing the insights the latter notion provides.
Since what patterns look like in relation to other patterns is what is
representationally important, the notion of discourse-pattern similarity is by
definition a strong device in the analysis of textual/discoursal coherence. Thus,
the task of assigning a discourse pattern specification to a (sub)genre can
fruitfully be approached from the point of view of already established

discourse patterns. For instance, horoscopes have certain things in common
with the Recipe, Guide-book, and Direction-giving patterns, since a horoscope
specifies certain ingredients (money, love, work) and how much there will be
of each, but rather than giving directions, a horoscope has certain things in
common with economics articles about future prospects: it gives you advice for
the future.
So far we have seen - both in theoretical terms and through illustration -
what the concept discourse pattern can contribute to our understanding of
discourse and text coherence, and thereby also to our understanding of text and
discourse structure. The rest of this article will show the usefulness of the
concept for concrete discourse analysis.
7 Conflating discourse patterns in the American press:

Clinton and Yeltsin in Helsinki
In order to illustrate the usefulness of the notion discourse pattern in practice -
both for the analysis of language, i.e. texts and discourse in general, and in
order to get a deeper understanding of the English (Anglo-American) language
and culture through an understanding of the discourse patterns that it implicitly
makes use o f - I will supply a brief analysis of the differing reports in some US
and UK newspapers on the Helsinki Summit that took place on March 20-21,
1997. The reports are from the March 22 issues of the papers.
In March 1997, Presidents Clinton and Yeltsin met in Helsinki to discuss
major issues in world politics: the future of NATO; the issue of arms control,
preparing for Start 3; the status and future of the G7 group, and to a lesser
extent the WTO and the Paris Club. In addition, the meeting was of course a
personal meeting between the Heads of State of the USA and Russia: a
meeting to enhance the friendship between Bill and Boris.
Any student of newspaper reports will have come across the upside-
down triangle (figure 7) depicting continuously decreasing information. The
heading (cf. the salient blackness of the READ ME! part of the Ad pattern)
needs to capture the essence of the contents of the report to follow, and at the
same time draw the reader's attention to the news item. After the heading(s)
comes the lead, summarizing the main points of the report; with the heading
and the lead, the gist has been conveyed.
After the lead, the actual news report starts; first with more important
information, then, in decreasing order, comes less and less important
information, background information is added - so that, in principle at least,
the newspaper editor can cut off the story (cf. the broken line in figure 7) at
92 Jan-Ola Östman
virtually any place and give space for the arrival of an important last-minute
news item.7
Figure 7. The News pattern.
The picture that has emerged as an upside-down triangle is for all intents
and purposes also a feasible way to present the discourse pattern of traditional
news items.
There is, however, also another discourse pattern in news reporting that
has come to compete with the traditional triangular pattern. This is the Human
interest discourse pattern introduced in figure 5c, which visually presents a
story as containing several golden nuggets throughout the news report. This is
a very different kind of pattern. There is no way one can cut off at will a news
item that has been written using the Human interest pattern, because, as we
saw earlier, what is very typical of this pattern is that it has one of its largest
golden nuggets at the end of the report.
This human interest manner of writing news items has for long been the
typical way for evening, tabloid papers to tell an interesting story - very often
about people. British tabloids like The Sun and the Daily Mail virtually neglect
topics like the Helsinki Summit. But if they do mention such topics, they dress
them up in a human interest fashion. Thus, The Sun's reporting on the Helsinki
Summit was confined to a brief item hid away on p. 7 of the paper with the
headline Boris fury at 'threat by NATO' (March 22, 1997), concentrating on
Yeltsin's feelings: Mr. Yeltsin called the plan "unacceptable" ... "It is a
serious mistake ".
In Great Britain8 this division of having different discourse patterns
being associated with different kinds of newspapers still functions pretty well:
the triangular News pattern is dominant in news reports in broad-sheets, the
Human interest pattern in tabloids. However, in the United States, tabloid
newspapers do not have the same prominent status for the everyday reader as
in Great Britain. There are a number of tabloids in the USA, too, like the
National Enquirer, but typically the 'soft' news reporting of British tabloids is
to be found in commonly read weekly magazines like People and TV-Guide in
the United States.
Although the triangular pattern has been dominant in US newspapers,

attempts to produce news reports according to the Human interest pattern can
also be found.9 For instance, when the San Francisco Chronicle and the San
Francisco Examiner moved under the same ownership, for a while the two
papers kept appearing, but with the Chronicle being more serious, and the
Examiner coming closer to being a tabloid. At present, the Examiner is only to
be found as a section in the Sunday edition of the Chronicle. The fact that the
Examiner was discontinued might suggest that there is not enough public
demand for tabloids in newspaper format in the United States.
At present, however, what seems to be happening more and more is the
following. First, for market-economical reasons the US broad-sheet
newspapers would try to get a share in the market that is constituted by readers
interested in lighter stuff - including human-interest stories. That is, the
situation is not optimal for the broad-sheets if the weekly published magazines
govern a large part of a flourishing market. At the same time, as we have seen,
people are not used to buying daily newspapers that are 'light3, i.e. tabloid
newspapers. A newspaper is still a newspaper; newspapers seem by definition
to have to be 'respectable' transmitters of news 'objectively.'
In order to cope with this situation, what seems to have developed in the
United States is a discourse pattern that is a combination of the two basic
discourse patterns for news items. A visual representation of this combination
is given in figure 8. (Since it is neither a pattern for 'hard' nor for 'soft' news, I
have termed it 'friendly.')
Figure 8. The Friendly news pattern.
The Friendly news pattern is thus a new, additional frame of coherence

for news reporting. This pattern is used to a varying extent by different US
newspapers. For instance, newspapers like the Houston Chronicle still use the
very traditional triangular pattern, whereas the Los Angeles Times utilizes the
combinatory, Friendly news pattern very frequently. I will illustrate how the
Friendly news pattern is used with a brief analysis of how the Los Angeles
Times reported on the Helsinki Summit.
94 Jan-Ola Östman
The LA Times of March 22, 1997 starts by treating all the four basic
issues of the Helsinki Summit (NATO; Arms control; G7; Bill & Boris) in a
front-page article entitled "U.S.-Russia Talks End in Arms Breakthrough". The
coverage is then split into two items (plus a "News Analysis" item) inside the
paper (pp. 18-19). One article deals with the NATO issue, and the other article
with the Arms control issue. Both of these articles gradually peter out with
respect to their informational content, and thus one interpretation of the
structure of these news items would indeed be to say that the triangular-shaped
discourse pattern (figure 7) has been followed in these reports.
However, a closer look at the reports indicate that, although both of them
have 'less important' information at the end, this 'less important' information
is not wholly of the 'supplying background' type, but it is, precisely, of the
human-interest kind. For instance, the Arms control article, "SUMMIT: Arms
Breakthrough at Talks", ends with an account of Clinton's problem as he had
not been able to sleep; there had seemingly been some disturbing noise in the
hotel. These final paragraphs of the news report are given in text sample 4
below (the low-case letters a-c have been added for ease of reference).
(a)
... seek new, clearer tax laws.

"Russia, in the end, cannot be the
strong partner that we seek ...
Unless ordinary Russian citizens re
ceive the benefit of free markets and
democracy," Clinton said.
Although Clinton did not play the saxophone at this summit, as he did at an
earlier one, and there were no reports of joint singing, the two leaders did share
some laughs.
During a bit of banter in the morning session Friday, Clinton
(b)
told Yeltsin that he had a hard time sleeping between midnight and 2 a.m. because
of a "loud thumping" from the ceiling above him, White House Press Secretary
Mike McCurry said later.
"Boris, I thought you had hired an extra-large Finn to stomp on my
(c)
roof," the president joked, and Yel
tsin laughed, according to McCurry.
McCurry said he was unable to determine the real source of the noise that
interrupted the president's sleep and left him red-eyed and tired-looking during the
news conference.
Text 4.
The (a) segment of text 4 shows how suddenly the break comes between
the spread-sheet kind of reporting on taxes and other business of political
importance, and Clinton's habit of playing the saxophone at some point in his
free time during official meetings. What follows from here onwards is - in a
discourse-pattern interpretation - one big golden nugget of human-interest
reporting at the end of the news item.
Section (a) of text 4 runs almost to the bottom of the page (p. 18). Below
it we find a two-column sized picture of the two presidents seemingly in a
good mood. The caption given to this picture is President Clinton shares a joke
as Russian leader adjusts headset. It is unlikely that the journalist or publisher
actually heard or otherwise knows what went on between the two presidents,
but the picture is clearly a picture of the human-interest type: 'sharing a joke,'
'doing something insignificant like adjusting one's headset,' 'laughing,' etc.
If news items are organized (partly) according to the Human interest
discourse pattern, readers are not expected to stop reading after the lead.
Readers also, of course, very soon learn that it pays to read the whole article,
since at the end one will get a 'lighter shade' of the topic.10 The traditional
manner of presentation would have been to publish a separate Human-interest
insert recounting the incident of Clinton's sleeping problems. Combining the
two stories which have different kinds of news value, the contents of both
stories change: the stress falls on the protagonists as being the same persons in
private and in public, rather than presenting them as being separate individuals
when involved in separate activities. The frame of coherence that is used is
thus not merely a combination of the News pattern and the Human interest
pattern, but a pattern of its own, a new Gestalt.
The manner in which the Friendly news pattern has been presented so far
would indicate that the heading-plus-lead part and the golden-nugget part need
to be of different kinds: here, information vs. humour and other light aspects.
This is not, however, necessary. The International Herald Tribune (March 22-
23, 1997), for instance, makes use of the same discourse pattern, but stays
solidly within the realms of transmitting information. Here we first - starting
on p. 1 - get the issues mentioned: NATO, the G7, the Arms control debate
and START 3, and then at the end (p. 7), when all these have been reported on,
we get an informative golden nugget summary, where all these three are
connected into one - cf. text segment 5.
96 Jan-Ola Östman
It is arguable, of course, whether texts like these instantiate the Friendly

news pattern, or whether they should be seen as argumentative texts dressed up
in a Plot pattern (figure 5b), being similar to the pattern displayed in a
detective story.
... Implementation of
START-3, which would reduce strategic warhead stockpiles for both countries to
2,000 to 2,500, would be completed by the end of 2007.
Mr. Yeltsin seemed particularly concerned that he would be attacked at home for
having been bought off on NATO expansion with economic inducements that
include American backing of Russia's application to join the world Trade
Organization and membership in the Paris Club, a group of nations dealing with
debt and international credit.
Text 5.
As has been repeatedly stressed, the demarcation of what to classify as

the same or as a different discourse pattern will have to be decided on the basis
of empirical research. What is important, however, is that the concept
discourse pattern does provide us with an additional analytical tool, which is
furthermore cognitively plausible, for the analysis of discourse coherence
globally. Thus, as a further example of its usefulness, we can note that it is
clear that a particular pattern, once established, can be used for strategic, even
manipulative, purposes. For instance, once the readership is well-acquainted
with the Friendly news pattern, it will read golden nuggets into any story; and,
by the same token, news reporters and editors will be able to decide what is
printed in golden-nugget position.
The discussion of the different patterns that are utilized in news reporting
should make it clear not only that the notion discourse pattern is a very useful
and powerful device, but also that all three categorizations of text/discourse are
needed in order to get a full picture of discourse coherence: it is important to
know that the genre is news reporting; it is equally important that the text type
is narrative or argumentative. But these, even when taken together, make up a
very blunt instrument for analysis and understanding. The further
understanding that can be achieved through the knowledge of whether a news
item is presented - or, indeed, understood as being presented - as a News
pattern, a Human interest pattern, or a Friendly news pattern is definitely not
negligible.
8 Conclusion
In this study, I have attempted to approach coherence through socio-cognitive
understanding. I have furthermore argued that understanding on discourse level
can best be understood in relation to the concept discourse pattern. Part of
what we conceive of as the coherence of text/discourse is anchored in the kind
of global, cognitive, partly codified understanding we have of how to
categorize the text/discourse in question and how to hook this text/discourse
onto the cognitive frame of understanding that I call a discourse pattern.
Obviously, at this stage of the investigation, the viability of the notion
'discourse pattern' is a hypothesis; but I have shown that it is a plausible
hypothesis in need of further study.
Being socio-cognitive11, the discourse-pattern approach is neutral with
respect to speaker's/writer's intention and addressee/audience effect. But at the
same time, discourse coherence is dependent on the interplay of both. One of
the most important aspects that contributes to the effectiveness - and thus
coherence - of discourse/text is the extent to which an audience recognizes the
discourse/text as being of a particular type. The suggestion of this study is that
the notion 'being of a particular type' can be explicated in terms of discourse
patterns.
The suggestion that discourse patterns play an important role for
understanding and for coherence is in line with, and thus receives support
from, research indicating the importance of scripts and cognitive schemata, the
importance of formulas, the importance of means like discourse topics for
creating expectations in interpreting and construing discourse, and the
importance of what in dialogic terms would be that communication has to do
with being constantly engaged in 'projects' of focused understanding. The
feasibility of the notion discourse pattern can be further appreciated by noting
the large number of fields of applicability it has. For instance, the notion of
discourse patterning offers an additional level for detailed analysis of
intertextuality and recontextualization.12 Secondly, discourse patterns are akin
to 'grammatical constructions' - conglomerates of structural, semantic and
pragmatic information - on sentence level,13 and could from this point of view
be talked about as 'discourse construct(ion)s.' This perspective also enhances
the point that language is more formulaic than we might at first think - also on
discourse level. And thirdly, misunderstandings can profitably be investigated
from the point of view of what negotiative strategies interactants use when
their discourse patterns conflict and (temporarily) hinder understanding.
Analyses in terms of discourse patterns are, however, only one road to a
deeper understanding of discourse coherence. By taking this stance, I indirectly
indicate that coherence might be too large a notion to conceptualize as a whole
98 Jan-Ola Östman
in one go. The discourse-pattern solution is to split up the task into an

understanding of conceptualisable (culture-specific) units that are basic-level
concepts on discourse level, and then to attempt a prototype 14 definition or
specification of these concepts.
Further elaboration of the status of discourse patterns will reveal that
they pertain to what is commonly talked about in terms of ideology. A
culture's network of discourse patterns will constitute a vital part of that
culture's ideology.15 Still, variability, flexibility and negotiability has to be the
norm in studying coherence and ideology: members of a society and culture
have different ways of expressing themselves, of depicting their reality in
language, and of construing their reality - i.e., of being coherent.
Notes
1. In this study I am not concentrating on matters of local coherence - i.e. the connexity
between clauses.
2. The physical shape of texts - in the sense I talk about it here, as a 'pattern' - has naturally
received attention in literary studies, especially in poetics. In linguistics, however, the
shape of texts tends to be either dismissed, or taken for granted without further notice.
Some text-books do mention the importance of the shape of texts (cf. e.g., Bülow-Møller
1989: 51-52), but mainly in reference to, precisely, poetry.
However, in anthropological linguistic studies of indigenous languages and cultures,
shape has a more respectable status. For instance, Hymes (1996) suggests that the way
lines pattern in a narrative shows how human experience gets shaped differently cross-
culturally.
3. Werlich (1976) distinguishes between the narrative, instructive, argumentative,
descriptive, and expository text types. Other classifications include those by Longacre
(1976) and Kinneavy (1971). For an overview, see Östman and Virtanen (1995).
4. The Aesthetic Pattern needs to be dealt with in much more detail than is possible here, it is
the topic of a separate, forthcoming article.
5. Some others that I have found useful are, for instance, the In medias res pattern, and the
Legal document pattern.
6. One way to test the feasibility of the notion 'discourse pattern' as linked to cognition
would be to measure subjects' manner and easy of comprehension of a text when it is
presented directly after another text utilizing the same vs a different discourse pattern. The
hypothesis would naturally be that sameness in pattern enhances comprehension - but
sameness in text type or genre do not.
7. For instance, Cragg Hines's Houston Chronicle report (March 22, 1997: 1A, 16A) on the
Helsinki Summit was reprinted in the Santa Barbara News-Press (March 22, 1997: 1, 2)
with the last two paragraphs left out.
8. Similar distinctions are, of course, to be found in many other countries; the ones that first
come to mind - due to my personal bias - are Finland and Sweden,
9. At tins point I want to express my gratitude to Diana ben-Aaron for several lengthy
discussions that have greatly enhanced my understanding of news discourse.
10. Readers who would not normally admit that they read tabloids, can also in thns way read
'tabloidy' stories without changing newspaper. If my analysis is correct, this also means
that some readers - e.g. somebody who is not at all interested in US - Russian
relationships - might be inclined to jump straight to the end section of the news item, to
read the news report 'funnies' - the 'friendlies' if you like. It may be a coincident, but
sections (b) and (c) continue the news item, next to each other, up on the same page. Even
though this is speculative, we might suggest that these sections were moved up in order to
be easily accessible for readers heading straight for the lighter sections.
11. On the need for both embodied and situational cognition, see e.g. Östman (forthcoming).
12. For a recent overview of the importance of recontextualisation, see Linell (1996).
13. On Construction Grammar, see e.g. Fillmore, Kay and O'Connor (1988).
14. 'Prototype' here specifically refers to the fact that discourse patterns have not been set up,
nor related to each other in terms of necessary and sufficient conditions and similarities.
15. For a view of coherence as one of three parameters of pragmatics as implicit anchoring,
see Östman (1986). The relation between pragmatics and ideology is also touched upon in
Kuusisto and Östman (1998).
References
Auer, P. 1995. "Context and contextualization". In J. Verschueren, J.-O. Östman and
J. Blommaert (eds), Handbook of Pragmatics, 1995 Installment. Amsterdam:
Benjamins. (19pp.)
Biilow-Møller, A.M. 1989. The Textlinguistic Omnibus: A Survey of Methods for
Analysis. Copenhagen: Handelshøjskolens Forlag.
Fillmore, Ch.J., Kay, P. and O'Connor, M.C. 1988. "Regularity and idiomaticity in
grammatical constructions: the case of let alone". Language 64: 501-538.
Fries, U. 1990. "A contrastive analysis of German and English death notices". In J.
Fisiak (ed), Further Insights into Contrastive Linguistics. Amsterdam:
Benjamins, 539-560.
Galtung, J. 1979. Papers on Methodology. Copenhagen: Ejlers.
Goodwin, Ch. and Duranti, A. 1992. "Rethinking context: an introduction". In A.
Duranti and Ch. Goodwin, Rethinking Context. Cambridge: Cambridge
Hymes, Dell 1996. Ethnography, linguistics, narrative inequality: Toward an
understanding of voice. London & Bristol: PA: Taylor & Francis.
Kaplan, R. 1972. The Anatomy of Rhetoric. Philadelphia: Center for Curriculum
Development.
Kinneavy, J.L. 1971. A Theory of Discourse. Upper Saddle River: Prentice-Hall.
Kuusisto, P. and Östman, J.-O. 1998. "The media as mediator. How foreigner
discourse constructs ideology in Finnish newspapers". In J. Blommaert (ed),
Political Linguistics. Amsterdam: Benjamins.
Lakoff, G. 1987. Women, Fire, and Dangerous Things. Chicago: University of
Chicago Press.
Linell, P. 1996. Approaching Dialogue. Talk and Interaction in Dialogical
Perspectives. Linköping: University of Linköping.
Longacre, R.E. 1976. An Anatomy of Speech Notions. Lisse: Peter de Ridder.
100 Jan-Ola Östman
Morris, Ch.W. 1938. "Foundations of the theory of signs". In O. Neurath, R. Carnap

and Ch.W. Morris (eds), International Encyclopedia of Unified Science, Vol 1.
Chicago: University of Chicago Press, 78-137.
Östman, J.-O. 1978. "Introduction: text, cohesion, and coherence". In J.-O. Östman
(ed), Cohesion and Semantics. Åbo: Akademi, 101-106.
Östman, J.-O. 1986. Pragmatics as Implicitness. Ann Arbor: University of Michigan.
Östman, J.-O. and Wårvik, Brita 1994. "The 'Fight at Finnsburh' - Pragmatic aspects
of a narrative fragment". Neuphilologische Mitteilungen 95(2): 207-227.
Östman, J.-O. (forthcoming). "Dialogism and cognition: peaceful vs. confrontational
co-existence". Proceedings from the 16th International Congress of Linguists
(Paris, 1997).
Östman, J.-O. and Virtanen, T. 1995. "Discourse analysis". In J. Verschueren, J.-O.
Östman and J. Blommaert (eds), Handbook of Pragmatics. Amsterdam:
Benjamins, 239-253.
Werlich, E. 1976. A Text Grammar of English. Heidelberg: Quelle & Meyer.
Semiotic Spanning at Conferences: Cohesion and
Coherence in and across Conference Papers and their
Discussions
Eija Ventola
Martin-Luther- Universität, Halle- Wittenberg
1 Introduction
This paper proposes a new term semiotic spanning and suggests that it might
be useful when we try to describe semiotically what is going on in
linguistically complex situations such as conferences. I have for a long time
been interested in academic writing in English and improving non-native
writing, especially by young scholars, and thus I have in this context been
interested in cohesion and coherence. Cohesion operates within texts and is
realised by cohesive links; coherence is a situationally determined relation,
linking up with the interpretative contexts of texts. I have also for a long time
been interested in how conference papers are written and presented and how
the discussion evolves after the paper. Here, too, the notions of cohesion and
coherence play a role - no participant at a conference wants to transmit an
image of her/himself as an 'incoherent' presenter/discussant. Yet, there are
times when we as conference participants wonder what the presenter's/
discussant's point really is and how it links to what has previously been said.
The notions of cohesion and coherence are helpful when showing individual
writers/presenters how to improve their texts. Yet for my present purposes
these notions seem to be too text and context-bound, i.e. too much tied to an
individual text and its relation to the immediate context of production.
This paper aims to demonstrate that we must extend our interests. We
should aim to go beyond textual cohesion and coherence and focus on how
texts are interrelated semiotically. Some might argue that we have already
developed such perspectives, as the notions of intertextuality and genre
suggest. But the kind of aspect that I would like to focus upon here is not, in
my view, covered by those terms. The notion of intertextuality, grossly
simplified, seems to cover the 'taking over' of certain features (mostly
ideational content) from one text to another, an appropriate example being an
102 Eija Ventola
editorial text that is based on one or two articles that have previously appeared
as news reports. The notion of genre seems to cover the relationship of texts of
the same kind as tokens of a certain class of texts. So, for instance, articles in a
linguistic journal may belong to the same genre (research articles) or agnate
genres (research articles vs. review articles). But what I am trying to present in
this paper is a more extensive view on how texts relate to each other by
spanning semiotically, i.e. linking up with various kinds of existing and
experienced texts (and other semiotic modalities) and creating new semiosis
through these links. This approach is necessarily dynamic as it spans across
various texts and situations. The closest thinkable comparison might be
'surfing in the Internet in the hypertext world of links' and creating something
new through the process, but 'new' not in an individualistic sense (as for
example a path created by the hypertext reader), but in a collective sense.
I shall approach semiotic spanning by looking at why the other above-
mentioned terms cohesion, coherence, intertextuality and genre are not
sufficient in explaining the kind of phenomena I have in mind. Sometimes the
spanning can, for example, be perceived between a conference paper and its
discussion, but it can be understood more extensively as well. Semiotic
spanning can be seen to function between various instances of genres within
the speech event - the different conference papers and discussions within one
section or during the whole of the conference when the discussions are built
upon previous papers and discussions. It is also part of the other kinds of
genres at conferences, e.g., when papers become part of other conference
genres - dinner-table talk, coffee-time chats - what we call talking shop. And of
course semiotic spanning exists between the presented paper and its source
materials as well as with the final written version of the paper. In this paper I
will focus on a number of issues: 1) semiotic spanning between a conference
talk and its sources, 2) the spanning between the talk and its discussion, 3) the
spanning between these texts and the texts of other talks, their sources and their
discussions and the conference as a whole, 4) the spanning between the
forthcoming written and spoken texts after the presented talks and their
discussions at a conference and the influence they in turn have on other papers
written for presentations and publication in the future.
2 Why conference papers and discussions as data?

My initial interest in collecting conference data was to see how English is used
for communication at conferences by scholars from various cultural and
language backgrounds at a conference. Thus, for three days I videorecorded
conference data at an international Georg Forster colloquium that was
organised in Germany. My interests soon started developing beyond English as
Coherence and Conference Language 103
an international language and non-native and native negotiations to the kind of

semiotic spanning issues that have been introduced above. The paper uses
mainly examples from one of the talks, its videotaping (presented in English)
and its published version (translated into German) for purposes of illustration
of semiotic spanning.
3 Cohesion and coherence
When we think about a conference paper delivered in front of an audience, we

are concerned with its functioning as a piece of cohesive and coherent
discourse in its communicative context. Usually the responsibility for the
cohesiveness of the text lies with its producer, the presenter. The coherence of
the text relies on the work that both the presenter and the audience do in
interpreting how the text 'fits' the context. Let us look at a short example and
consider its cohesiveness and coherence. (See notes for translations of
German.)
(1)
1. Chair: Wenn nicht würde ich vorschlagen, daß wir fortfahren er mit dem
Vortrag von Frau XXX der diesen Sektor der wissenschaftlichen
Hinterlassenschaft in Form der Sammlungen... der Biographie also eine ganz
wichtige Sache behandelt
'Chair: if none {more questions} I would suggest that we go on with Mrs.

XXX's talk which deals with that part of the scientific legacy that has been
left to us in the form of collections ... of the biography, that is a very important
subject matter'
2. Presenter: Good morning ... At first I thought I would give my talk in a

Polynesian language which might be more relevant for today. However I
thought I'd be a bit more intelligible and do it in English ... So I begin with
'Cook, the Forsters and Pacific Islanders: Three Visions of Science, Curiosity
and Art'. In the London Gazetteer of August eighteenth eight- seventeen
sixty-eight Lieutenant James Cook noted that the aim of his first Pacific
voyage was to attempt some new discoveries in that vast unknown track. Five
years later during Cook's second voyage in December seventeen seventy-three
Johann Reinhold Forster, speaking for himself and ultimately his son George,
noted that his motivation for joining the voyage ... was the thirst for
knowledge [and] the desire of discovering new animals and new plants for the
benefit... which should accrue to science ...and the additions to human
knowledge ... in general. That is ... while Cook's vision was originally
focussed on new geographic and astronomal {Presenter searching for the
right word} astronomical knowledge, ... Forster was searching for biological
104 Eija Ventola
and anthropological knowledge. Although both Cook and the Forsters

succeeded in making new discoveries ... their successes do not necessarily lie
... in the visions of Cook, the Forsters, or their patrons but... at least partly in
luck of being the right people ... in the right place ... at the right time. {the
presentation of the paper continues}
At the beginning of example (1), since no more questions are asked about
the previous paper, the chair hands the speaking turn over to the next presenter,
Frau XXX (a native speaker of American English), who after the initial
remarks about the language of the presentation starts producing what we in the
academic world would consider an appropriate beginning of a paper. Let us
first focus on its cohesion.
3. J Cohesion
We can begin with Halliday and Hasan's (1976) notion of cohesion as an

expression of semantic continuity which is realised by the lexicogrammatical
means. They recognise four kinds of cohesion, reference, ellipsis, substitution,
lexical cohesion, conjunction. We can certainly find examples of these kinds of
semantic cohesive relations in example (1), e.g. referential relationships
between entities like Johann Reinhold Forster, himself, his son Georg (himself
and his referring to J.R.F.); such lexical relations as geographic, astronomical,
biological, anthropological (all being co-hyponyms), conjunctions like that is,
while, although, and so on, which link the messages logically. These make the
presenter's text cohesive and thus hopefully more understandable to the
listeners.
Halliday and Hasan (1976: 298-299) look at cohesion as unity and
continuity "that exists between one part of a text and another" and as "the
points of contact with what has gone before". Cohesion in this sense does not
seem to go beyond the boundary of the individual text created. Yet example (1)
has interesting links with other texts, but the notion of cohesion will not
explain these, namely the linking between the presenter's text and with what
Cook and Forster had allegedly said.
A further question is, whether or not we can consider the talk and its
discussion as parts of the same text and thus cohesively linked. In example (2)
a member of the audience asks a question about the Forster collections which
Frau XXX had studied and which the chair had actually mentioned when
introducing her in (1) and also when, after the talk, he invited the audience to
present questions: ... Sie sollten die Gelegenheit auch wahrnehmen gerade
auch zu den Sammlungen Fragen zu stellen [you should take the opportunity
and ask questions about the collections]. The question in (2) concerns a
subcategory of the Forster collections.
(2)
1. Discussant: I'd like to ask a question. You can rather ignore it probably. Can
people visit erm Göttingen the the collection the collection?
2. Presenter: Pardon?
3. Discussant: Can people visit the- these collections which are at Göttingen. I
mean, if I- Is there- And everyone can go there?
The identity of the Göttingen collection has been established by the

speaker already during her talk in (3) when she listed the places where the
items collected by the Forsters are stored.
(3)
The next most important collection is the Institut und Sammlung für
Völkerkunde in Göttingen. This collection was sold to Göttingen in 1799, after
Reinhold's death. Its some 157 listed pieces is similar in scope and variety to the
collection in Oxford and was apparently Forster's own systematic collection that he
had retained for himself, although it does not include many unique objects and
diminished over the years by giving away and selling pieces.
How far can we extend the cohesive interpretation of points of contact? If

we apply the terms of linguistic referencing and lexicalisations to 'the
collections' here, we would have to extend the notion of cohesion across three
different kinds of language use: the chair's introduction, the talk, and the
discussion (examples 1-3). Yet, as will be shown later in (5), the presenter
actually carefully constructs the identity of the collection and its subparts to her
audience. But to her and the chair (and perhaps to many in the audience, too)
'the collections' seem to have points of contact outside of the present text and
the context of situation. They have read about 'the collections' in other texts
and talked about them even before coming to the conference. These links are
difficult to explain within the boundary of the text unfolding as a paper and
discussion. Such links can perhaps be considered in terms of coherence.
3.2 Coherence
I shall again use Halliday and Hasan's views for discussing coherence (but I
also acknowledge that other scholars have explored coherence extensively
from its various aspects, cf. this volume). Coherence can very briefly be said to
be about the relationship the text has to its context of situation:
106 Eija Ventola
...texture involves more than the presence of semantic relations of the kind we refer
to as cohesive, the dependence of one element on another for its interpretation. It
involves also some degree of coherence in the actual meanings expressed: not only,
or even mainly, in the CONTENT, but in the TOTAL selection from the semantic
resources of the language, including the various interpersonal (social-expressive-
conative) components - the moods, modalities, intensities, and other forms of the
speaker's intrusion into the speech situation... A text is a passage of discourse
which is coherent in these two regards: it is coherent with respect to the context of
situation, and therefore consistent in register; and it is coherent with respect to
itself, and therefore cohesive.
(Halliday and Hasan 1976: 23)
In (1), we notice that when the presenter starts her presentation she
produces a speech act which we may interpret either as an apology (feeling
apologetic because she cannot present her paper in German) or as a joke. This
part of her presentation does not relate in any of the ways described above as
cohesion to the chair's turn or to her own paper, except for one lexical item
talk. Yet, no one with experience of international conferences considers such
speech acts as 'social niceties', jokes, etc. as odd and out of place. So,
contextual coherence in the communicative situation explains how the parts of
the presenter's text which lexicogrammatically do not seem to go together can
actually be seen to be linked. The same applies to the chair's turns, when he
introduces the speaker and opens and leads the discussion. His turns are
"coherent with respect to the context of situation, and therefore consistent in
register" (Halliday and Hasan 1976: 23), in terms of Field (relevant social
action), Tenor (relevant social and interactional roles) and Mode (channels of
communication, medium), and thus seem to form a unity that is contextually
coherent. Yet at times, especially during discussions, this contextual coherence
is put to test.
Achieving cohesion and coherence seems a complex process in the
context of conference papers and demands hard work from all of the
participants. When the presenter drifts off his/her topic or keeps jumping back
and forth topically, we may find it hard to build up links with what the
presenter is saying and the initial topic of the paper, and we may consider this a
failure in maintaining textual cohesion. But we may also experience breaches
in contextual coherence. Sometimes at conferences we find it hard to link up
the content of a paper with the global topic of the conference (if there is one),
with what the other speakers present in their papers and how their own views
come into play. As discussants we do not always see things in the same way
and even seem to have difficulties in understanding what the other discussants
want to say. In spite of the context of situation, coherence seems to be lost, as
in (4) where the presenter not only has difficulties in understanding what the
discussant is saying, but also finds it difficult to establish common ground with
her. His linking differs from the discussant's.1
(4)
1. Chair: Ja bitte
'Chair: Yes please'
2. Discussant: Ich wollte nur... eine kurze Be Bemerkung. Sie haben von der
erm team teamship oder oder sagen wir nicht so erm oder die enge erm
Beziehung zwischen Vater und Sohn wissenschaftlich jetzt. Und meine Frage
wäre in wie fern man das auch so erm ... erweitern könnte also jetzt politisch
gemeint also ich meine, daß Georg Forster, der Sohn, dann eine völlig andere
erm einen völlig anderen Weg gegangen ist erm jetzt was seine politische
Entscheidung anbelangt. Also das heißt nicht daß der Vater da auch eine Rolle
spielen erm könnte aber ich glaube schon, daß erm solche biographische
Erklärung Erklärungen nicht ausreichen, um dann Forsters politische, jetzt
Georg Forsters politische Entwicklung zu erklären. Das ist wieder mein
Problem.
'Discussant: I just wanted a brief remark. You have {no main verb 'talked'}
about the erm team teamship or or should we not say about so or a close erm
relationship between father and son now scientifically. And now my question
would be in how far one could erm expand this, now in a political sense, that
is I mean that Georg Forster the son then went {'chose'} a totally different
erm a totally different way in erm what now concerns his political decision.
That does not, however, mean that the father would not be playing erm a role
there, too, but I think that erm such biographical explanation explanations are
there not sufficient to explain then Forster's political now Georg Forster's
political development. That again is my problem.'
3. Presenter: Have I understood you rightly that you think that Johann Reinhold
Forster had no influence on the political thinking of George
4. Discussant: No I meant perhaps he could have had
5. Presenter: *Oh I'm -
6. Discussant: *but not that much because when you erm read when you know
of erm Johann erm Reinhold's erm position after George Forster's erm erm
you said that in English erm
7. Presenter: Arrival back in Germany
8. Discussant: No I mean in Mainz erm
9. Presenter: Ya
10. Discussant: And after his death its erm its quite complicated and well... But I
think we can't discuss about this because it's erm we are not erm it's not
possible to discuss what was going on in their minds
11. Presenter: *No other than -
108 Eija Ventola
12. Discussant: *And I think that my problem is only how far if we consider this I
I think your task is very important I mean erm what concerning Johann
Reinhold Forster because he's been very erm erm
13. Presenter: *neglected
14. Discussant: *neglected ...**aha
15. Presenter: **and is still ya
16. Discussant: in England...where you you can still- well I'm Portuguese so I
can speak sort of erm
17. Presenter: Ya
18. Discussant: as an outsider
19. Presenter: Ya
20. Discussant: But I think if we insist on that point I think that we forget Georg
Forster as a revolutionary
21. Presenter: *Ya
22. Discussant: * That's my point. It's too much
23. Presenter: **Ya
24. Discussant: **scientist and
25. Presenter: Yes
26. Discussant: and zu wenig Revolutionär.
'Discussant: and too little of a revolutionary'
27. Presenter: Ya. I I I have to say quite frankly in the Southern hemisphere
that some of us here are interested in him primarily as a scientist. You can
keep your revolutionary activity if you like I mean we are interested but for us
erm and I'm speaking not speaking personally because I'm interested in the
whole of the life but here's the problem. ... They were thrown out of England,
their forefathers were thrown out of England for political reasons
{the asterisks and the underlinings indicate simultaneous speech}
Here the discussant attempts to build up links between the presenter's

views on the relationship between the Forsters and her own views. The
discussant's native language is neither German nor English, and therefore she
has difficulties in constructing her messages. Creating text dynamically during
the discussion 'in real time' is hard for all of us and may sometimes make our
discussions seem less cohesive and coherent. In our written papers we can
check cohesive ties and thus hopefully guarantee coherence to our readers. In
(4), however, although the presenter tries to help, the dialogic cohesion and
coherence does not seem to be operating well.
But the problem we see in the above example is not merely a matter of
cohesion and coherence. Obviously the two participants also move in two
slightly different semiotic spheres of knowledge, understanding, and interests
about the topic, and as a result they do not seem to succeed bringing them
together. This difficulty in linking becomes obvious in the presenter's last turn
(27), where the presenter contrasts his interest in the subject to that of the
discussant's (i.e., G. Forster as a scientist and his life as a whole vs. as a

revolutionary). It seems that this kind of linking between different
views/approaches/semiotic worlds cannot be considered simply a matter of
contextual coherency, but rather as a matter of relating someone's whole
semiotic world (accumulated through reading/listening/seeing/experiencing) to
another person's world. To understand how such linking operates, we need to
go beyond the cohesion of the present text and beyond the notion of coherence
of the text in the context of situation. This will be done later in the section on
semiotic spanning, but let us first summarise the discussion so far.
When measuring and evaluating cohesion and coherence in texts,
scholars compare individual texts with others and consider what makes one
text more coherent than another. Little attention has, however, been paid to the
way texts influence one another in the process of text creation. Examining the
conference paper provides us the possibility of understanding the paper
coherently and allows us to see the discussion that follows it in its relation to
the paper. But what we also have to consider is how links are built not only to
the presented paper but to many other papers and their discussions at the
conference and, furthermore, to many other texts within the conference
participants' sphere of experience. Are we now leaving the sphere of cohesion
and coherence and entering the sphere of intertextuality?
4 Intertextuality
Intertextuality is mostly seen in terms of'sharing ideational content'. A typical
example of intertextuality in the context of conferences is perhaps seen in
terms of quotations and references to other authors' work in the papers. Parts
of texts become parts of other texts; the information is condensed, reformulated
and reshaped to fit the purposes of the author (cf. an editorial based on news
reports or the background and theoretical sections of academic papers).
According to de Beaugrande (1980: 14), this kind of intertextuality involves
"experientia knowledge of specific, actual occurrences" (his italics), and he
gives the following as examples of where we find intertextuality of this kind:
summaries, protocols, continuations, replies, parodies. The notion of
intertextuality has generally been used to describe the relationship between
written texts - texts that are products (for a detailed discussion, see e.g. de
Beaugrande and Dressier 1981: 188-214). Thus, looking at intertextuality at
conferences would mostly mean considering the relationships between the
presenters' written papers and those texts they have referred to for writing their
own texts. But this view is hardly sufficient since what is presented and
discussed at conferences not only relies on written texts, but on experiences
gathered through other semiotic means as well.
110 Eija Ventola
As already suggested, during a discussion interactants frequently link

what they say to what unfolded dynamically in the previous discussant's turn.
Unless the discussion is taped and transcribed, there is no record of this
dynamic unfolding. Thus, we cannot always be certain of what kind of
intertextual linking takes place and what kind of other semiotic knowledge gets
reshaped and transmitted when the presented papers and the discussions which
follow are talked about at later times (i.e. at coffee breaks and at dinner table,
and after the conference with other colleagues who could not attend the
conference). An attentive and enthusiastic academic may eventually build the
content of the talk and the discussion s/he heard into his/her next lecture series
and visualise it instead of verbally recounting the said/heard, and so on.
Intertextuality, too, seems limited for our present exploration, because it
appears to be based mainly on ideational/experiential (content) function of
language and its written records. Its interest is in how parts of written texts get
'transferred' to other contexts of situation and to other texts and the processes
that take place when this happens (e.g. in translation, in summarising, in recall
tasks used in psychology; see de Beaugrande 1980). This seems too simplified
a view when we consider what is going on in conference discourses and what
happens before and especially afterwards when the total 'intellectual
contribution' of the conference starts living a life of its own and can be traced
back in other texts, i.e. in other papers yet to be written, in discussions led at
the conference and afterwards, at lectures at universities, and so on. It seems,
then, that we somehow have to work out how texts, spoken and written, can be
related to one another, as types, not just in terms of shared ideational content.
Perhaps the model that helps us to sort out the various kinds of linking is the
genre model, looking at linkages in terms of the generic qualities of texts.
5 Genre
Martin and others have elaborated Halliday and Hasan's (1976) views on
textual unity in terms of cohesion and coherence and text construction and text
relatedness in terms of genre and register (see e.g. Eggins 1994; Martin 1992;
Ventola 1987). The benefit of the notion of genre in this context is that it could
offer us a dynamic view to generic unfolding of goal-oriented social activities,
thus allowing discourses which unfold slightly differently still be considered as
realisations of the same genre. It also allows different registerial realisations, at
least to a degree, during the unfolding process and guarantees appropriateness
in terms of field, tenor, and mode choices.
The choices made on the genre and register level have repercussions to
the ways texts are realised as discourse, lexicogrammatically and
phonologically (for examples and discussion, see Martin 1992). When we think
of social behaviour of various members of our culture we perceive them

behaving alike on many occasions. When we see videotapings of conference
papers and the discussions, we usually have no difficulty in recognising them
as examples of their genre. Discourses with the same kind of generic structures
belong to the same genre, although in actual instantiation of texts certain
variation naturally occurs due to the dynamic unfolding of interaction.2
Generic unity and its analysis allows us to go beyond examination of
individual texts and their wholeness. The notions of genre and register allow us
to look at texts in terms of how they are the same and how they are different,
but it does not necessarily help us to understand the kind of linking that has
been suggested here within and outside the contextual framework of the genre
in question. In other words, we can demonstrate that although Frau XXX's
paper displays slightly different registerial choices (e.g. in terms of field - the
subject matter) it generically unfolds in a similar fashion as e.g. Herr ZZZ's
paper. The same can be said of the discussions that follow the two papers.
They unfold in their own way as dialogues, and differ from papers and e.g.
from casual conversations (where the dialogue is not controlled by the chair).
They also have the peculiarity of not appearing as independent unities. At
conferences we may not have a discussion without the presentation of the
paper (except when we change the genre, e.g. into 'a round table discussion'),
but we can have a paper without the discussion. And the discussion can
develop to the direction where it is hard for the listeners to see it as 'the
discussion of that particular paper'. The discussion can drift off the topic of the
paper (and perhaps even off the field of the conference).
Where does this leave us in terms of genre classification of these two
types of discourses? Linguistically speaking, the paper and its discussion
should be considered as separate genres - so different are their unfolding and
the consequent linguistic choices and realisations. However, we also find links
and thus need to explain how the discussion is linguistically and semiotically
related to its paper. And, if during the discussion (or in the paper) other papers
or their discussions from the same conference or other contexts are also taken
up and discussed, we then have to explain these connections as well. The genre
and register model does not seem to capture all the aspects of this kind of
linking. We still need to work on and develop a notion which enables us to
capture and explain how links are established between texts and other semiotic
modalities created in totally different contexts for different purposes. Below it
will be suggested that semiotic spanning might be used for describing those
aspects of papers and discussions where presenters and discussants seem not
only to connect within the text they are creating (the paper and its discussion),
but to the views presented by previous presenters or to something which they
themselves have written/read/seen/photographed/videorecorded, etc. elsewhere
112 Eija Ventola
- in other words, for explaining the various kind of linking and expanding of
semiotic information to other communicative contexts.
6 Semiotic spanning
What has been suggested above is that the previously discussed notions of
cohesion, coherence, intertextuality and genre are insufficient to explain in
terms of linking what goes on before, during and after conferences, what takes
place at conferences at all levels (including cafeteria talk about the papers) and
what keeps conferences together as social events. We do not sufficiently know
how such events as a whole contribute to 'the discourse world', and as linguists
we have never really attempted to describe all these links. The proposed term
semiotic spanning might be useful to explain this kind of complexity in the
world of discourses and the various combinations of semiosis that we
encounter. We should be trying to explain what makes conferencing
linguistically possible, how its influences are felt and developed afterwards and
what changes conferencing consequently initiates in the world of discourses.
We know very little about what it is that allows us to bring in so many aspects
into conference situations and yet (mostly) achieve some unity in knowledge
and understanding during and after the conference, and then take a step
forward in our understanding how the world (of discourses) functions.
Conferences as social events provide 'life' to conference papers and their
discussions (formal and informal). After the conferences, texts 'live on' in
form of other similar professional genres: in following discussions, in
conference volumes, articles, in the quotations in other people's articles,
reviews, in popularisations of papers, and so on. We even need to be able to
describe what is going on between related and non-related genres, i.e. how, for
example, 'scientific knowledge' presented in a paper at a conference gets to
become a part of a coffee/dinner-table discussion or a part of a newspaper
report or a school essay. This is where perhaps the notion of semiotic spanning
will prove to be useful.
It seems that so far we have hardly begun to adequately address all the
aspects of conferencing. Although some research has been done on the drafting
stage of writing papers, we still do not know enough about what conferencing
preparation entails - i.e. how the presenters combine the material they obtained
from different sources through different semiotic modalities (including
internet) with their own earlier reading and acquired knowledge in preparation
of their papers. Further, no systematic study has been conducted to see how the
activities of the presentation, giving handouts, showing transparencies or
slides, playing audiorecordings or videorecordings, etc. link with the
conference situation and the papers the participants present there, and how
these various modalities then enhance the view of knowledge/world

constructed at the conference. We also know little about how the feedback we
get for our papers at conferences influences the (re)writing of the paper for its
publication.
All such linking happens largely through what I call semiotic spanning,
aspects of cohesion and coherence, register and generic similarities being part
of the realisations of semiotic spanning, but they in my view concentrate either
on individual texts or the texts of the same kind, whereas semiotic spanning
takes place between various kinds of texts independent of their generic
qualities and their realisation modalities.
I shall now illustrate what might be the operational areas of semiotic
spanning, beginning with the examples already presented. In (1), the presenter
makes links with other texts by quoting other writers. This is not so obvious
when one reads the transcript of the presentation in (1), although the wordings
Cook noted and Johann Reinhold Forster, speaking for himself and ultimately
for his son Georg naturally suggest a quote. In the transcript, the source for the
Cook quotation seems to be the London Gazetteer, 18.8.1768, whereas no
source for J. R. Forster's saying is given. The published version of the paper
makes the spanning clear, but also shows that linking is somewhat more
complicated. The published version gives (a), to attempt some discoveries in
that vast unknown track, as the 'writing' of Cook, and (b), The thirst for
knowledge [and] the desire of discovering new animals & new plants [...], and
(c), The benefit which should accrue to science, and the addition to human
knowledge in general [...], as the 'writing' of J. R. Forster. The writings of
Forster, (a) and (b) are, however, mediated by another text, an introduction to a
book titled The 'Resolution' Journal of Johann Reinhold Forster, 1772-1775,
and this is made clear in a footnote in Frau XXX's published article.
The source for (b) is originally Forster's diary 21.12.1773 and for (c) a
letter to the Earl of Sandwich, 24.2.1777. So, in Frau XXX's presentation we
have semiotic spanning to three texts from three different genres: a newspaper
article, an entry to a diary, and a personal letter. The original wordings in those
texts have been created for these generic purposes, and they have been taken
out of that context and made into parts of another text with another purpose, a
scientific article. In the case of the Forster quotations they are mediated
through a text the purpose of which is 'an introduction to a book'. The study of
semiotic spanning should clarify what the relationship is between the
presenter's manuscript, her talk, and the published work and the sources for
these quotations. For example, science and human knowledge in the Forster
quotation (c) seem to be linking up with the presenter's own lexical realisations
of summarising Cook's vision as geographic and astronomical knowledge and
Forster's as biological and anthropological knowledge. Presenters at
114 Eija Ventola
conferences generally use wordings created for other textual functions for their
own purposes and thus create a semiotic span between the original text and the
one created by them.
With this kind of spanning the presenters often bring 'knowledge/views'
established by others and incorporated with their own to the attention of the
audience. Sometimes, however, the presenters slowly build up the views/things
that they want to share with the audience. Thus, we notice for instance in (1),
that for the chair (and perhaps for many others in the audience) 'the Forster
collections' seem already to be 'shared knowledge' (as indicated by the
wording when introducing Frau XXX). But the presenter cannot rely on this
kind of semiotic spanning functioning for all of the audience. She must build a
'shared view' with his/her listeners, and we see it happening in (1). The
presenter establishes technical terms and gets them accepted as 'referable
things', which then can be taken as topics for discussion by the audience. In
this way the process contributes to the semiotic spanning between the paper
and its discussion. If we look at Frau XXX's presentation as a whole, the most
important items discussed in the whole paper and in the discussion are the
Forster Collections - the artefacts that the Forsters collected during the second
Cook voyage around the world. They have been distributed to various places as
separate collections: Oxford, Göttingen, Görlitz, Mittau and Stockholm.
The collections can be related back to the transitivity structure that goes
back to the events in the Forsters' time, namely the fact that the Forsters
collected artefacts, which implies the following congruent transitivity
structure: Active  Material Process ˆ Goal3. At the beginning of Frau XXX's
talk, this structure is also prominent, although grammatically it does not always
get realised congruently as a material clause (and as we have already seen the
chair uses the nominalised form in his introduction of the presenter). Example
(5) shows the first mentions of the activity of 'collecting' in Frau XXX's
paper.
(5)
As far as the Forsters were concerned their purpose on the second voyage was to
contribute to the scientific knowledge of the world through collecting and
publication. Strictly speaking, there were problems here also. Many of the new
plants and animals had been previously discovered and collected during the first
voyage by Banks and Solander. Most significantly the Forsters' biological and
anthropological specimens, although collected systematically, were widely
scattered and there was no systematic publication of the botany, zoology, geology,
or anthropology by the Forsters or anyone else... (later in the text). Thus our focus
on the present day is the distribution of the ethnographic collections made by
George and Reinhold Forster and the knowledge that can be gained from the
objects and the writings about them.
When one examines the final stage of (5) and examples (2) and (3), it is
obvious that the material process has been established as a referable Thing. The
activity sequence has become a technical term in Frau XXX's presentation and
is, for example, realised in its nominalised form as the Forster Collection, i.e.
Deictic the  Classified Forster  Thing Collection or as the largest, widest
ranging and most important collection made by the Forsters (Deictic  (Epithet

Epithet  Epithet)  Thing  Qualifier). This Thing is further subclassified
into different kinds of collections, named after the places where the items are
kept today (e.g. in Göttingen in example (2)).
By the time the discussion starts, this technical term and all the other
ones similarly created, have been accepted by the audience as things that one
can refer to, talk about, argue about. For instance, in (2), when the discussion
starts one of the conference participants asks Frau XXX: Can people visit eh
Göttingen the collections?. The created classification becomes the focus in the
discussion. This kind of 'establishing shared view' (mostly through what
Martin 1992 called ideation) and then referring to it (Martin's identification)
helps us as conference participants to think that we are talking about the same
things and that we understand each other and the argumentation. It establishes
links between the paper and its discussion and the other papers and their
discussions presented afterwards. In our example text, other researchers can
refer back to the collections mentioned by Frau XXX in their papers and thus
create a new node in semiotic spanning that helps to unify the conference as an
event.
Frequently, however, discussants do not just establish semiotic spans
about shared information, as in (2), but rather make links with their 'semiotic
worlds', as in (4), where the discussant tries to bring forth the political aspects
of Georg Forster into the discussion, but this spanning is rejected by the
presenter. Here then we have a confrontation of two semiotic spannings - both
the discussant and the presenter link to the semiotic worlds which they have
experienced and do not seem to be able to reconcile their views.
In the discussion following Frau XXX's paper we indeed have several
instances where semiotic linking is made to some other things than the shared
'world' that she establishes in her presentation. Other expertise is brought into
light in (6).
(6)
1. Chair: {after a fairly long turn by the chair himself and the silence, he gives
the turn to the discussant} ... (7 secs) Ja?
'Chair: Yes'
116 Eija Ventola
2. Discussant: Gestatten Sie mir einige Bemerkungen zu dem Besuch auf der
Osterinsel... erm die beiden Forsters wie de- wie der größte Teil der ganzen
Besatzung waren sehr krank. Erlauben Sie mir, daß ich das sage. Ich bin
nämlich Mediziner. Sie hatten alle Skorbut fast alle und der alte Forster
beschreibt ja wie mühselig er gehen konnte. Das lag an den dicken Beinen, die
er hatte...Ödeme, die sich da gebildet hatten in Folge dieser C-A A-
Vitaminose, das war's übrigens nicht allein. Die Nahrung war qualitativ und
quantitativ unzureichend. ... Deswegen ... ist ja auch Kapitän Cook so rasch
wieder abgereist... weil sie nicht genug zu essen bekommen haben ... und der
... jüngere Forster Georg ... der war ja damals noch gar nicht so alt noch. Der
Junge war ja sehr jung ... erm der eine war ja erst 18 Jahre ... da ... erm der
jüngere war so krank, daß er also kaum sich an Land schleppen konnte.
Vielleicht ist es doch nicht ganz uninteressant zu sagen wie diese Fülle von
Materialien, die die beiden schwerkranken Leute in zwei Tagen gesammelt
haben. Das finde ich doch immerhin beachtlich.
'Discussant: Allow me {to make} a few remarks concerning the visit to the
Easter Island ... erm both Forsters as well as the- as well as the main part of
the whole crew were very sick. Allow me to say that. I am namely a doctor.
They all suffered from scurvy, nearly all and the elder Forster describes how
difficult it was for him to walk. That was due to the swollen legs he had.
oedema that had developed for all as a result of this C-A A-Vitamin
deficiency, that wasn't by the way the only reason. The nutrition was
qualitatively and quantitatively insufficient... Therefore ... Captain Cook
actually left so quickly again ... because they did not get enough to eat... and
the ... younger Forster, Georg...who at that time was not that old yet. The boy
was very young ... erm he was just 18 years ... when ... erm the younger was
so sick that he could hardly drag himself on the shore. Perhaps it is not too
uninteresting to mention what a wealth of material these both of these two
seriously ill people collected in two days. That I still find all the same
remarkable.'
The discussant is a doctor, and he brings a medical aspect into the

discussion. From his point of view his remarks are coherent in the context of
situation and relevant to the topic. Yet, the conference participants may
initially experience his comment as incoherent and irrelevant when listening to
him, because the link to Frau XXX's paper and the Forster collections is made
only at the end of the discussant's turn (the amount of collected material) and
even then may not be considered a relevant link by many in the audience.
Thus, the rhetorical force of the comment is largely wasted. How many in the
audience can actually relate to what this discussant is saying depends very
much on their ability to establish appropriate semiotic links with such medical
expertise for themselves.
In the next example, (7), the coherence and the semiotic spanning is
further complicated by the fact that the discussant's remark is not at all
addressed to the presenter, Frau XXX, but to the chair who had presented his
paper before her at the conference. Thus, to see the discussant's turn it in the
right light, the audience (and we as the readers of the transcript) have to
establish links with the chair's paper and his semiotic world as well as the
discussant's world.
(7)
1. Chair: Ja noch weitere Fragen (7secs)
'Chair: Yes, some further questions'
2. Discussant: Keine Frage, sondern eine Ergänzung zu Ihnen Herr H. {=the

chair} zu ihrem sehr wertvollen Beitrag ... Ich möchte sagen daß was Sie
herausgearbeitet haben, daß die ... Bedeutung der Aussagen und ... erm
Aufnahme von den beiden Forsters auf der Osterinsel... und wie Sie die
Kultur interpretieren und vorher in die (Kultur ?) eingebaut hat. Das gleiche
gilt natürlich auch für die Gesellschafts-inseln und gilt auch zum Beispiel
für...und ich möchte da das Beispiel, daß Sie selbst angeschnitten haben kurz
noch ergänzen. Sie haben gesagt mit ( Ihs ?), daß es im 18. Jahrhundert sehr
stark mit dem Zauberer vielleicht in Vebindung war ... Wir haben die
Volksfrömmigkeit heute noch auf den Inseln, die ich neunzehnhundertzwei-
dreiundsiebzig und fünfundachtzig auf Forschungen besucht habe, ist der
Glaube an den (...?) Totengeist nach wie vor... sehr präsent der
zusammenhängt mit einer Art Volksreligiosität, die sich bis heute erhalten hat.
Darauf hat Forster im Jahr 1774 darauf hingewiesen. Danke.
'Discussant: Not a questions but an addition to you Mr. X (=the chair) to your
very valuable presentation...I would like to say that what you have worked out
that the importance of the statements and erm notes {written down} by both of
the Forsters on the Easter Island ...and how you interpreted culture and put it
into the ( culture?) before. The same applies also to the Gesellschaftsinseln
and also applies as a an example of ...and I would then like to add briefly to
the example that you yourself gave. You said that with (ee's ?) ... that in the
18th century there perhaps was a very strong connection with magic We have
this religiousness of the people still today on the islands, when I visited the
islands nineteen-seventy-two ... seventy-three and eighty-five during my
research trips, following the patterns of the belief of overall (?) spirit of the
death ... still existing in connection with the kind of religiousness of the
people that has been kept until today. This is what Forster referred to in the
year 1774. Thank you.'
We must ask ourselves to what extent the discussants can rely on the
conference participants' willingness to see their remarks, questions, etc. as
118 Eija Ventola
relevant and coherent and whether the audience can establish the appropriate
links to understand the point, especially when we are moving from discussing
one paper to another without much warning and prearrangement. Examining
semiotic spanning thus involves looking at how we link with other sources,
with other experiences and their expressions, and still keep what we say as
relevant and focused in terms of the presented paper and the point of the
discussion.
Discussions are examples of dynamic unfolding of discourse. They are
problematic in relation to their papers at least in two ways. Firstly, they cannot
occur independently as texts. If there is no preceding text, the paper, there
cannot be any discussion of it. Consequently, the discussion should always in
some respects be related to the paper and to the whole context of situation;
otherwise it cannot be perceived to be coherent. I have demonstrated how, in
the discussion, links are made far beyond the boundaries - to participants'
semiotic worlds. Often participants also 'prevent' successful semiotic spanning
with their actions. Questions do not get relevant answers, or the presenter
employs 'avoidance techniques', as s/he does not think him/her to be an expert
who can do the semiotic spanning in certain areas of knowledge.
Secondly, discussions have a capacity to develop far from the original
subject matter of the paper. The discussants simply perform long 'self-
presentations' that ideationally may have little to do with what has been said in
the paper - or at least the other participants do not seem be able to follow the
intended semiotic linking. Such comments may lead the discussion off the
intended focus and, consequently, the 'connection' with the presenter's original
messages is lost. Sometimes the discussion seems to 'side-track' dynamically
along its own paths and begins forming its own semiotics independent from the
paper, as almost happens in (6-7). Discussants start making other links to the
previous papers and their discussions, to the texts that conference participants
have read, to the other goings-on at the conference or in the outside world and
so on. It is then naturally the fonction of the chair to bring the discussion back
to its original course. Sometimes the questions asked offer the presenter the
possibility to side-track as well, for instance to present new material. This
happens in Frau XXX's case when one of the discussant asks her whether she
could see herself publishing a complete list of the items in the Forster
collections some time in the future, to which she replies: I actually- in my
paper in the section I didn't read to you is- I have a chart erm where the
collections are like islands they come from and in fact what objects what object
types there are and so....
So far semiotic linking has largely been exemplified in terms of evidence
of linking as seen through the linguistic realisations in the paper and its
discussion. The last example, (8), displays a link made evident to the audience
through the spoken as well as through the visual mode. Frau XXX shows slides
to the audience twice during her talk. Frau XXX has visited the museums
where the Forsters' collections are located. Her experience of what she has
seen is captured in her slides, and here she is trying to share that semiotics with
her listeners. Example (8) displays part of the first SLIDE SHOW.
(8)
So I begin with- to show you some slides of the Pacific collections from the
Forsters and erm especially the collection in the Pitt Rivers Museum ... (27 sec
pause) can you hear me if I stand over here?...I start with Wörlitz. This is the
country house in Wörlitz if any of you have not been there...and these are the little
buildings in which the objects from erm the Forsters who have their own special
little building for the objects made for them and if as I said it is a really excellent
collection and when I visited it was still in quite good condition...This is the Easter
Island hand that was collected. It is a unique piece and now in the British Museum
but was given by the Forsters to erm the British Museum ... it's okay? This is the
Pitt Rivers Museum which is erm as you can see a rather old-fashioned museum
which is arranged by artefact type rather than by area ... And this is the catalogue
of curiosities that was written apparently by George Forster that accompanied the
collection when it went erm to the Asmolean Museum in the 18th century. Some of
the important objects can al- you can only figure out what they were used for by
looking at illustrations. This is an illustration by Hodges and note the little human
images on the top... of the erm I'm sorry ... the er what I wanted you to look at was
the- the- erm the man wearing this wonderful feathered head-dress and the one of
these unique pieces was collected by the Forsters and is now in the Pitt Rivers
Museum ... (7 sec) now I can't move it forward ... and look at the erm human
images and the front of the... erm erm the canoe and here are the two that were
collected by the Forsters both of them given erm to the Pitt Rivers Museum and
these were the only two human images that were collected apparently during the
second voyage.
This is an interesting example because it forces us to focus on the

problematics of combining the visual mode with the spoken mode in the
conference presentation and the role of both of them in indicating semiotic
spanning. We notice that quite different demands for language are in operation
in terms of cohesion and coherence when this section of the paper is realised in
contrast to the parts where she reads the prepared text (cf. (3)). The genre-
register-language analysis actually copes very well with the switches of this
kind. One does not, for example, have to consider the slide show as some kind
of an embedded genre on its own. It is sufficient when we describe the social
process at conferences in its generic element variation. This we do by bringing
in the different possibilities for realising generic elements when we are
achieving goals dynamically - here a choice combination of visual and spoken
120 Eija Ventola
discourse mode, which then puts different linguistic systems at stake in the
production. Frau XXX uses language differently in this part of her talk because
both visual and spoken discourse modes are active. This has several linguistic
consequences. Firstly, the realisations from the IDENTIFICATION systems at
the level of discourse are exophoric rather than endophoric4. The presenter
refers to the entities with exophoric pronouns; the reference relation has to be
interpreted contextually, e.g. this is the country house in Wörlitz. Secondly,
although the experiential content (the artefacts collected by the Forsters) does
not vary, the IDEATION is now also realised differently lexicogrammatically.
The prominent transitivity structure is a relational, identifying process with the
exophoric pronoun being the Token and the nominal group after the relational
process being the Value (see Halliday 1994), as in: this (Token) is (Relational,
Identifying Process) the country house in Wörlitz (Value). Furthermore, in the
slide show the logical function gets realised largely by additive
CONJUNCTIONS, as in: and (additional conjunction) these are the little
buildings... (Note in contrast that Frau XXX's prepared manuscript includes
various other types of conjunctions and metaphorical realisations of logical
relations.) Finally, the interpersonal aspect becomes more prominent in the
slide show - more choices from the NEGOTIATION systems are made and
realised lexicogrammatically: can you hear me if I stand over here
(MODALITY: MODULATION; mental process; Question; polar
interrogative); and note the little human images on the top of the erm... (mental
process; Command; imperative).
The factors of cohesion, coherence, intertextuality, and genre do not
explain, nor were they meant to explain, the ways in which Frau XXX
incorporates the slides which were taken under entirely different conditions in
which originally formed a different kind of semiotics. Through asking
questions about semiotic spanning we may find out what she had hoped to
achieve through this kind of linking and to what degree she was successful
Obviously she felt that this kind of realisation of linking could bring the
audience closer to the 'semiotic world' she had experienced throughout her
research. Yet, in the written paper she is forced to leave this kind of linking
out, and the reader of her article simply gets information about the collected
items in the appendix of her paper, where the items are listed according to the
museums where they can be found. It is certain that the interrelatedness of this
kind needs more detailed exploration, especially when we consider what lies
behind the 'preparing a paper for a conference' and what happens after the
conference when the conference paper is (re)written into a publication, when it
takes on a life of its own by being discussed, reviewed and quoted, and so on.
7 Conclusion
When writing papers, we build various links with other discourses - i.e. the
sources, etc. of the papers. The discussions of papers at conferences then link
with the presented paper or other papers preceding it as well as the discussants'
own experiences of discourses (or other realisations of their semiotic world); a
cumulative increase of perspectives of knowledge and understanding is
achieved at conferences. How 'conference presentation texts and their
discussions' come about seems to involve a complexity of phenomena: reading
of other texts, viewing visuals, making experiments, questionnaires, films,
taking pictures and slides, etc. The process requires a new approach for its
study, and thus semiotic spanning has been suggested as a term that could
cover all of the aspects of linking I have addressed. Linking in conference
situations clearly takes place largely through linguistic realisations, although
frequently visual mode combines with the spoken or the written mode (as in
transparencies or slide shows). However, we also need to see how the papers
exist after their instantiation, how they participate in the semiosis of other texts
of different kinds of genres. We must start to research how different recipients
experience the same information presented in a paper: what is obvious to one
conference participant about the presenter's paper can to another person be
incoherent and devoid of semiotic linking with his/her semiotic world. The
function of a conference is to establish and build common ground. In my view,
as linguists we still have a fairly limited view on how we succeed or fail in
doing this. Extensive research has been carried out on both cohesion and
coherence in individual texts and/or similarities/dissimilarities of texts as
classes of texts. However, very little research has focused on how texts are
actually related, not in the sense of formal or generic relationship, but in the
sense of how they create a social event as a whole and how they build the
social event as a chain of events over time (e.g. how a paper is written over a
period of time and how it lives on later).
One might say we are dealing with something like an internet hypertext:
the way in which, at a conference, someone else has previously conducted the
research and built up the text resembles the text readers/users create when they
do the linking in internet. The presentation of the paper at a conference is
similar to that of an internet-user getting the first text on the screen. As soon as
the text starts unfolding, the conference participants start making their own
links. They relate what they hear to their own individual 'semiotics', and where
one establishes links, another may see no reason to establish any links at all
(and thus switches off). This process of semiotic spanning influences the flow
of the discussion: what questions are eventually asked and what comments are
made. It may also influence the next presenters, because they, too, have been
122 Eija Ventola
building up links in their thinking with what they have heard. Consequently
their 'prepared' text may go through radical changes during the oral
presentation. This kind of semiotic spanning on the part of the discussants and
the presenters actually results in a very interesting and innovating conference,
which ultimately motivates the participants to carry on their spanning on their
own. On its own each paper may not have a dramatic effect on the conference
participant, but what makes the conference valuable is the effect of the
interplay between all of the papers and their semiotic spanning within and
outside the social event of a conference. This spanning is cumulative as the
conference proceeds from one paper to the next, but due to the dynamic nature
of unfolding of papers and discussions, this aspect of conferencing has, to my
knowledge, not been researched at all.
Frequently we find that the sense of semiotic spanning created during a
conference during the papers and their discussions (formal and informal) is not
of lasting nature. Actually most conferences merely create an illusion of
temporary understanding and unity of an academic discourse community. Even
immediately after the conference the links between discourses by the members
of the community are reformed and changed. The conference situation creates
a momentary feeling of intellectual companionship and sense of common
understanding and experience, but unfortunately its effects may, remain very
short-lived and local. Yet, one could perhaps enhance the effects by closer co
operation in the publication and planning of future events around the theme of
the conference. When papers are published their effect should be cumulative;
in other words, they should influence the thinking of the conference
participants and other scholars and help them to understand the phenomena (of
whatever kind) around us. But ensuring that effect is practically impossible in
the publishing world.5
What remains to be said is that, when viewed semiotically, conferencing
remains an unexplored territory. Although not entirely defined and exemplified
in this paper, the term semiotic spanning has been suggested in order to allow
linguists and researchers to explore different kinds of relationships between
different social activities, activities which are sometimes not closely related,
yet influence and enhance our thinking in other fields.
Notes
1. The asterisks and the underlinings indicate simultaneous speech.
2. For a detailed discussion, see e.g. Ventola 1987; Martin 1992, especially Section 7.3.
3. For theory and analyses of transitivity structures and other kinds of verbal processes, see
Halliday 1994.
4. See Martin 1992 for a reference to these systems.
5. A good conference publication should also perhaps resemble a hypertext in the sense that
there should be enough textual and semiotic spanning between the texts. Too frequently
conference publications are not built upon semiotic spanning and, consequently, the
publishers frequently are reluctant to publish conference volumes. They do not tend to sell
well because, out of the many articles, the potential buyers may only be interested in one
or two articles; i.e. as a whole, the conference volume does not form a 'semiotic whole'.
Today it is simply too easy to photocopy the few articles that one is interested in the
conference volume. One does not need to buy the whole book (or one gets the articles
from the internet). To create a publication that enhances 'semiotic spanning' is very
demanding for the authors and the editor. What the editor can naturally do is to put the
articles as they come in on his/her homepage and then ask the authors to rewrite their
papers by establishing necessary links with the other papers. Ultimately, then, the readers
will make their own connections when they read the final product.
References
Beaugrande, R. de 1980. Text, Discourse and Process. London: Longman.
Beaugrande, R. de and Dressier, W. 1981. Einführung in die Textlinguistik. Tübingen:
Niemeyer.
Eggins, S. 1994. An Introduction to Systemic Functional Linguistics. London: Pinter.
Halliday, M.A.K. 2 1994. An Introduction to Functional Grammar. London: Arnold.
Martin, J.R. 1992. English Text: System and Structure. Amsterdam: Benjamins.
Ventola, E. 1987. The Structure of Social Interaction. London: Pinter.
Coherent Keying in Conversational Humour:
Contextualising Joint Fictionalisation
Helga Kotthoff
Universität Konstanz
1 Introduction
Conversational humour is chiefly situated practice, resting on inference-based

interpretations. Although most speakers cannot explain how they use prosody,
wording, mimics, etc. to make utterances humorous, they produce coherent
episodes which often have a generic structure. I suggest an empirical discourse
analytic procedure for the study of humour. In the area of humour research
particularly, the focus on standardised, written jokes has been responsible for
our limited knowledge of the complex interaction of verbal forms, background
knowledge, and social and contextual factors. In many linguistic approaches
the humorous is reduced to the lexico-semantic punchline (Raskin 1985; Ruch,
Attardo and Raskin 1993). It is conspicuous that many typical oral features are
missing in written jokes. Such features include code switching, social stylistics,
poetic features, repetition, marked wording, prosody, sighs, laughter, etc.,
which Cook-Gumperz and Gumperz (1976) refer to as "contextualisation
cues." These cues index the constantly changing contextual presuppositions on
which situated interpretations in oral discourse depend. I favour adopting an
interactional perspective, because humour is basically a locally structured,
social phenomenon; especially in informal discourse interactants select various
strategies to evoke unconventional, comical perspectives.1 In situational
humour 'being there' becomes a very important part of 'getting it' (Boxer and
Cortés-Conde 1997). As discourse analysts we have to reconstruct what 'being
there' means.
Not just topics, but also the modalities of discourse topics are jointly
constituted (and changed) in real time.2 Above all in conversation analysis,
empirical pragmatics, and interactional sociolinguists coherence achievement
has been shown to involve a turn-for-turn interplay of speech activities and
reactions (cf., e.g., Bublitz 1989; Jucker 1993; the studies in Dorval 1990).
126 Helga Kotthoff
Laughter particles in utterances are important modality markers; they

index that the text is to be interpreted as humorous. Particularly through the
works of Gail Jefferson we have learned that laughter in conversation is not
just a reaction to something amusing or 'funny', but actively creates amusing
(funny) perspectives on what is said. It can have a metapragmatic function3 as a
kind of commentary, as if to imply, "What I am saying is not totally serious."
The recipient is thereby invited to create a second perspective on what has been
said. Beyond this, speakers co-construct various genres which can be described
as teasing, mocking, anecdote, irony, banter, or joint fictionalisation.
Conversation analysis and the ethnography of speaking deal with humour
and joke forms in natural surroundings. Forms of lexical, mimic and prosodic
marking of the humorous interaction modality have been studied (Schütte
1991; Müller 1992; Norrick 1993; Kotthoff 1996b). The concept of interaction
modality (Kallmeyer 1979) is of central importance for the analysis of
conversational humour. It corresponds to the concept of keying which is
employed in the ethnography of communication (Goffman 1974; Hymes 1974;
Straehle 1993). Pathos, exaggeration or humour can be regarded as examples
of keying; they contribute to the creation of a specific text understanding
(Kallmeyer 1979). Interaction modalities are procedures which regulate the
particular reality- and coherence relations of utterances. In humour the
relationship to reality is loosened and special inferences are expected for the
purpose of creating "sense in nonsense", to use Freud's expression
(1905/1985).4 Loosening relations to reality means widening the possible scope
of imagination. This could mean that in the humorous modality coherence
becomes weaker. But this seems true only at first glance. At second glance,
coherence seems to be governed by principles other than those governing
ordinary discourse, namely principles of punchline creation and of creating
comical perspectives.
Utterances are marked as humorous through special contextualisation-
procedures. Laughter particles inserted in utterances, prefacing or ending them,
are thereby particularly important. Contextualisation cues steer cognitive
inferencing into unusual bisociations of perspectives.5 Serious and non-serious
modalities of interaction should be analytically separated from one another,
although both modalities include a wide range of activities which do not
always belong specifically to only one or the other mode (Mulkay 1988); they
can be held open strategically (Kotthoff 1996a).
2 Punchlines, humorous keying and relevance

As many articles in this book underline, there are contextual differences in
coherence standards. I am interested in coherence-forming strategies in
Coherence and Conversational Humour 127
humorous discourse because they are more creative than in serious discourse,
both in production and reception. Humorous utterances are often not what is
contextually expected; violating normative expectations is indeed an essential
characteristic of humour. In conversational humour we quickly find extremely
complex inferential paths used to create "sense in nonsense". In explaining
what happens in humorous discourse, we must employ conversational
inferencing in the sense of Gumperz (1982) (i.e., 153: "the situated and
context-bound process of interpretation, by means of which participants in an
exchange assess others' intentions, and on which they base their responses")
and Cook-Gumperz and Gumperz (1994), based on contextualisation
procedures.
The keying of an activity as "humorous" is largely accomplished by the
use of marked prosodic cues, in conjunction with laughter, rhetorical
procedures of surprise organisation, pragmatic incongruities, shifts in
politeness standards (Kotthoff 1996a), and a general decrease in truth-value.
Through the above-named processes, speakers index a play-space: Bateson
(1954/1972) has shown that even primates can distinguish between a serious
bite and a playful one. Play framings are highly significant in ascribing a
humorous keying to an event. Keyings influence the reception of an activity;
they are closely related to affect management and the manipulation of truth
values.
Many theories of discourse coherence work with the notion of relevance
(e.g., Grice 1975; Dascal 1977; Sperber and Wilson 1985; Jucker 1993).
Discourse is coherent if relevance can be constructed for its parts. Sperber and
Wilson (1985) define relevance in terms of contextual effects and processing
effort. Contextual effects are achieved when newly-presented information
interacts with a context of existing assumptions. In humour, newly-presented
information often does not interact with the context of existing assumptions;
otherwise it would not be surprising (Mulkay 1988; Kotthoff 1998).
Furthermore, Sperber and Wilson theorise that the greater the contextual
effects, the greater the relevance, and also the smaller the processing effort, the
greater the relevance. Hearers pay attention only to information that enriches
their set of assumptions at low cost.6 However, we find that in humour listeners
must make an extra processing effort, which means that their costs are not low.
Why then are they willing to pay these extra costs?
It is necessary to distinguish between jokes which depend on punchlines
(German Witzigkeit) and comical discourse (German Komik). Only punchline
joking permits written representation. Because it is strongly based on lexico-
semantics, its humour potential can be inferred within the text. Everyday oral
humour often combines punchlines and comical keying. Punchline-joking is
based on lexico-semantically induced cognitive inferencing, the comical works
128 Helga Kotthoff
with keyings which play down the seriousness of utterances and modify their
reception, thereby influencing cognition. When speakers cannot rely on lexico-
semantically steered cognition alone, the humorous must be contextualised on
the level of performance. Contextualisation focuses on the performance7 aspect
of humour. Our attention, as researchers, thus shifts from the structural analysis
of bounded texts and lexical semantics to situated discursive practice. In
informal humour discourse, special contextualisation procedures lead to special
cognitive inferencing through which coherence is constructed for the
humorous. Comical discourse alludes much more to exclusive group
knowledge, as readers of transcripts can testify. Often they can only understand
humour potentials after being given background information by the researcher.
Both forms of humour demand extra processing efforts. Punchline-joking
works with the surprising bisociation (Koestler 1964) of frames or scripts;8
comicality works with double perspectivation (more on this below).
Let us look at one sort of effort typical of humour based on creating
punchlines.
He: Your nagging just goes in one ear and out the other.
She: That is because there is nothing in between to stop it.
Normally, a metaphor such as the one used by the man is understood

holistically as having a conventionalised meaning: Your nagging doesn't
impress me. The woman shifts the conventionalised frame (extra effort) by
taking the utterance literally. This makes sense, although not in the expected
way. The listener of the dialogue can build coherence with the extra effort of
constructing an unusual context for the first utterance, the context of literal
meaning. The extra processing effort is worthwhile, because the relevance is
upgraded: The answer makes even more sense than a "normal" answer,
because it is a quick repartee and tops the unfriendly intention of the initial
utterance. For the female speaker and a sympathetic audience, the woman's
response is a source of fun and laughter (worth the extra processing cost).
Sperber and Wilson (1985) do not deal with humorous relevance
construction, but they could theoretically integrate it into part 6 of section 4 on
"implicatures and style: poetic effects" (217 ff.). There they deal with
"emphatic effects" of style features, such as repetition, parallelism or
explicitness/ implicitness. From the point of view of relevance theory, the extra
processing effort incurred by stylistic or poetic features must be outweighed by
some increase in contextual effects triggered by the features. Poetic effects can
be too salient to go unnoticed; they may require an effort of imagination to
construct non-stereotypical assumptions. Sperber and Wilson (1985/1995: 224)
write that taking the hearer beyond standard contexts and premises is typical
for poetic effects. The same can be said for humorous effects and might be
integrated under the rubric of aesthetic pleasure.
There is agreement among coherence researchers that we accept extra
efforts if we can thereby expect to gain extra effects such as joy, aesthetic
pleasure or enlightenment. But in the literature this is chiefly reserved for the
'higher' arts.9 Thus Graesser, Person and Johnston (1996) consistently limit
aesthetic text dimensions to canonical literature. They cite numerous authors
who support the following thesis: "According to the polyvalence assumption, a
reader constructs multiple interpretations of a sentence or excerpt in a literary
text, whereas there is only a single bona fide interpretation in a nonliterary
text" (1996: 9). My article questions this delimitation. Or expressed differently:
Everyday discourse cannot be equated with nonliterary texts. Humour is a form
of oral art integrating many aesthetic strategies.
Like supporters of the contextualisation theory, Sperber and Wilson
(1985: 137-140) disagree with the traditional view of context as given prior to
the act of communication. They assume that processing new information
involves the selection of an adequate set of background assumptions which
constitute the context. They do not, however, sufficiently clarify the manner in
which a frame of interpretation is jointly set up for the rest of an utterance's
linguistic content. Taking conversational processes of contextualisation into
consideration is a way to enrich the cognitivist approach by clarifying the way
inferencing is steered in talk-in-interaction. Sperber and Wilson's approach to
communication and cognition is similar to Grice's (1975), in that they
underestimate the surface form of utterances. Humorous discourse illustrates
how implicatures and inferences depend on the concrete surface form of
utterances.
Conversation analysts have made valuable contributions to the analysis
of the concrete surface of speech activities. They focus on such questions as
when precisely people laugh in conversation, how this changes discourse and
to what extent laughter represents an orderly conversational activity
(Schenkein 1972; Sacks 1974; Jefferson 1979, 1985).
I would like to show the peculiarities of humorous coherence in a genre
which is not categorised in everyday discourse. We speak of such activities
simply as joking; and yet they display patterns which can be uncovered from a
scientific etic perspective.10 It is central for the coherence of the genre of "joint
fictionalisation" to extend the initiated humorous potential, to extract more and
more new comical aspects from the preceding talk. It has been variously
argued that there is not just one principle of coherence formation, but several
(Giora 1997). My work supports this claim. The coherence of keying and the
co-construction of the comical are special dimensions of oral humour.
130 Helga Kotthoff
When speakers jointly construct absurd fictions, their coherence

formation is specific to the time spent within a given frame. When a speaker
fancifully leaves the realm of reality, another speaker often joins in to co-
construct fictions or even top the absurdity. The second speaker expands on the
specific potential of the preceding utterance. Every utterance is characterised
by a diversity of potential relevancies. It is interesting to recognise how
speakers work together to expand a coherent realm of absurd fiction.
3 Discussion of Data 1
Let us examine a conversational episode which contains joint humorous
fictionalisations. The episode is taken from a corpus of twenty dinner
conversations among friends taped in Austria, Germany and Switzerland in
1994-1995. The data is analysed using conversation analytic techniques
enriched by ethnographic methods (Kotthoff 1998).11 Some of the joking
activities have recognisable patterns, others do not. Overt joint fictionalisation
is regarded here as a joking genre if comical perspectives or punchlines are
created in its course.1 But there are naturally also covert and pre-patterned
fictionalisations in other forms of joking. Every standard joke is fictional;
already in the initial stage of telling a joke it becomes clear that the
interlocutors are about to enter into a closed fictional world with its own
coherence rules (Sacks 1974). Joint fictionalisations, however, leave the realm
of reality step by step, one fictionalisation building the ground for next, thereby
enhancing comicality.
The following episode was taken from a conversation recorded in a post
graduate student group in Vienna. In conversational joking we must always
share the background knowledge of the interlocutors in order to understand the
humorous potentials (this is part of coherence work). The group is chatting
about Hermes Phettberg. 13 At the time the recordings were made, (1995),
Hermes Phettberg was a very popular television moderator in Austria and
Germany. 14 His popularity was based partly on the fact that he was unusually
corpulent, especially for a media personality; he was also a confessed
homosexual masochist and appealed to a more intellectual public.
(1) (Conversation 19 Episode 9)15
Conrad (C), Hugo (H), Lilo (L), Peter (P), Renate (R), two or more persons (m)
1 C: wieviel Kilo der hat, waaß a kana,

nobody knows how many kilos he weighs,
2 wieviel Kilo der wirklich hat.

how many kilos he really weighs.
3 R: hundertvierundsieb[zig?
hundred forty-sev[en?
4 P: [jenseits der zwahundert sicher.
[over two hundred surely.
5 H: [(? ?)
6 P: [i mein, des
[I mean, this
7 C: [i bin ka Bravo-Leser mit SteckbhefsammluH [Ung.
[I am not a Bravo-reader with a collection of fan cards
8 L: [HE[HEHEHEHE
9 m [BEHEBE
10 P: was?
what?
11 H: woher weißt du das über[haupt.
how do you know that anyway.
12 P: [na, aber zwahundert des könnt
[no, but two hundred could
13 wirklich sein, ja
be possible, yeah
15 C: amal hob is glesen.
once I read that.
16 P: ↑des wär was. ↑Hermes Phettberg lebensgroß.
that would be something. Hermes Phettberg life-sized.
17 ↑Sta:rschnitt in der Bra:vo, [na?
star cutouts in Bravo, [huh?
18 m: [HABABABA[BABABABABA
19 L: [kommst a Jahr lang aus.
[printed for a whole year.
20 H: na, na im Playgirl, no viel besser
no, no in Playgirl, much better
21 P: a Jahr
one year
22 m: BABABABAHA
23 L: als eine Ausgabe fünfundzwanzig [Meter hoch. BEBEBEBEBE
as an edition twenty-five [meters high. HEHEHEHEHE
24 P: [der fallt eh net
[he doesn't come under
25 unters Jugendverbot, =
the minimum age restrictions=
26 L: BEBE
27 P: =weil Genitalien sichst bei dem eh kane
=because with him you don't see genitals anyway
132 Helga Kotthoff
28 durch den Bauch, also

because of his belly, well
29 C: na der kann nackert ruhig sein, [des des
no it's all right for him to be nude [the the
30 H: [der kann ruhig
[it's all right for him
31 nackert sein.
to be nude.
32 C: malen nach Zahlen. [Hermes Phettberg zum selber malen.
painting by numbers. [Hermes Phettberg to paint yourself.
33 P: [das Phettberg Puzzle.
[the Phettberg Puzzle.
34 L: HAHAHA pfui Teufl.
HAHAHA yuck.
35 C: da hast a Lebensaufgabe.
that's a lifetime job.
Before the start of the transcript, someone had commented that due to his
excessive weight, Hermes Phettberg would never live to be seventy. He was
thus selected as a topic, but not necessarily for a humorous modality. Conrad
agrees in lines 1/2 and notes that no one knows how much Phettberg weighs.
Renate, with a slight question intonation, makes a concrete guess. Peter in line
4 offers an even higher estimate. So far the talk is serious. In line 7 Conrad
makes a statement about himself which demands extra coherence work to fit it
into the topical context. The last word contains a laughter particle. Some
participants react with laughter (8, 9), which suggests that they have processed
Conrad's statement as comical. Coherence is formed by them, although a bona
fide understanding of line 7 would not make sense. Conrad is referring to a
youth magazine, Bravo, which publishes so-called Steckbriefe (which in
English could be called "celebrity trading cards") containing all sorts of
information about film, pop and rock stars and celebrities. Here we have an
unusual combination of elements from the real life world. Phettberg does not
fit at all in the sensationalistic Bravo, which does not deal with intellectual and
unconventional personalities such as Phettberg. The interactants can thus create
a perspective in which it is funny to imagine Phettberg in Bravo. Line 7 does
not create a classical punchline, however. There is no sudden frame bisociation
with the typical implicit possibility to construct meaning in a frame that has not
been developed yet (Koestler 1964; Norrick 1993; Attardo 1994). Combining
Phettberg and Bravo sheds an unusual light on both, as will be discussed
below. Peter does not understand something (perhaps he could not process the
comical dimension), Hugo's question in line 11 is directed at Peter, who in line
4 made an assertion about Phettberg's weight. In the following Peter softens
his assertion somewhat. In line 15 Conrad agrees with him. Up to here some
interactants have reacted in a way that vaguely suggests they have been
constructing a sense for line 7. Other speakers have elaborated the subtopic of
Phettberg's weight. In multiparty talk it is not essential that everybody
understands everything at all times. We see that Conrad's remark in line 7 is
greeted with laughter and is then otherwise passed over.
In lines 16/17 Peter returns with fantasising to the topical area of Bravo.
He draws on a topical potential which Conrad introduced in line 7. In lines
16/17 we witness special keying procedures. The phrase des wär was/that
would be something fonctions as an introduction to something which is marked
as unusual. The turn is syntactically and prosodically subdivided into three
phrase units which all have the same rhythm and intonation. The accent is on
the first syllable in each phrase; the intonation falls at the end of each phrase.
Rhythm and intonation shifts can indicate a new context.16 Here they go
together with semantic comicality; in this way a humorous keying is set up
which is not totally new in the context, but is intensified. The youth magazine
is known for celebrity cutouts (Starschnitt), in each edition sections of a picture
of a celebrity are printed, and readers can gradually piece together life-sized
star portraits like a puzzle. This sort of world knowledge is activated here.
Extended laughter by several persons in line 18 shows that something
considered funny has been processed.
Let us look at some details of formulation. In lines 16/17 Bravo is
introduced in headline style. This means that elements of Bravo (a journai has
headlines) are used not only to denote this magazine semantically, but also
simultaneously to stylistically evoke it. The syntactic and prosodic form itself
creates the semantic content via iconicity. This "likeness on several levels"
(Jakobson 1960: 369) characterises aestheticised speech.
Furthermore, the continuation of a very elliptical speech style is striking.
We talk about ellipses if at least one part of a phrase is omitted that can be
completed contextually. Ellipses usually contain the rheme, the new
information (Schwitalla 1997). In line 20, for example, the suggestion to put
Phettberg in Playgirl, a magazine takeoff on Playboy which features photos of
nude or scantily-clad males, as well as the evaluation no viel besser/much
better, are quasi brief spotlights on an already-created stage; lines 21 and 23
also cohere in form and content with lines 16/17.
The presented fictionalisation is absurd, since in the real world Phettberg
absolutely would not come into question as a teenage idol to be popularised in
Bravo. In line 18 the women in the group laugh. Lilo comments on the fantasy
that in the case of Phettberg it would take an especially long time to collect all
the parts (a Jahr/a year); she is alluding to his enormous girth. Through their
demand for active coherence formation, allusions (Wilss 1989) contribute to
further aestheticising the discourse. The active reception here shows that the
134 Helga Kotthoff
comicality was not only understood but built upon. In conversational humour
recipients directly participate in further shaping the talk-in-interaction as
humorous. Humour and fictionalisation thus become an ongoing joint
achievement. In line 20 Hugo further intensifies the fiction about Phettberg in
Bravo in the topical direction of Playgirl. Previously created images are now
rounded out with specific details. Peter would like to see Phettberg presented a
year long in this magazine as well. Several persons laugh (22). In line 23 Lilo
enlarges the life-size poster to one twenty-five meters high and laughs at this
absurdity.
Starting in line 24 Peter takes up a different aspect of Phettberg's girth
and erotic self-display. The star's vast, sagging belly conceals his genitalia
from view. Conrad and Hugo affirm this image. Phettberg and 'minimum age
restrictions' are a new combination which is now developed by other
interlocutors. Conrad in line 32 alludes to a children's pastimes: Malen nach
Zahlen/painting by numbers11 Peter formulates a further fictionalisation from
children's pastimes {das Phettberg Puzzle). Lines 32 and 33 again display
headline-style. The games are presented like an ad. Lilo laughs and utters an
interjection of disgust. Conrad, in pointing out that it is a
Lebensaufgabe/lifetime job, once more alludes to Phettberg's enormous girth.
All the fictionalisations draw on a common knowledge of entertainment
media, thereby making coherence easy. The interactants successively intensify
the absurdity. The fictionalisations have a meta-message: Hermes Phettberg,
who with his "Nette Leit-Show'V'Nice People's Show" 18 carries on an
entertaining, witty type of media conversation, is relentlessly marketed in the
fantasies of the young Viennese. His body shape is the pivot point for
numerous ideas. The interlocutors do something Phettberg himself often does;
but they do it so-to-speak in a diametrically opposite manner. In numerous
interviews, Phettberg has referred over and over to his body, body feel and
sexuality. He achieved his popularity to a considerable degree by staging
himself as an anti-type. Phettberg contradicts several norms of the yellow press
world. He notoriously presents himself in interviews as overweight, unkempt,
homosexual and masochistic - thus aiming at a shock effect which appeals to
certain segments of the intellectual public. It is amusing to imagine him
integrated into the yellow press world as though he were a typical TV star or
celebrity. Thereby the Viennese interlocutors also communicate that they find
Phettberg's self presentation inconsistent and contradictory. Consequently,
distance can be simultaneously displayed toward both Phettberg and the yellow
press. The participants indirectly show both their knowledge of media contents
and also their critical attitude to them without explicitly making an evaluation.
By means of joking the evaluation is jointly performed rather than denoted.
Here the norms of the magazine world are violated and, at the same time, social
norms of self-presentation are negotiated. I regard implicit norm negotiation as

a metapragmatic function of humorous communication in the sense of an
indirect indexing of shared values in orientation to each other. As Gumperz
(1996: 366) writes, indirect (not overtly lexicalised) signalling mechanisms are
for the most part culturally or subculturally specific.
It is apparent from the joint fantasy enhancing of the interlocutors, that
not only has the topic been inferred and developed, but also the direction of the
fictionalisation and combination of elements.
As we have seen, selection and humorous combination occur when the
topicalised TV moderator, Hermes Phettberg, who is not an attractive person, is
juxtaposed with a media context (Bravo), where positive appeal and
appearance are marketed. This combination opens up a new perspective for
viewing Phettberg. He is imagined as a life-sized idol, as a poster hanging on a
wall. On the one hand, Phettberg contradicts the usual criteria for attractiveness
and, on the other, he constantly thematises his private life in the press. By
constantly revealing intimate details of his life to the press, he behaves like
many typical TV celebrities. The humorous discourse permits the participants
to jointly produce an implicit evaluation of Phettberg's social practice, which
is, however, kept vague. It is an advantage of the humorous that such
evaluations cannot be easily pinned down and are not intended to be.
Fictionality can be optionally combined in humour with other procedures
of oral art, e.g., allusion, vagueness of attitude, rhetorical stylisation, dialogue
dramatisation (which will be illustrated in the next episode), sound parallelism,
and creative word play.
In order to explain how humorous discourse is understood, we need a
model of conversational inferencing which goes far beyond semantic
explanations of how a joke is processed. Linguistic humour theories which
view the written joke as a basic model are unsatisfactory, as they totalise text-
intrinsic lexico-semantic script-oppositions or incongruencies as the chief
factor of the humorous. Script oppositions play a role in conversational
humour, but an integrated one - seldom an exclusive role.
The genre of joint fictionalisation involves creating unusual linkages
between referents, which are jointly developed. The fictions are not fully
formulated, but rather alluded to and evoked. The space for fantasising is
thereby kept open. The turn-taking is so open that everyone can at any time
insert a further intensification or shift of focus into the humorous fictions.
There is no temporally coherent narration, but rather coherent image creation.
Continual laughter, unusual perspectives, concrete images, and high
involvement show that humour is being produced. Furthermore, details - such
as that people cannot see Phettberg's genitalia - are welcome. Evaluations can
introduce the closing of fictionalisation, as occurs in lines 34 and 35.
136 Helga Kotthoff
Instead of working with surprise punchlines, the humorists here develop

unusual perspectives (points of view) starting from everyday interaction. These
perspectives result in quite new, complex meanings, as I have tried to show.
Comical perspectives are built up systematically with the aid of various
framing procedures. Recipients show in perspective-taking how they interpret
the perspective and how they relate their own perspective to it.
I do not maintain that the comic potentials of utterances are always built
upon. The range of keyings need not always extend over several sequences, but
can be limited to a single turn or even a single phrase. First, some contexts do
not permit the sort of free topic development that we see in the data discussed
here. Second, the topic of joking remarks can also be seriously pursued. It is a
sign of mutual familiarity when a group succeeds in jointly producing
humorous fictions in a longer period of talk. Keyings are valid as long as they
are marked by clusters of indexicals which create them.
4 Discussion of Data 2
The next episode of humorous joint fictionalisation was taped in the same
Viennese group. Lilo begins to tell a story from her childhood in a marked way
with interspersed laughter particles. Laughingly she describes some difficulties
her parents had in feeding her. These are removed from the domain of the
realistic through picturesque exaggeration {einmal im Monat t Küche
ausgwaschen/once a month scrubbed the kitchen from top to bottom). Someone
in the group immediately reacts with brief laughter.
(2) (Conversation 20 Episode 9)
Conrad (C), Elisa (E), Gerda (G), Martin (M), two or more persons (m), Lilo (L),
Peter (P), Sabine (S)
1 L: wie ich klein woa, ham meine Eltern einmaHAl

when I was small, my parents
2 iHlm Monat t Küche ↑aHAusgHwaschen.
scrubbed the kitchen from top to bottom once a month.
3 ?: HEHE
4 L: meiHene Mutter, ich weiß nicht, die hat a schlechte Technik ghabt.
my mother, I don't know, had a poor technique.
5 augenschHEeinlich.
obviously.
6 sie hat mich scheinbar immer schlecht hingsetzt.
she apparently always sat me down poorly.
7 nämlich immer mim GesIHT zuHUr WaHAnd HAHAHA. HEHE

namely always with the face to the wall.
8 ?: HEHE
9 M: ja des is innaKlaanKüche a Problem, ne?
yeah that is a problem in a small kitchen, huh?
10 L: ja.
yes.
11 (1.0)
12 M: °is dann besser im Badezimmer futtern. °
is better to feed in the bathroom.
13 L: ja
yes
14 (--)
15 : also ich hätt mich im Badezimmer gefüHttert,
well I would have fed me in the bathroom,
16 weHEnn ich meiHEne MuHtter gweHEsen wäHr.
if I had been my mother.
17 ich hätt ihr, ich würd ihr das auch nicht ↑übelnehmen.
[°im NachhineHEin. °
I would have, I wouldn't have held it against her
later on.
18 S: [am besten in die Dusche ste[llen.
best of all put in the shower.
19 ?: [HEHEHE
20 P: ne Duschkabine, jaHE.
a shower stall, yeah.
21 S: [und nach dem Essen (? schnell die [Dusche anstellen?)
[and after eating (?turn on the shower quick?)
22 C: [oder no besser bei laufendem Wasser, eigentlich.
[or still better with the water running, somehow.
23 m: [HAHAHAHAHAHAHAHAHA
24 M: in die Duschkabine, ja
in the shower stall, yes
25 P: so wie in die Labors mit der Glaswand
like in laboratories with glass walls
26 C: [hast alles wieder ausgspuckt oder was?
[did you spit everything out again or what?
27 L: [meine Mutter hat den Löffel hineingestopft und gfragt,
[my mother stuffed the spoon in and asked,
28 schmeckts denn? ich hab angeblich an breites Grinsen
does it taste good? I apparently put on a broad grin
29 aufgesetzt, gsagt, ömhöm, pfffff HEHEH[EHEHEHEHE
and said, omhom, pfffff
30 m: [HAHAHAHAHA
138 Helga Kotthoff
31 P: über die Lautsprecheranlag. a Löfferl fürn Phapa,

over the loudspeaker system. one spoonful for papa,
32 a LHöfferl für die Mama.
one spoonful for mama.
33 (--)
34 S: kann i ma garnet vorstellen,
I cannot imagine
35 [daß du des Essen alles ausgspuckt hast.
that you spat all the food out.
36 L: [gell?
can you?
37 M: [und der rote Rübensaft schmückt auf einmal die weiße Wand.
and the red beet juice suddenly decorates the white wall.
38 S: na, des versteh ich sowieso nicht, warum man Kindern
well, I don't understand that anyway, why people
39 nicht nur was Weißes zum essen [gibt. HEHEHEHE
don't give children only white things to eat.
40 m: [HEHEHEHEHEHEHEHE
41 M: aber immer Spinat. =
but always spinach. =
42 C: Gipsbrei
plaster porridge
43 M: =unhamlich grün.
=incredibly green.
44 G: intravenös am besten.
preferably intravenous.
45 P: rote Rüben hab ich nie mehr gekriegt oder sonst was,
I have never gotten red beets or anything like that again,
In lines 1 to 7 the topical context is not explicitly stated to be that of

feeding a child, but since prior to this episode the conversation centred on plant
food, the topic of feeding can be easily inferred. The expression Gesicht zur
Wand/face to the wall evokes an amusing incident; laughter particles increase.
One can immediately imagine the child Lilo spitting her food out on the wall.
By laughing loudly Lilo herself contextualises the evoked image as funny. One
person from the group laughs (8). Martin, however, in line 9 reacts to serious
content potentials: he takes into account the limited space in the kitchen. We
see here that humorous keyings need not be immediately confirmed. Lilo
accepts this; a short pause ensues. Then in line 12 Martin makes a novel
suggestion, collaborating on the topic combination of feeding and space limits
(feeding the child in the bathroom). This affirms the keying introduced by Lilo.
His fictionalisation evokes an unusual scene, but without employing a
punchline. Lilo agrees laughingly (13). After a brief pause she indirectly
portrays herself as a difficult child. Sabine's suggestion in line 18 intensifies
Martin's idea of feeding children in the bathroom. Her formulation is strongly

oriented to Martin's in line 12: she produces a similar elliptical construction,
leaving out not only the pronoun, but also the finite verb. We see not only
content, but also format tying here. Sabine raises Martin's comparative better
to the superlative best. Someone laughs (19). Peter, also using an elliptical
form, makes Sabine's comment even more specific (20), in accordance with
the topical component of convenience in cleaning up after feeding. Sabine
herself then adds a further elaboration of the suggested fiction, leaving out the
subject and the finite verb. The subject is omitted in all the phrases up to line
26 and again starting at line 31. Deleting themes is a strategy used to direct
attention to rhemes. Ellipses are not just economical means of omitting
syntactical elements listeners already know, but are also rhetorical strategies
for focusing rhemes. Schwitalla (1997: 71) writes that we often find ellipses at
the start of increased detailing in narration. In data 2 the elliptical constructions
also present details. Each elliptical phrase visualises a pictorial detail, thereby
enhancing the absurdity of the fantasies. It is unclear whether the imagined
subject of the ellipses is Lilo in particular or children in general. Several
participants laugh (23) after Conrad's suggestion to feed children (or Lilo?)
with the water running. Martin affirms Peter's and Sabine's suggestions. In
line 25 Peter augments the fiction. Conrad, returning to Lilo's narration, asks
her a question which she answers beginning in line 27. She describes further
incidents from the days when she was spoon fed. A sound drawing iconises the
activity of spitting food out (29), ending in laughter; the listeners join in. Lilo
puts on stage an imaginative scene from her childhood. With his utterance in
31/32 Peter seems to embroider the fantasy scenario created by his comments
in line 25 and Lilo's narrative. He specifies a technique with which parents
could talk to a child sitting behind a glass wall in a shower and then offers
stereotypical parental talk as constructed speech, without making the cited
speakers explicit. The words are spoken as though reported, but are actually
constructed, as Tannen (1989: 109 fif.) and Couper-Kuhlen (this volume) have
emphasised for reported speech in general. Citing ideas as stemming from
somebody other than the speaker herself is a discourse strategy for enhancing
involvement. Constructed dialogue enables listeners to create their
understanding by drawing on their own history of associations. The little
scenario Peter presents creates a vivid contrast to Lilo's narration where the
mother did not speak over the loudspeakers.
In lines 34/35 Sabine appends a commentary which alludes to the idea
that Lilo likes to eat; this is a form of teasing19 to which Lilo reacts with a
particle equivalent in sense to "isn't it astonishing?" In this context this is a
form of agreement with Sabine's doubt. Martin (line 37) suddenly shifts to
Standard German (not Viennese), seeming to play at being a news announcer.
140 Helga Kotthoff
He formally presents details of the scene of throwing food at the wall. His use
of alliterations is striking (weiße Wand, roter Rübensaft). Line 37 is again
constructed speech. With a remarkable register change Martin indicates that he
is not speaking in his own voice.20 He seems to be commenting on the scene
described by Lilo. It does not matter who would speak like that in what
situation, and thus does not create a coherence problem. The following
utterance by Sabine is wonderfully related because Martin has focused on the
colour of the food and the contrast it makes on the wall. Starting in line 38
Sabine claims not to understand why people don't simply give children only
white things to eat. She solves the problem of the color contrast absurdly. This
is greeted by laughter. Martin thereafter confirms Sabine's lack of
understanding (41) by presenting yet another colour contrast (43).
In line 42 Conrad's fantasy moves further in the direction of black
humour: it can be inferred that plaster porridge would not contrast with a white
wall. Gerda augments the black humour (44); her suggestion would even make
feeding unnecessary. No one laughs at this, however. While it may be that
black humour is enjoyed in a rather "dry" manner (without laughter), perhaps
the humorous potential of the topic has simply been exhausted by this point.
Peter says, with a slightly complaining tone, that he has never been served red
beets since his childhood. The topic is gradually moved from a humorous
modality back to a serious one. Several persons have participated in the
amusing joint fictionalisations.
Both episodes have shown that it is completely unproblematic to insert
fantasies into discourse which are not yet worked out. The interlocutors carry
on the necessary coherence work by jumping from scene to scene. This
"loosening" of coherence is balanced by speakers' stylistic format tying
(speaking in headline style or using other elliptical constructions).
5 Topic shift vs. keying shift

The fact that not only topics but also keyings must be ratified in order to be
maintained is seen in that both can be changed independently. In Data 1 the
speaker explicitly changes both domains.
5.1 Shifting topic and keying
In Data 1 the amusing episode ends at line 35. There is a short pause, and then
Lilo explicitly introduces a new subject.
36 L: Du, jetzt wenn i i will net die Konversation unterbrechen,

well now, though I I don't want to interrupt the conversation,
37 P: mocht nix.
doesn't matter.
38 L: aber ich bräucht ma den eh (- -) Rotterengel.
but I would need the Rotterengel.
In topic and keying Lilo returns to business talk. She asks for permission
to "interrupt the conversation". Peter agrees to that. The new topic is different
from the previous one, both in content and keying, which might make it
necessary to mark the shift explicitly. She wants to borrow "the Rotterengel", a
special dictionary she needs for interpreting. The group starts to talk about their
interpreting classes.
5.2 Shifting the keying
In Data 2 the topic is briefly maintained, but the keying is shifted into the
serious domain.
45 P: rote Rüben hab ich nie mehr gekriegt oder sonst was,
I have never gotten red beets or anything like that again,
46 auàer I woa jung.
except I was young.
47 I mein als Kind hob i roten Rübensalat gessen.
I mean as a child I ate red beet salad.
48 S: [(? ?)
49 ?: [(?also gut?)
okay well
50 E: ja, des hab ich aber, als Kind hab i des gemocht.
well, but I, as a child I liked that.
51 P: ja?
really?
52 E: jetzt kannst mich jagen damit.
now it scares me off.
53 P: i waaß a net. aba als Kind hob is gessen.
I don't know either. but as a child I ate it.
From line 45 on the group starts to seriously reflect on what they ate and
liked to eat as children, compared with the tastes they now have as adults. The
new topical aspect of taste differences between children and adults is
developed without creating comical perspectives and without integrating
laughter particles. Later the topic is dropped.
142 Helga Kotthoff
5.3 Shifting the topic, keeping the keying
In the next data we witness how a topic is changed, but the keying remains
stable. The fictionalisation here serves to tease the interactant Gisela, who is
humorously razzed for not having bought enough soft pretzels for the party
held after their judo practice. Teasing requires that the conversational joking be
directed at someone present (Straehle 1993; Günthner 1996; Boxer and Cortés-
Conde 1997). Any deficit or shortcoming can serve as a reason for teasing.
Very often, the teased person is spoken about in the third person (tangential
address), the others present being the direct addressees of the amusing attack.
Teasings can be short or elaborated. Fictionalisations can represent a form of
elaborated teasing. The teased person is the protagonist of the construction.
(3) (Conversation 16; episode 7)
Fritz (F), Gisela (G), Helmut (H), Klaus (K), Nadine (N), Oskar (O), Susanne (S),
Willi (W).
1 H: is sie in den Laden gegangen [und hat gesagt=

she went into the shop [and said
2 K: [HAHAHAHAHAHA
3 H: =[dreißig Brezeln, die Verkäuferin hat sie angeguckt,
=[thirty pretzels, the sales clerk looked at her,
4 hat gesagt=
said=
5 ?; [(? ?)
6 H: = ↑jawolL [(- -) nicht, und sie is losmarschiert mit zwölf
= yes. [ well, and then she marched out with twelve.
7 ?: HAHAHA
8 F: und zahlt für vierzig, gell
and paid for forty, didn't she.
9 H: ja. zahlt für vierzig,
yeah. paid for forty,
10 K: HEHEHEHE
11 H: undfür sechzig Butter gekauft hier.
and bought butter for sixty.
12 G: HA[HAHAHA
13 H: [wo wir doch sowieso alle gesagt haben,
[even though we had all said anyway
14 wir wolln nich soviel Butter.
we don't want so much butter.
15 a: HAHAHAHAHAHAHAHAHA
16 N: die is Dein Problem (? ?) HEBE
that's your problem (? ?) HEHE
17 G: HAHAHAHAHA
18 H: ha is doch wahr,
well that's true.
19 soviel Butter das is doch wirklich mich gesund.
so much butter is really not healthy.
20 G: [Du hasch ja Butter im Gsicht.
[you've got butter on your face.
21 N: [HEHEHEHEHEHEHE
21 H: is doch wirklich nicht gesund so viel Butter.
it's really not healthy so much butter.
22 A: die viele Butter.
so much butter.
23 N: die Butter im Gesicht.
the butter on your face.
24 K HAHAHAHAHAHAHA
25 H; komm Butter. (wipes his face)
come butter.
26 (--)
27 H: ach nee, das sind ja die elektronschen Dinger, nich,
well no, those are the electronic things, aren't they,
28 combutter.
29 a: HEHEHEHEHEHEHE
30 ?:
Computer
computers
31 a: HEHEHEHE
32 A: Du bisch doch echt der letzte Vogel combutter.
you're really one of a kind. combutter.
In Data 3 the fictionalisation, predominantly by Helmut, has the function

of teasing Gisela. Gisela was supposed to buy pretzels for the party and got too
few. Because this mistake had already made her the butt of many humorous
sallies that evening, the interactants can expect something humorous again
when Helmut starts to visualise the scene in the shop with Gisela as the main
figure. Klaus starts to laugh (2) as early as the start of the fictionalisation. Since
Helmut did not help shop for the party, it is immediately clear that he is
inventing a scenario. He first imitates Gisela with quoted speech (3), then the
shop assistant (6). The constructed dialogue immediately visualises the
imagined scene. The shop assistant's speech is constructed with an enthusiastic
intonation (high onset). Helmut continues with the detail that Gisela walked out
with only twelve pretzels (6). Here again he obviously departs from the realm
of reality. Someone laughs. Helmut fantasises Gisela as the shop assistant's
gullible victim. Fritz adds that Gisela paid for forty (8), thereby coherently
elaborating the line of fabrication started by Helmut in line 6. The fabrication is
144 Helga Kotthoff
jointly constructed by Helmut and Fritz. Fritz participates in expanding the

absurdity of Helmut's perspective on the invented scene.
Helmut agrees and Klaus laughs (10). Then another humorous topic of
the evening is reintroduced by Helmut. They not only have too few pretzels,
but there is not even enough butter for them. Fictionalising that Gisela was
overcharged and cheated, Helmut claims that she paid for enough butter to
spread on sixty pretzels; Gisela laughs (12). She obviously does not feel
insulted by Helmut's critique. His impertinent complaint about the small
quantity of pretzels and butter implied by his fabrication is just another starting
point for an amusing performance. Helmut's impoliteness is expressed with
exaggeration, imagery and funny details, thereby mitigating the face
threatening potential (Kotthoff 1996a). Helmut adds that everyone has said
they do not want so much butter (13-14). The whole group laughs (15). Here
Helmut is being openly ironic, since the contrary is the case. Everybody
laughs. Nadine laughingly tells Helmut that butter is his problem, alluding to
Helmut's weight problems. She counters his implicit 'bites'. Gisela laughs. In a
playfully convincing tone of voice, Helmut claims that so much butter is really
not healthy. Gisela discovers that he has accidentally smeared butter on his
face. Nadine laughs. The two seem to form a coalition against Helmut. Helmut
repeats his ironic claim that so much butter is not healthy. Anton repeats just
the ironic content so much butter. Nadine combines the attribution of a large
quantity of butter to the butter on Helmut's face, thereby changing the aim of
the tease. Klaus laughs. Helmut, wiping the butter from his face, says komm
Butter/come butter. A short pause follows. Then, as if correcting himself,
Helmut makes a pun using the phonetic similarity of komm Butter and
computer. The ironic butter and pretzel topic is closed, and a new topic is
coherently introduced by punning. The humorous keying, of course, thereby
remains stable. Now the topic of computer is jokingly exploited.
6 Closing remarks
Humorous fictionalisations can fulfil many functions. A group can implicitly
negotiate shared norms and evaluations as in Data 1. People can also involve
themselves in playful expansions of absurd scenes as in Data 2 -just for fun. In
addition, fictionalisations can be used to tease other people. Here as well, we
find people participating in the creation of a comical scene. But in humorous
keying we also find counter-attacks; the teaser always runs the risk of
becoming the teased. Counter-teasings re-establish the humorous keying.
It is obvious that topics and keyings have separate identities. Humorous
keying can be coherently produced beyond topic boundaries, and topical
coherence as well can be maintained beyond the boundaries of keying. Keying
coherence can be organised by specific means. In serious discourse wordplay

would not be accepted as a means of stepwise topic development.21 To keep the
humorous keying stable over several topics it is necessary to find some element
that also allows a double perspective or even a punchline and is still somehow
connected to the previous talk.
Interactional analysis aims at identifying strategies which help
interlocutors to construct and reconstruct contexts as humorous. Linguistic,
paralinguistic and nonlinguistic cues are constituents of activities which allow
contextual inferences to be directed not only toward humour, but also toward
specific humorous genres.
Joint humorous fictionalisations obey their own genre-specific rules. In
contrast to jokes, which offer a closed world of fiction, in this genre the
departure from reality occurs gradually. This is achieved by evoking frames
situated outside the normal fields of association of a topic (i.e., Bravo and
Steckbriefsammlung/trading card collection; feeding children under the
shower; buying 30 pretzels, taking 12, but paying for 40). The unusual frame
linkage is then worked out. In example 1, lines 16 and 17 help to stabilise the
unusual frame combination. This is then successively assigned a joint meaning,
whereby the absurdity of the combination is intensified (Playgirl) and can also
be shifted (Hermes Phettberg zum selber malen/to paint yourself). The
communication creates scenes which, however, are never fully formulated. The
headline settings evoke objects which are given in reality and whose relevance
is now conversationally created for an object. The simple naming of the
magazines Bravo and Playgirl, for example, challenges the speakers to realise
a possible relationship between them and the fiction's main character
(Phettberg). Elaborated details (with him you don't see genitals anyway or the
interaction in the shop) are functional in the process of visualisation. Humorous
fictionalisations establish unusual perspectives on concrete images and scenes.
Listeners do not just process the constructed scenes but try to co-construct the
keying or even enhance their comicality. Many psychological and linguistic
humour analyses have explained how listeners process standardised jokes. The
processing of conversational humour, however, demands a more complex
model, because 1. listening is not just listening, but listening for speaking, 2.
the humorous works with allusions to shared knowledge and presuppositions of
shared values, 3. it involves the social images of the speakers and 4. it is
indexed by specific turn shapes and contextualisation cues.
146 Helga Kotthoff
Notes
1. Perspectivation is used here similarly to Graumann (1989) and Kallmeyer and Keim
(1996: 286), who claim that all human beings' perceiving and acting is done from a
specific viewpoint. A process of perspective setting and taking is always somehow
involved in verbal interaction. A speaker must reveal her/his perspective, at least to a
certain extent; recipients must show how they interpret the manifested perspective.
2. Coherence and discourse topicality are not equated here; see Fritz (1982) and Schegloff
(1990) on the differences.
3. See Verschueren (1995) on metapragmatics.
4. See on humorous interaction modalities also the articles in Kotthoff (1996b).
5. I use Koestler's (1964) terminology of bisociation of contexts, frames or perspectives
because it is a more general concept for explaining humour than the concept of opposing
basic lexico-semantic scripts used by Raskin (1985). An extensive discussion of the
concepts can be found in Kotthoff (1998 in preparation).
6. The principle of "least effort" is also found in the otlier relevance-tlieories mentioned
above.
7. With Hymes (1974) and Bauman (1986) we see performance as a multilevel activity.
Performances exhibit temporal brackets and frames; they aim at producing an emotive
effect.
8. Different tenninologies are used in the field of punchline humour; see Attardo (1994) and
Kotthoff (1998) for a discussion.
9. Complexities in the management of point of view, polyvalence of utterances, and poetic
effects, which demand more coherence work, are often either implicitly (Wienold 1983) or
even explicitly (Lászlo 1988) associated with literature.
10. In Kotthoff (1996b) various authors discuss the patterns inherent in humorous genres such
as mocking, teasing, joint ironising and personal anecdotes. See Sornig (1987) on fooling
around (German: blödeln).
11. One aim of the study was to differentiate different genres of conversational joking.
12. Naturally, there are also joint fictionalisations which are not humorous. Sometimes
speakers jointly embroider a journey they are planning or some wish they have.
13. Phettberg is a pseudonym which refers self-ironically to the bearer's corpulence. In
German Phettberg means literally "mountain of fat." The spelling "Phett" instead of "Fett"
is old-fashioned and is intended to imply a high-level of education.
14. "Phettbergs nette Leit Show."
15. All interlocutors speak Austrian German. The transcription takes account of this. Laughter
particles are noted wherever they occur, and the vowels in which they appear (HIHI or
HUHU etc.) are indicated.
16. See Couper-Kuhlen (this volume) for framing properties of prosody.
17. Children thereby learn to deal with colours and colour nuances. The numbers on the not-
yet coloured pictures are assigned colours which children are to use in filling in different
sections of the picture.
18. "Leit" corresponds phonetically to English "light" in the sense of a "light show," but as a
dialect term it corresponds to the standard term "Leute"/"people."
19. Without background knowledge we could not figure out the teasing dimension of Sabine's
remark. The teased do not necessarily defend themselves against the teasing.
20. In his essay "Footing" (1981), Goffman elaborated the possible division of roles within
speakers and listeners. He distinguishes degrees of responsibility for speaking and
listening. The listener can be the addressee, the public or an eavesdropper. For Goffman,
"speaking" contrasts above all with "animating." This is, for example, the case with
quoted speech, where one puts one's own words into the mouths of other people and can
still manipulate the prosody so that in direct quotation a social type is simultaneously
created, as Couper-Kuhlen shows in her article in this volume. The animator presents
herself only as a sounding box for the words of others; Goffman further distinguishes the
principal, who bears responsibility, and the author, who is the creator of the text. See also
Günthner (1997) on reported speech.
21. Sacks (1972, lecture 5) showed that the best way to move from one topic to the next in
conversation is not by one topic closely followed by the next topic, but rather by what he
calls a stepwise move, which involves connecting what we've just been talking about to
what we're now talking about. He even writes "that what's thought to be a 'lousy
conversation' is marked by the occurrence of a large number of specific new topic starts
as compared to a conversation in which, so far as anybody knows, we've never had to
start a new topic, though we're far from whatever we began with, we haven't talked about
just a single topic; it just grew".
References
Attardo, S. 1994. Linguistic Theories of Humor. Berlin: de Gruyter.
Auer, P. and di Luzio, A. (eds). 1992. The Contextualization of Language.
Bateson, G. 1954/72. "A theory of play and phantasy". In Steps to an Ecology of
Mind. San Francisco: Chandler, 177-193.
Bauman, R. 1986. Story, Performance and Event. Cambridge: Cambridge University
Press.
Boxer, D. and Cortés-Conde, F. 1997. "From bonding to biting: conversational joking
and identity display". Journal of Pragmatics 27: 275-294.
Bublitz, W. 1989. 'Topical coherence in spoken discourse". Studia Anglica
Posnaniensia 22: 31-51.
Capone, A. 1995. "Dialogue analysis and inferential pragmatics". In F. Hundsnurscher
and E. Weigand (eds), Future Perspectives of Dialogue Analysis. Tübingen:
Niemeyer, 147-163.
Cook-Gumperz, J. and Gumperz, J. 1976. Context in Children's Speech. Papers on
Language and Context. {Working Papers No. 46). Berkeley: Language
Behaviour Research Laboratory.
Cook-Gumperz, J. and Gumperz, J. 1994. "The politics of conversation:
conversational inference in discussion". In A.D. Grimshaw (ed), What's Going
on Here? Complementary Studies of Professional Talk. Norwood: Ablex, 373-
395.
Couper-Kuhlen, E. This volume. "Coherent voicing: On prosody in conversational
reported speech".
Dascai, M. 1977. "Conversational relevance". Journal of Pragmatics 1: 309-328.
Dolitsky, M. 1983. "Humor and the unsaid". Journal of Pragmatics 7: 39-48.
148 Helga Kotthoff
Dorval, B. (ed). 1990. Conversational Organization and its Development, Norwood:

Ablex.
Fritz, G. 1982. Kohärenz. Tübingen: Narr.
Giora, R. 1985. "Notes towards a theory of text coherence". Poetics Today 6: 699-716.
Giora, R. 1997. "Discourse coherence and theory of relevance: stumbling blocks in
search of a unified theory". Journal of Pragmatics 27: 17-34.
Goffman, E. 1974. Frame Analysis. An Essay on the Organizations of Experience.
New York: Harper & Row.
Goffman, E. 1981. "Footing". In E. Goffman, Forms of Talk. Oxford: Blackwell, 124-
159.
Graesser, A.C., Person, N. and Johnston, G.S. 1996. "Three obstacles in empirical
research on aesthetic and literary comprehension". In R.J. Kreuz and M.S.
MacNealy (eds), Empirical Approaches to Literature and Aesthetics. Norwood:
Ablex, 3-23.
Graumann, C.F. 1989. "Perspective setting and taking in verbal interaction". In R.
Dietrich and C.F. Graumann (eds), Language Processing in Social Context.
Amsterdam: North Holland, 95-122.
Grice, P. 1975. "Logic and conversation". In P. Cole and J. Morgan (eds), Syntax and
Semantics, Vol 3. New York: Academic Press, 41-58.
Gumperz, J. 1982. Discourse Strategies. Cambridge: Cambridge University Press.
Gumperz, J. 1996. "The linguistic and cultural relativity of conversational inference".
In J. Gumperz and St. Levinson (eds), Rethinking Linguistic Relativity.
Günthner, S. 1996. "Zwischen Scherz und Schmerz. Frotzelaktivitäten in Alltags
interaktionen". In H. Kotthoff (ed), Scherzkommunikation. Opladen:
Westdeutscher Verlag, 81-109.
Günthner, S. 1997. "Stilisierungsverfahren in der Redewiedergabe". In B. Sandig and
M. Selting (eds), Rede- und Gesprächsstile. Berlin: de Gruyter, 94-123.
Hymes, D. 1974. "Ways of speaking". In R. Bauman and J. Sherzer (eds),
Explorations in the Ethnography of Speaking. Cambridge: Cambridge
Jakobson, R. 1960. "Closing statement: linguistics and poetics". In Th. Sebeok (ed),
Style in Language. Cambridge, MA: MIT Press, 350-377.
Jefferson, G. 1979. "A technique for inviting laughter and its subsequent acceptance/
declination". In G. Psathas (ed), Everyday Language: Studies in
Ethnomethodology. New York: Irvington, 79-96.
Jefferson, G. 1985. "An exercise in the transcription and analysis of laughter". In T.
van Dijk (ed), Handbook of Discourse Analysis, Vol. 3. London: Academic
Press, 25-34.
Jucker, A. 1993. "The structure and coherence of discourse". In H. Löffler (ed),
Dialoganalyse IV. Referate der 4. Arbeitstagung, Basel 1992. Tübingen:
Niemeyer, 71-78.
Jucker, A. 1995. "Discourse analysis and relevance". In F. Hundsnurscher and E.
Weigand (eds), Future Perspectives of Dialogue Analysis. Tübingen: Niemeyer,
121-147.
Kallmeyer, W. 1979. "(Expressif) Eh ben dis donc, hein' pas bien': zur Beschreibung
von Exaltation als Interaktionsmodalität". In R. Kloepfer (ed), Bildung und
Ausbildung in der Romania, Vol. 1. München: Fink, 549-568.
Kallmeyer, W. 1981. "Gestaltungsorientiertheit in Alltagserzählungen". In R.
Kloepfer and G. Janetzke-Dillner (ed), Erzählung und Erzählforschung im 20.
Jh. Stuttgart: Kohlhammer, 409-429.
Kallmeyer, W. and Keim, I. 1996. "Divergent perspectives and social style in conflict
talk". Folia Linguistica 30: 271-299.
Koestier, A. 1964. The Act of Creation. London: Hutchinson.
Kotthoff, H. 1995. "Oral performance in interactional sociolinguistics: humorous
stories among friends". Toegepaste Taalwetenschap in Artikelen 52: 23-55.
Kotthoff, H. 1996a. "Impoliteness and conversational joking: on relational politics".
Folia Linguistica 30: 299-327.
Kotthoff, H. 1998. Spaß Verstehen. Zur Pragmatik von konversationellem Humor.
Tübingen: Niemeyer.
Kotthoff, H. (ed). 1996b. Scherzkommunikation. Beiträge aus der empirischen
Gesprächsforschung. Opladen: Westdeutscher Verlag.
Laszló, J. 1988. "Readers' historical-social knowledge and their interpretation and
evaluation of a short story". Poetics 17: 461-481.
Mulkay, M. 1988. On Humour. Its Nature and Its Place in Modern Society.
Cambridge: Polity Press.
Müller, K. 1992. "Theatrical moments: On contextualizing funny and dramatic moods
in the course of telling a story in conversation". In P. Auer and A. di Luzio
(eds), The Contextualization ofLanguage. Amsterdam: Benjamins, 199-223.
Norrick, N.R. 1989. "Intertextuality in humour". Humor 2: 117-139.
Norrick, N.R. 1993. Conversational Joking. Humor in Everyday Talk. Bloomington:
Indiana University Press.
Raskin, V. 1985. Semantic Mechanisms ofHumour. Dordrecht: Reidel.
Ruch, W., Attardo, S. and Raskin, V. 1993. "Toward an empirical verification of the
general theory of verbal humor". Humor 6: 123-136.
Sacks, H 1974. "An analysis of the course of a joke's telling in conversation". In R.
Bauman and J. Sherzer (eds), Explorations in the Ethnography of Speaking.
Sacks, H. 1992. Lectures on Conversation. (Ed. by Gail Jefferson.) London:
BlackwelL
Schegloff, E. 1968. "Sequencing in conversational openings". American
Anthropologist 70: 1075-1095.
Schegloff, E. 1972. "Notes on a conversational practice: formulating place". In D.
Sudnow (ed), Studies in Social Interaction. New York: The Free Press, 75-119.
Schegloff, E. 1990. "On the organization of sequences as a source of 'coherence' in
Development. Norwood: Ablex, 51-78.
Schenkein, J.N. 1972. "Towards an analysis of natural conversation and the sense of
heheh". Semiotica 6: 344-377.
150 Helga Kotthoff
Schütte, W. 1991. Scherzkommunikation unter Orchestermusikern. Tübingen:

Niemeyer.
Schwitalla, J. 1997. Gesprochenes Deutsch. Berlin: Schmidt.
Sornig, K. 1987. "Blödeln". In E. Weigand and F. Hundsnurscher (eds),
Dialoganalyse II, Vol. 1. Tübingen: Niemeyer, 451-460.
Sperber, D. and Wilson, D. 1985. Relevance. London: Blackwell.
Straehle, C.A. 1993. "'Samuel?' 'Yes, dear?': Teasing and conversational rapport". In
D. Tannen (ed), Framing in Discourse. Oxford: Oxford University Press, 210-
229.
Tarmen, D. 1989. Talking Voices: Repetition, Dialogue, and Imagery in
Conversational Discourse. Cambridge: Cambridge University Press.
Verschueren, J. 1995. "Metapragmatics", In J. Verschueren, J.-O. Östman and J.
Blommaert (eds), Handbook of Pragmatics. Amsterdam: Benjamins, 367-371.
Wienold, G. 1983. "Linguistische Aspekte des Erzählens". In W. Sanders and K.
Wegenast (eds), Erzählen für Kinder - Erzählen von Gott. Stuttgart:
Kohlhammer, 79-128.
Wilss, W. 1989. Anspielungen. Zur Manifestation von Kreativität und Routine in der
Sprachverwendung. Tübingen: Niemeyer.
Part II
How to Negotiate Coherence: Degrees of Coherence
Disturbed Coherence: 'Fill me in'
Wolfram Bublitz and Uta Lenk
1 Introduction
In more than two decades of assiduous research into coherence, very little
attention has been paid to cases of coherence disorder or breakdown, i.e. of
what we propose to call disturbed coherence. Such neglect cannot readily be
explained, especially since disturbed coherence is by no means an infrequent and
unknown phenomenon in everyday communication. Quite the contrary: as
hearers we are used to appeal to the speaker with fixed phrases such as I'm lost
or fill me in or, more informally, hunh, whenever we are unable to reach a
coherent understanding of what we hear. And as readers we are arguably even
more often at a loss for how to make an essay, a poem, a newspaper
commentary, a technical manual coherent; and, to make matters worse in this
case, there are no writers vis-à-vis to whom we. could appeal to help us
negotiate the coherence of their texts.
In this paper, we would like to discuss disturbed coherence as a common
phenomenon of communication, which, on the one hand, must be kept distinct
from incoherent speech of persons whose command of language is impaired
(among them schizophrenics, cf. below), and, on the other hand, should not be
treated as simply the opposite of coherence. On the grounds that coherence is a
scalar notion, we argue that texts are frequently only partly coherent for their
hearers, or, to put it more strongly, that the hearer's understanding of coherence
can resemble but never totally match the speaker's. Thus, partial coherence is
not necessarily disturbed coherence. It turns into disturbed coherence at a point
when the extent to which the text is only partly understood is no longer
tolerated by the hearer. We distinguish between different types of disturbed
coherence (deliberate and accidental) and will present in this paper a detailed
account of some of its sources, such as topic drifts, topic changes, lack of
reference, frame and register breaks.
Our original intention was to keep within the confines of authentic and
attested spoken face-to-face discourse. However, relevant data is not easy to
154 Wolfram Bublitz and Uta Lenk
come by. To broaden our data-base, we therefore went hunting for examples in
all kinds of territories of discourse, ranging from available corpora like The
London-Lund-Corpus of Spoken English (LLC) to yet unavailable ones like The
Santa Barbara Corpus of Spoken American English (CSAE), as well as from
our own personal recordings of everyday talk to written fiction.
2 Coherence as a scalar notion

Both in the preface to this reader and elsewhere (Bublitz 1988, 1994; Lenk
1998) we have spelt out our hermeneutic understanding of coherence. The
following is a very brief summary, in which we also adopt the distinction
between text and discourse made by Seidlhofer and Widdowson (this volume).
Referring to ongoing spoken discourse, we see coherence as an interpretive
notion, which is intrinsically indeterminate because it is relative to participants
ascribing their understanding to what they hear. Though not given in the text,
i.e. not a text-inherent and invariant property at all, coherence nevertheless
'comes out' of the text, is based on the text as it is based on additional
information. The speaker normally helps create coherence by subtly guiding the
hearer to an understanding of coherence which comes close to or, ideally, even
matches her own.1 The hearer, on the other hand, uses the speaker's guiding
signals as instructions to re-align his interpretation with what he takes to be her
intention. Or, to put it differently, he assembles a (possibly tentative) view of
coherence which he assumes to come closest to that of the speaker. Coherence
is thus not a state but a process. It is also a cooperative achievement because it
depends on both the speaker's and the hearer's willingness to negotiate
coherence (in the same way as they negotiate ideational meaning and
illocutionary force).
To adequately describe cases of disturbed coherence, we need to
supplement this snapshot account with two more aspects. We contend that as a
rule participants operate on a generally shared default assumption of coherence
(cf. below), and we see coherence as a gradable, scalar notion. In literary
criticism, it is a pet topos that a text can be highly coherent for one reader but
much less so for another. Consider the opening lines of the poem in (1):
(1)
Observe this unit, gentlemen, which
(for want of a better word)
I call a self
It consists of an envelope enclosed by a void
and enclosing a void
Disturbed Coherence 155
and lined inside out with nothing but eyes

most of which are shut.
(Gray, Alasdair. 1989. Old Negatives - 4 Verse Sequences by Alasdair Gray.
London: Jonathan Cape, p. 18.)
The poem certainly allows for a number of different readings of

coherence. For some readers, figuring out the coherence of literary as well as
non-literary genres may put a far greater strain on their interpretive competence
than for others, and they may end up with items that will not fit into the overall
picture. On a scale of coherence, then, one and the same piece of written text
can be at a point further up the scale for one reader and further down for
another.
Essentially the same applies for ongoing spoken discourse. Of two
addressees, one may reach only a partial understanding while the other may
promptly come up with a (in his view) complete and adequate interpretation of
coherence. Thus, in the 'ears' of hearers, ongoing discourse can occupy any
position on a scale whose polar opposites are coherence and incoherence.2
It is tempting to equate disturbed coherence with partial coherence.
However, if we accept that it is not texts that have meaning, force and
coherence but rather speakers and hearers who ascribe meaning, force and
coherence to a text, we may safely argue that coherence is always only partial
coherence. Indeed it may be said that the 'hearer's coherence' at best comes very
close to the 'speaker's coherence'. Furthermore, we know that even the speaker's
interpretation of the coherence of her own discourse is not inevitably complete.
When asked to explain what they meant, speakers are in retrospect not always
able to say what they had intended to achieve and how their utterances were to
fit together and make sense. Often, speakers re-interpret their own utterances in
alignment with what they take to be their hearers' interpretation.
Any interpretation of the coherence of a text is restricted and only
"approximate" (Seidlhofer and Widdowson, this volume) and, accordingly,
partial in different degrees. And this is where disturbed coherence comes in. We
talk of disturbed coherence when the extent to which a text is only partly
understood is no longer tolerated by the hearer. This may be the case either
when he assumes that his interpretation of coherence clearly diverges from the
speaker's or when he fails to come up with a plausible interpretation on a global
level (if not on a local level). For the speaker (and for the analyst), coherence is
observably disturbed when the hearer demonstrates his inability to ascribe (a
sufficiently high degree of) coherence, either by commenting on it and asking
for clarification or by obviously opting out of a conversation (showing his
disinterest, etc.). Clearly, we must assume that disturbed coherence is best
described by adopting a hearer-knows-best stance.
3 Default principle of coherence

For the hearer to create coherence is a pervasive task, which involves making
complex acts of inference. He must attend continually to a wealth of incoming
linguistic and non-linguistic information in order to update his state of
knowledge. Instead of somehow extracting coherence from texts, he makes
hypotheses, which he matches against his knowledge as part of his own
contextual world. If we are agreed that ascribing coherence is such a complex
and tricky business, then we must wonder why cases of disturbed coherence are
not abundant. Another puzzling question which arises is: why do hearers rarely
struggle to find coherence even though the data to go by often looks sadly
insufficient, at least to an outsider? As an example of this, consider the following
extract from a family conversation:
(2)
ALMA: ...() look at this,
... isn't [this pretty]?
SUSAN: [these are] good.
... [2<X lime or the green X>2].
DICK: [2what is that2].
ALMA: .. [3Kerry always wears them3].
ROGER: [3bok bok3] bok bok bok.
ALMA: ... She brought some,
ROGER: bok [bok].
ALMA: [I al]ways comment [2on them2].
SUSAN: [2well you2] have to be
nice to me,
[3if you want to talk3][4me in4][5to shar5][6ing
my candyó].
JESS: [5<VOX %uh
VOX>5],
ALMA: [3She brought some3] «
DICK: [3aw=3],
[4that's4][5cute5] [6Kerry6].
JESS: ... [7<VOX %uh VOX>7],
ALMA: [7 X <X the table for them X>7].
BOB: [7it's not7] cute at all,
X: [8 XX <X get yourself X>8].
BOB: [8it's very elegant8].
JESS: I am,
.. I am not mean to you all the time.
BOB: ... Look how mean you're being.

(CSAE: Guilt)3
The participants themselves appear to be undisturbed by this hotchpotch

of truncated contributions, which may or may not be relevant for whatever
topics they are talking about. They seem to find coherence in this conversation.
And analysts will follow them and assume that this text is coherent despite the
fact that the analysts themselves may be at a total loss to see how its parts could
fit together well. The reason for their assumption, and thus the answer to our
two questions, is that in our societies we operate on a general default principle
of coherence.
As a matter of course, we proceed from the assumption that what we hear
and read is coherent, even if this involves making extremely remote and unlikely
connections. Even though there are deviations, we permanently assume that our
fellow-speakers are creating coherent discourse. The speaker is likewise aware
of the principle and expects her hearer to look for coherence in what she says.
Both the hearer and the speaker share and accept the default principle of
coherence because it follows from the general principle of cooperation (in the
Gricean sense) and is therefore an essential normative basis of communication,
which leads to rational behaviour as the only path to efficient communication.4
We can now describe these notions more clearly: for the hearer, the
coherence of a text is disturbed when he is unable to make it coherent but,
operating on the default principle of coherence, assumes that it could be made
coherent because he has no reason to believe that the speaker is opting out, i.e.
not abiding by the general principle of cooperation.
4 Some examples
It is precisely because he continues to operate on the default assumption of
coherence that the hearer appeals to the speaker for elaboration and
clarification, i.e. to help him overcome the disturbance and find coherence. To
this end, he resorts to fixed phrases such as you 've lost me, what do you mean, I
do not understand, fill me in, how does that tie in with... Here is one example
(for more, cf. below):
(3)
S: Who never stopped talking or something - urn - but we
really need to talk about this.
D: About what?
K: Dan -I mean - it's not like Dan - he just doesn't

talk.
S: Mark Twain
D: Oh about Mark Twain
S: Yeah what'd ya think I was talking about?
M: It's your topic.
D: You just told me to go take a walk.
K: (cat screechy sound) It's like about WHAT!
D: I'm lost I swear
S: No I don't know what they're talking about.
(Joan Wallace Mixed Sex Discourse, unpublished data, 1994)
To prevent intolerable degrees of coherence disturbance, speakers in

everyday face-to-face discourse, who permanently make assumptions about
their hearers' state of knowledge, essentially resort to two strategies: guiding
and monitoring.
Speakers guide hearers by continuously weaving a tight net of intra-
textual relations, using various kinds of cohesive clues, as for example discourse
markers (cf. Lenk 1998). A shortage of guiding clues can hamper the hearer's
efforts to infer coherence, though such a shortage may be balanced or
circumvented by information taken, for example, from prior knowledge. But
usually such clues are liberally applied to secure comprehension (of referents
referred to, of scenarios opened, of localities, dates etc) and to ease the hearer's
burden of interpretation (cf. Bublitz 1996a). A fairly obvious means of 'putting
the hearer in the picture' through guiding his interpretation are digressions:5 the
speaker digresses from the current topic and offers additional information in
order to prevent a disturbance of coherence. In the conversation from which the
following example is taken, the main topic of discussion, or supertopic (Chafe
1994) has been environmental issues when Marilyn introduces the subtopic of
this great hook. She briefly digresses from the subtopic to give additional
information about the author of the book and signals the end of the digression
with the discourse marker anyway and then returns to the topic of the book:
(4)
MARILYN: I-1 really,
... we read this great book.
.. called,
.. The End of Nature?
PETE: .. Oo=
@ [Sounds wonderful].
MARILYN: [by this guy],

.. this .. punk,
.. you know,
ROY: Bill [McGivon].
MARILYN: [twenty-eight year ol]d,
PETE: ..<X R[2ightX>2],
MARILYN: [2Harvard2] guy,
who lives out in the woods?
..al[3a Thor3]eau?
PETE: [3Uhuh3].
MARILYN: (H) Anyway,
he's smart.
...()and,
uh,
...() it's called,
yeah,
it's called The End of Nature,
... and it's just really this <@ scary book
(CS AE: Conceptual Pesticides)
In order to monitor, i.e. to check on her hearer's interpretive process, a

speaker uses meta-discoursal phrases such as are you still with me, see what I
mean, know what I'm talking about, get me. Here is an example (in which,
however, the monitoring appears not to have helped a lot):
(5)
b [...] I've been thinking [...] over writing this
dictionary [...] and looking [...] at the work of
slightly older people [...] - all their examples and ■
half of them are military -[...] they do escort in
terms of warships I do it in terms of red roses - -
A o\h - m\
b again and again and again [...]
A m\
b (do you) see what I mean
A \m (coughs .)
b it's the collocaters that have changed
A ye\s
(LLC 1.10.226-235)
Incidentally, we also conceive of an interesting phenomenon which we

could call over-coherence, i.e. extracts whose coherence is secured by an
unnecessarily composite cluster of cohesive means, paraphrases and the like. In
the following example, a guides A through the entire stretch of what he is

saying, trying to assure comprehension. And even though A asserts early in the
exchange that he has understood and keeps repeating this assertion, a goes on
to explain providing a lot of additional, by then superfluous data:
(6)
a [...] the meaning ofthat little diagram is that
everybody's got to do the central three he's got to
pick another seven on top of that and he's got to pick
up his seven by following the paths -
A *how do you m\ean he (he)* +m+
a *and if he wants* if he wants to take say this one
+he's+ got to do everything in bet**ween**
A **oh** I\ see . ye\s .
a so that if he wants to do this this commits him already
to seven three four five *six seven*
A *I f\ollow* y\es . y\es
a so that he can't really do that and that - -
A no I f\ollow - . I f/ollow [...]
(LLC 2.2a.376-388)
Normally, participants accept mutual responsibility for collaboratively

achieving coherence and seek to keep their "collaborative effort" (cf. e.g. Clark
and Wilkes-Gibbs 1992: 486) at a minimum. From this point of view, both the
hearer and the speaker are to blame for any occurring coherence disturbances in
ongoing discourse.
5 Disturbed coherence versus coherence impairment

Before analysing some such instances of disturbed coherence, we need to make
a clear distinction between the latter and a related concept that we find in
literature on a particular kind of incoherence. Their authors tend to take their
examples from two sources: either from the speech of young children,
schizophrenics and aphasics, whose command of language is said to be
somehow deficient (with children not yet and schizophrenics no longer
possessing a sufficiently high degree of communicative competence).6 Or else
they refer to passages taken from literature, preferably poetry and drama "from
purposefully non-standard (or 'experimental') authors such as James Joyce,
Gertrude Stein and Eugen Ionesco" (Roberts and Kreuz 1993: 455). Here is one
example of what we propose to call coherence impairment, taken from
schizophrenic discourse:
(7)
Swiss pride must be deserved. Salu K..., I am the nun. If that's enough, you are still
his. That is a brave cavalier, take him as your husband. Karoline, you well know,
though you are my Lord, you were just a dream. If you are the dove-cote, Mrs. K. is
still beset by fear. Otherwise I am not so exact in eating. Handle the gravy carefully.
Where is the paint-brush? Where are you, Herman?
(Roberts and Kreuz 1993: 456f.)
Listening to this (or, indeed, reading it), you would most certainly regard
(7) as a piece of totally incoherent discourse. However, if we had listed (7) as
an example of disturbed coherence - not telling you that it was produced by a
schizophrenic - or appealed to your default assumption of coherence, your
reaction would most likely have been very different. You would have suspended
your judgement until it had dawned on you that it was not for a lack of data that
you failed to infer the missing links (and thus coherence), but that there were no
links missing because the speaker either had not intended to or else had been
unable to produce coherent discourse in the first place. The hearer who has
come to such a conclusion will no longer operate on the default principle of
coherence. He will not give the speaker, or rather the text, the benefit of making
sense and will refuse to further engage in cooperative interaction. So the
question to ask, then, is not whether or not a text like (7) is coherent but
whether or not its hearer or reader is willing to assume that it is coherent, i.e.
whether or not he is willing to make it coherent. This is entirely his choice (as
well as in accordance with the hearer-knows-best principle of describing
coherence), and is independent of the cause of the speaker's failure to secure
coherence; she may, for example, be schizophrenic or aphasic, too young or too
old, drunk or drugged.
The explanation above accounts for those cases of incoherence that are
caused by some kind of impairment. But it also helps us to understand why a
reader, after having failed to understand its coherence, may end up concluding
that some literary text is incoherent (indeed, as incoherent as schizophrenic
talk), because its writer did not abide by the default principle of coherence.
There are several reasons for the reader's readiness to come to some such
conclusion (unfamiliarity with the genre or the author, insufficient interpretative
competence among others). Other readers, for instance those who have greater
interpretive skills or who have had a chance of conversing with the writer and
thus of negotiating meaning, force and coherence (i.e. of establishing a common
context) might find the same text to be either fully coherent or might regard it
merely as a case of (more or less) disturbed coherence.
6 Types of disturbed coherence

Let us return to cooperative interactions without coherence impairment. At this
stage, we want to introduce a distinction between two types of disturbed
coherence: deliberate and accidental. Consider (8):
(8)
A ich glaube ich bin gegen Sesam allergisch, ich hab
nämlich neulich - oder gegen Mohn. ich hab -
B grüne oder weiße?
A hä?
B grüne oder weiße?
A grüne oder weiße?
B grüne oder weiße Bohnen?
A Mo = =hn!
(Uta Lenk, personal data, March 1997)
'A I think I am allergic to sesame seed. the other day I - or to poppy seed. I had -
B green or white?
A hunh?
B green or white?
A green or white?
B green or white beans?
A po==ppy seeds!'
B's interrupting question grüne oder weiße? is totally incoherent for A,

who rightly understands that grüne and weiße in the elliptical noun-less
utterance are qualifying adjectives but cannot figure out what the qualified noun
is to which B is referring. This is a case of what we call accidentally disturbed
coherence. Though intent on creating coherence for ,A, B fails to do so because
she mistakenly assumes that A is aware of what the implied referent is, or is at
least able to figure it out.
Examples of our second type, deliberately disturbed coherence, are rather
difficult to come by although we believe that they are not that infrequent in
everyday discourse. Example (9) is taken from fiction, a short story in which the
character Frank Fencepost is testifying in a trial.
(9)
When it is Frank's turn to testify, the bailiff try to swear him in. 'Do you swear to
tell the truth, the whole truth, and nothing but the truth?' 'I'll take the first one,' say
Frank. [...] 7 do will suffice,' say the Bailiff. 'Why would you give me a choice if I
got to choose all three?' ask Frank. The appropriate reply is I doj the bailiff say. 'If
I say I do, I'm liable to wind up married to somebody. How about Okay?
'Whatever.' 'Let me hear those choices again?'
(W. P. Kinsella. 1994. "Bull." In: Brother Fran's Gospel Hour. Toronto:
HarperCollins, p. 9.)
The non-complying Frank Fencepost is obviously set on disturbing the

smooth flow of discourse expected by his audience, and he succeeds. How
exactly does he proceed? He is perfectly aware that all those attending the trial
share the stereotypical, socio-pragmatic knowledge of the situation, i.e. of the
court room frame. The audience's knowledge of the routine intrinsic to this
particular frame is shaped by its experience of prior instances of the same frame,
which will then give rise to a number of respective expectations. In order to
achieve his aim, Frank Fencepost deliberately violates one of these expectations
- the linguistic expectation - by deviating from the expected or one might say
preferred sequence. He keeps within the structural frame of each adjacency pair
opened by the bailiff (e.g. Ill take the first one filling the slot reply opened by
the preceding question) but fails to deliver the expected content. To adopt a
(Neo-)Firthean view: Frank's hearers realise that here the assumption they
normally operate on, namely that the meaning and the force of an utterance
depend on its place in ongoing discourse or in a text, does not hold. Coherence
is disturbed because while the speech act pattern fits the place in the structural
sequence, the proposition does not.
Unlike accidentally disturbed coherence, deliberately disturbed coherence
is the outcome of a meaningful choice. The speaker is intent on preventing the
hearer from ascribing coherence, usually by violating an expectation. Why
would a speaker do that? We can think of several motives. Some of them are
clearly unfriendly with the speaker wishing to unveil the hearer's ignorance or to
make him look like a fool etc. Others are clearly friendly with the speaker
seeking to tease the hearer, to surprise him pleasantly, to create suspense, to
achieve a humorous effect etc. There is an example in (11) below, which we will
discuss presently, and another one in (10), in which C's interruption later turned
out to be meant as a bit of friendly ironic teasing:
(10)
A While I was taking Chinese, during my year as teaching
assistant, I was planning to go to China later on, to
really pick up the language there. But that was the
year of the Tien An Men massacre, and that kind of
changed my plans. I -
C So you must have been pretty happy to find out

that it never happened.
A What do you mean it never happened?
B (...)
C The Tien An Men massacre never happened. At least
that's the government's version.
(Uta Lenk, personal data, March 1997)
Speaker A's reference to the Tien An Men massacre opens the frame of a
totalitarian regime where she would not want to spend any amount of time to
learn the language. C's statement that the massacre never happened is
completely incoherent for A because it negates the existence of one particular
component of this historical frame, an existence which she presupposes to be
true as it is well-known and widely documented. This is totally unexpected and
so unbelievable that it triggers a question for clarification instead of a clear
contradiction.
And there is a third group of motives that are neither friendly nor
unfriendly. The speaker may, for instance, deliberately disturb coherence simply
in order to get the hearer's attention.7 Or she may be deliberately vague,
ambiguous, non-specific, thus preventing the hearer from an easy, obvious and
immediate ascription of coherence, because she wants to wait for the hearer's
reaction, for his interpretation of coherence. She leaves the question of
coherence open, as it were, leaves it to the hearer to answer it. Such behaviour
supports our view that participants accept mutual responsibility for making a
text coherent (in her workshop contribution, Kotthoff gave examples where
deliberate misunderstandings result in temporary disturbances of coherence and
which are employed as means of conversational humour).
To conclude, the speaker's motives when deliberately instrumentalising
disturbed coherence are opportunistic but she never aims at a total and
permanent breakdown of coherence. Incidentally, for the hearer who fails to
ascribe coherence it makes no difference whether or not the disturbance
happens accidentally or is deliberately caused by the speaker.
7 Sources of disturbed coherence

7.1 Disturbed coherence is not disturbed cohesion
The topic of our paper is not 'disturbed cohesion' or, indeed, 'incohesiori.
Taking a text and stripping it of its cohesive means could be described as
making it 'incohesive'. As there is no direct and mandatory correlation between
lack of cohesion and lack of coherence, we are not interested in 'incohesive

texts' per se. Likewise, it is by now common knowledge that the equation more
cohesion = more coherence is far too simple. Research into the correlation
between cohesion and coherence has come up with a number of well-known
counter examples (cf. Bublitz 1994).
However, having said that, let us add that research has also shown that at
least in some genres, among them everyday face-to-face conversations, the
wealth and abundance of cohesive means is quite extraordinary to the point that
to strip such a text of its cohesive devices would result in an incomprehensible
verbal skeleton (cf. Bublitz 1996a). Cohesive means are meant to ease the
hearer's job of trying to fit new information into his stack of old information.
Doubtless, a lack of cohesive means could be conducive to disturb the hearer's
interpretation of coherence. An unimpeded flow of information easily
interpreted as coherent depends not only on cognitive links but arguably even
more so on communicative and (as with gambits or discourse markers) meta-
communicative textual links. (And this holds true not only for everyday face-to-
face conversation but even more so for some written genres where the writer
has to do without immediate feedback.)
In this paper, however, we will not address disturbed coherence as a result
of the absence of cohesive means such as gambits, discourse markers, preferred
seconds or those described in Halliday and Hasan (1976). There is one
exception, though, which appears to be uncommonly frequent: reference, or
rather unclear reference. We will therefore turn to it, albeit very briefly, before
discussing other sources of disturbed coherence.
7.2 Unclear reference
For the hearer, the act of identifying a referent can fail or can be partly or fully
successful. It can be immediate or delayed and can take a shorter or longer span
of time depending on how much information is readily available and how much
cognitive effort is ultimately required. These options correspond to different
degrees of disturbed coherence. In (11), the coherence of B's first utterance is
strongly disturbed for d, who completely fails to identify the referent of feet:
(11)
A the thing to do is obviously to swing the PhD subject
round to something -. nearer what I'm being paid to do
*you see*
B *so you've* gone off feet -.
d you've what
A,B (- - - laugh)
B he's gone off *feet*
A Tm* no longer Peter's footman (- - +-+ laughs)
B +(-laughs)+ you've heard of the fo/ot [...]
d no
B Abercrombie's feet -. *rhythm . rhythm*
d *well yes but « n o t 1 syll»* fill me in yeah
B well he'd been doing a thesis on feet
d oh gosh yes of course
(LLC 2.4.1065-1083)
Whenever the hearer is faced with several alternative options, he can

postpone the ascription of coherence and wait for subsequent relevant
information to fill in the gaps in his contextual knowledge. Or, if he decides that
to know the referent is of no importance and will not impede his search for
coherence after all, he can simply accept the disturbance for the moment. Or he
can, in fact, ask for clarification, i.e. initiate the act of mutually negotiating
reference and thus coherence. This is the course of action chosen by d in (11).
He needs, however, two quests for clarification because his fellow-speakers
refuse 'to fill him in' after his first quest. Clearly, (11) is a case of deliberately
disturbed coherence. A and B successfully prevent d from finding coherence in
what they are talking about because they delight in a bit of friendly teasing and
eventually, after having 'filled him in', savour his dawning of understanding. To
this end, both speakers temporarily suspend a communicative principle
according to which a speaker should only use a referring expression when she
believes that her addressee can identify its referent (without resorting to
additional data). Here, the achievement of an interactional, interpersonal goal
overrides the observance of Grice's general principle of cooperation.8
7.3 Topic drift
Topic drift occurs when the global topic is temporarily neglected or even
permanently abandoned. Discourse then develops according to what is
associatively closest or "easiest to say next" (Hobbs and Agar 1986: 231) rather
than to what the speaker's projected goals of the global topic demand. This
complies with the hearer's view that for the interpretation of the current
utterance "the immediately preceding utterance has a special significance
because it is against the mutual cognitive environment as established by this
previous utterance that the oncoming utterance is going to be interpreted"
(Jucker 1993: 72). Each single step of slightly shifting the topic will not affect
local coherence, the two adjacent utterances are regularly understood as

coherent by the hearer.9 Nor will it affect global coherence at that point,
whereas a sequence of such steps will inevitably lead to disturbed global
coherence, in the end; consider (12):
(12)
a you were telling us a . a long complicated story about
Eileen's sons last night -I hadn't quite got them in
order
B well she has four boAys
[...7 lines leading up to their names]
B BeVn Don Luke and Co\in
[... 88 lines, in which B starts to talk about Eileen having moved into a rather
outback kind of place and then, in long monologues, drifts from topic to topic;
when finally she comments on the immense costs you have to invest in dental
chairs, a steers her back to the original topic]
a and how does this tie in with Ben
(LLC 1.13. 1ff)
Another example shows how an assumption of global coherence can

constrain a participant from checking on local coherence:
(13)
Student: wär's möglich daß sie [seine Hausarbeit] eher
synchron nicht diachron ist
Professor: ja kein Problem
Student: mit synchron hob ich's nicht so
Professor: ja ja *aber Sie* könnten vielleicht
Student: *äh diachron*
(Wolfram Bublitz, personal data, April 1997)
'Student: would it be possible that it [his paper] is

synchronic rather than diachronic
Professor: yes no problem
Student: I am not so keen on synchronic
Professor: yes yes *but you* could perhaps
Student: *uh diachronic*'
The professor's ongoing interpretation of coherence is not disturbed

because he fails to notice that on a local level the student's adjacent utterances
do not fit. His expectations, which are geared both to a more general
conversational goal (giving advice to a student) and a more immediate one
(specifying requirements for a synchronic paper), clearly override his

registration of the student's incorrect phrasing.
7.4 Topic change
Another source of disturbed coherence which is related to topic shift is topic

change (cf. for a more detailed account Geluykens, this volume). It comes in
two easily recognisable types: abrupt topic change and announced topic
change. Topic changes serve admirably to show that coherence is a matter of
perspective, particularly for the analyst, less so for the participant. Focussing
merely on the utterances immediately preceding and following a change of
topic, analysts can be hard pressed to make them coherent, as in (14) and (15):
(14)
A well I haven't seen her this te\rm
B I \have I've only seen her o\nce since I he\ard this
news I think in in the in the refe\ctory and I didn't
actually spe\ak to her — I had a seminar today in
which . people hadn't read the stu\Aff because [...]
(LLC 1.4.1076ff)
(15)
c he's fast asleep -
B i/she
c yes -.
B oh go\od . Jean Piage\t what a what's the point of
having a book about hi\m around
(LLC 2.10.326fi)
For the participants actually present, however, these abrupt topic changes,
though definitely locally incoherent, do not lead to disturbed coherence for
various reasons. The new topic introduced after the topic change in (14) is
coherent on a global level because, like the preceding topic, it contributes to the
global topic 'exchanging university news'. Therefore, its introduction is not a
violation of the participants' expectations triggered by the global topic. The
local incoherence in (15) is likewise of no concern for the global coherence,
though for a different reason. There is not one superordinate global topic as in
(14) but both the preceding and the following topic in (15) relate to the general
conversational purpose or goal of maintaining and developing social contact.
Here, the exchange of information on the cognitive or ideational level is
secondary to the exchange of information on the interpersonal level. Analysts,
when looking only at truncated sections of some such conversations, may not be
aware of this, but those participating in the interaction check all incoming
messages, including those after an abrupt change of topic, against the main
conversational goal of making them coherent.
Abrupt topic changes can, however, prompt disturbed coherence.
Example (16) illustrates a case where both the next topic and the topic change
itself lead to disturbed coherence.
(16)
PETE: Yeah.
.. That would be good.
... Cause [all] of that stuff should go into the
compost pile to begin with.
MARILYN: [X]
..Yeah.
..(H) Actually,
you know,
.. Zeke the sheik.. is a local.
...You know,
the guy whose compost pile blew up?
PETE: ... Oh no I don't know a[bout this].
MARILYN: [Didn't you hear] about him?
PETE: No.
(CSAE: Conceptual Pesticides)
Marilyn takes up Pete's cue of compost pile to initiate a topic drift from
gray water systems, a subtopic of their overall topic environmental issues, to a
specific case of compost-pile-trouble as another subtopic. The abrupt change of
topic is introduced by dropping a name and commenting on its owner {Zeke the
sheik is a local). Pete's lack of reaction ('...' indicates a pause) is recognised as a
sign of his failure to make this utterance and thus the topic change coherent.
Obviously, he is unable to detect a superordinate global topic (though, in fact,
there is one) to which he could relate the utterance at the moment of hearing it.
Nor is he able to relate the new information to a general conversational goal. At
this point in the conversation, the utterance therefore comes totally unexpected
and disturbs the coherence of the ongoing discourse. This is recognised by
Marilyn, who then gives additional details in order to help Pete identify the
referent, thus enabling him to see the connection between the new and the old
topic and ultimately find coherence.
To conclude, abrupt topic changes are always locally incoherent but
coherent on a global level provided the new topic is a noticeable contribution
either to a superordinate topic or to a general conversational goal on the
interpersonal level of communication. If both conditions are not fulfilled, the

abrupt topic change will also be seen as globally incoherent, i.e. as something
which disturbs coherence.
But there is an option for the speaker who wants to change the topic
without disturbing coherence: she can announce her topical action, in which
case the hearer's interpretation of coherence is not disturbed but 're-directed'.
(17) is an example of such an announced topic change. The fragment
reproduced here is preceded by a third party's talk, in which, much to her
discomfort (and to the amusement of her interlocutors), she gets more and more
entangled in a topic which turns out to be embarrassing to her hosts (i.e. the
subject mentioned by c). Finally, and this is the point where the example sets in,
c comes to her aid with the suggestion of changing the topic, which is
immediately taken up and carried out by d:
(17)
c the only diplomatic way out is to change the subject
very interestingly -
d we're putting up a lot of shelves in the other room
(LLC 2.10.1019f)
Speaker c topicalises the possible action that would help in a situation as

his partner is in, which is then carried out by speaker d. Speaker c's
announcement prevents speaker d from being sanctioned for breaking the line of
topical talk and interrupting her hearers' interpretation of coherence.
7.5 Frame break
Frame breaks, as in (18), are a common source of disturbed coherence.
(18)
A: im German Department in Pittsburgh hat jemand eine Dissertation über
Aschenputtel geschrieben *und*
B: */Aschenputtel*
C: */worüber*
(Wolfram Bublitz, personal data, January 1997)
'A: in the German department in Pittsburgh somebody has written a dissertation on

Cinderella *and*
B: *Cinderella*
C: *on what*'
Frames are normally activated by keywords. In (18) the keywords are

German Department and Dissertation, which open up the frame 'PhD
dissertation at a university'. Operating on the default principle of coherence, B
and C assume that A has not left the boundaries of the conventions associated
with the recently opened frame. There is no reason to believe that they and A do
not share identical conceptions of this frame. For the two hearers, the coherence
gets disturbed when A mentions the dissertation topic Aschenputtel because it
does not fit an essential element of this particular frame, viz. that a university
dissertation is about a serious scientific topic. At this point, the question Why
that now?, which according to Schegloff (1990: 55) permanently accompanies a
hearer's process of understanding, cannot be answered and disturbs, i.e.
interrupts B and C's search for coherence. Their simultaneous interruption of
A's talk may be explained in different ways. They may think that they have
simply misunderstood, i.e. misheard. Or they may infer that in Pittsburgh or
even in universities throughout the US there exists a conceptualisation of the
dissertation frame that is substantially different from their own, in which case
their exclamatory reactions may carry tones of disbelief or even of
condescending disapproval.
7.6 Register break
When we gave the oral presentation on which this paper is based, we started
our talk with You know, the funniest thing happened to me this morning on my
way to our workshop. In more than two decades of assiduous research into
coherence ... We deliberately chose this onset to create a feeling of disturbed
coherence in our audience. Apparently, we were partly successful, and this was
possible because You know, the funniest thing happened ... is a typical story
"preface" (Sacks 1974: 340 ff) but not an appropriate preface of an academic
talk. It is in the wrong register and opens the wrong genre. Generalising the
default principle of coherence, we can say that as long as there is no reason to
assume otherwise participants in a conversation assume that the current
situation, frame, register and genre still hold (cf. Widdowson 1983: 45). A
noticeable deviation from their conventions disturbs the hearer's search for
coherence. Unfortunately, we cannot go into this aspect in detail here but will
leave it for further research.
8 Conclusion
Contrary to the general impression created by much research into coherence
disorder, it is not only schizophrenic talk that is incoherent. In our research, we
found disturbed coherence of a different kind in everyday face-to-face
conversations. This did not come as a surprise because, after all, ascribing
coherence is a highly idiosyncratic and thus easily disturbed activity. Cases of
disturbed coherence are, nevertheless, not abundant in conversations for two
reasons: speakers and hearers alike operate on a general default principle of
coherence and they accept mutual responsibility for collaboratively achieving
coherence. For the hearer, the coherence of a text is disturbed when he is unable
to make it coherent but assumes that it could be made coherent because he has
no reason to believe otherwise. Starting from the assumption that coherence is a
scalar notion, we argue that coherence is disturbed when the extent to which a
text is only partly coherent is no longer tolerated by the hearer. The hearer's
failure to make a text coherent is, inter alia, prompted by violations of
expectations as they appear in cases of topic drift, topic change, unclear
reference, frame- and register-breaks.
Notes
1. In this paper, we use she to refer to the speaker and he to the hearer.
2. For a similar view cf. e.g. Dahlgren as quoted in Roberts & Kreuz (1993: 455).
3. The following transcription conventions appear in the CSAE examples: a carriage return
indicates the end of an intonation unit, a comma stands for a continuing intonation
contour, a period indicates a final intonation contour, a ? stands for appealing intonation
contour; two dashes (—) indicate a truncated or incomplete intonation contour after (self-)
interruption; [ ] indicate speech overlap (numbered consecutively when several overlaps
occur close to each other); .. mark a short pause,... a medium pause and ...() a pause that
will be timed in the published version of the corpus; = marks the lengthening of the prior
sound; (H) is inhalation, (Hx) exhalation, @ stands for a syllable of laughter; <X word
X> indicates uncertain hearing, X an indecipherable syllable.
4. Ciliberti (this volume) shows that in certain situations, such as e.g. examinations, the
default principle is temporarily suspended. Typically, these are kinds of interactions
where all participants know that the relevant rules for this particular situation require the
suspension.
5. Unless the digression constitutes out-of-topic talk or is a so-called situational digression,
cf. Bublitz (1988: 94ff) and Lenk (1997).
6. For "in fact, one criterion for a diagnosis of schizophrenia is incoherent speech
(American Psychiatric Association 1987)" (Roberts and Kreuz 1993: 456). However,
even schizophrenic talk is not as uniformly incoherent as is often argued but may be
more coherent or less coherent. This is pointed out by Roberts and Kreuz, who have also
observed that "schizophrenics may be considered more coherent when they are assumed
to be speaking coherently" (456).
7. Occasionally, speakers even tell the hearer in so many words that what they are saying
does not fit. Or the incoherence may be the (intended) result of a clash of maxims: the
speaker may deliberately observe the politeness maxim at the expense of the maxims of
relevance or quality.
8. It is very interesting that, if several speakers are involved in such a process of teasing,
often a kind of agreement between the speakers as to the aim and method can be noticed
despite the fact that their intention was never verbalised.
9. "Local coherence" here means that two successive utterances are linked by coherence
relations such as "temporal succession, causal relation, explanation, semantic
parallelism, elaboration, exemplification, contrast, background [...]" (Hobbs and Agar
1986: 220f); cf. also Lenk 1998 and Sanders and Spooren (this volume).
References
Bublitz, W. 1994. "In the eye of the beholder: 'the rather mystical notion of
coherence'". In K. Carlon, K. Davidse and B. Rudzka-Ostyn (eds), Perspectives
on English. Studies in Honour of Professor Emma Vorlat. Leuven, Paris:
Peeters, 213-230.
Bublitz, W. 1996a. "T bought a freezer - you've bought a freezer - they've bought a
freezer': repetition as a text-building device". In C. Bazzanella (ed), Repetition in
Dialogue. Tübingen: Niemeyer, 16-28.
Bublitz, W. 1996b. "Semantic prosody and cohesive company: somewhat predictable".
Chafe, W. 1994. Discourse, Consciousness, and Time. The Flow and Displacement of
Conscious Experience in Speaking and Writing. Chicago: Chicago University
Press.
Clark, H.H. and Wilkes-Gibbs, D. 1992. "Referring as a collaborative process". In P.
Cohen, J. Morgan and M. Pollack (eds), Intentions in Communication.
Cambridge, MA: MIT Press, 463-493.
Hobbs, J. and Agar, M. 1986. 'The coherence of incoherent discourse". Journal of
Language and Social Psychology 4: 213-232.
Jucker, A.H. 1993. 'The structure and coherence of discourse". In H. Löffler, C.
Gromilund and M. Gyger (eds), Dialoganalyse IV Referate der 4.
Arbeitstagung Basel 1992. Tübingen: Niemeyer, 71-78.
Kotthoff, H. This volume. "Coherent keying in conversational humour. Contextualising
joint fictionalisation".
Lenk, U. 1998. Marking Discourse Coherence: Functions of Discourse Markers in
Spoken English. Tübingen: Narr.
Roberts, R.M. and Kreuz, R.J. 1993. "Nonstandard discourse and its coherence".
Discourse Processes 16: 451-464.
Rochester, S. and Martin, J.R. 1979. Crazy Talk: a Study of the Discourse of
Schizophrenic Speakers. New York: Plenum Press.
Sacks, H. 1974. "An analysis of the course of a joke's telling in conversation". In R.
Baumann and J. Sherzer (eds), Explorations in the Ethnography of Speaking.
London: Cambridge University Press, 337-353.
Sanders, T. and Spooren, W. This volume. "Communicative intentions and coherence
relations".
Schegloff, E. 1990. "On the organization of sequences as a source of 'coherence' in
Development. Norwood, NJ: Ablex, 51-78.
Seidlhofer, B. and Widdowson, H. This volume. "Coherence in summary: the contexts
of appropriate discourse".
Widdowson, H.G. 4 1983. Teaching Language as Communication. Oxford: Oxford
University Press.
Coherence and Misunderstanding in Everyday
Conversations
Carla Bazzanella and Rossana Damiano

Universita ' degli studi di Torino
1 Coherence and misunderstanding

What role does coherence play, besides supporting comprehension, in
monitoring for possible misunderstandings and recovering from a
misunderstanding? In order to answer these questions we have tried both to
investigate the matter theoretically, and to check our hypothesis against a
corpus of spoken Italian1.
Let us consider a standard case of misunderstanding taken from our
corpus2: the ambiguity3 between the title of a movie (by the Italian director
Carlo Mazzacurati), Il toro ('The bull'), and Toro ('The Bull'), which is the
nickname of a football team in Turin, caused the interlocutor to interpret the
speaker's utterance as referring to a football match instead of a movie. It is
worth noting that the interlocutor's reply in line 3 is pragmatically coherent with
her interpretation, since it is a ritual response to the announcement of a speaker
that she/he has watched a football match.
(1)
1 A. cos 'avete fatto ieri sera f
2 B. abbiamo visto il toro alla tele ↓
3 A. e cos 'ha fatto il toro T
4 B. (-)ma no(-)lLTOROi
1 A. what did you do yesterday nightt

2 B. we watched the bull on tv↓
3 A. what was the scoret
4 B. (-) no (-) THE BULL↓
176 Carla Bazzanella and Rossana Damiano
We know that in standard, successful interactions, coherence plays a

positive, crucial role, in so far as it works as a Tilter' that helps the interpreter
rule out a number of alternatives. In other words, the judgement of coherence is
in most cases essential to discriminate between different readings of a turn. With
regard to misunderstanding, coherence seems to play a double role. On the one
hand, the interlocutor's effort to ascribe coherence to the speaker's ongoing
discourse can deviate from the speaker's intention, thus creating
misunderstanding (in this case the ongoing speaker is misunderstood, cf. 2.1).
On the other hand, lack of coherence marks the occurrence of a
misunderstanding (cf. 2.2) and helps in triggering the process of repair, that is in
reconstructing coherence as a common ground (cf. 2.3). A diagnosis of the
misunderstanding and a search for the right interpretation must be accomplished
and the constraints posed by coherence are a valuable guide for human
participants in this process.
In the first part of this paper, we will examine the interactional
development of misunderstanding from the coherence point of view (2.1), and
the mismatch it produces between the speaker's and the hearer's interpretations
(2.1.1), and will also take into account the main feature of misunderstanding,
namely non-awareness (2.1.2), and the causes of misunderstanding (2.1.3). In
the second part, we will examine the symptoms of misunderstanding and the
repair process they trigger (2.2 and 2.3).
2 Interactional development of misunderstanding
2.1 "Diverging" coherence
Normally, for each turn, more than one interpretation would be suitable and
coherent, but the one intended by the speaker must be recognised. In other
words, the interlocutor has to reconstruct the interpretation and the coherence
as they were intended by the speaker, with a degree of tolerance that differs
from one conversation to another, partly depending on the goals of the speaker
and of the exchange itself. If coherence can be a requisite for correct
understanding, in some cases the interlocutor can fail in this task, assigning to a
turn a coherence that differs from the speaker's intended coherence, and an
interpretation that differs from the speaker's intended meaning. In these cases
we are faced with a misunderstanding, which we propose to characterise on the
basis of two main features (cf. Weigand forth.):
• the mismatch between the speaker's intended meaning and the interlocutor's
interpretation of a given turn,
Coherence and Misunderstanding 177
• the non-awareness of misunderstanding on the part of the person who has

misunderstood.4
2.1.1 Mismatch
The participant who fails to come up with the expected interpretation of a turn
normally succeeds in reconstructing another kind of coherence between her/his
'wrong' interpretation and the previous conversation. Up to then, we can
assume that the participants' views of the ongoing conversation have matched,
at least as far as the main points are concerned. As a consequence of the
misunderstanding, the speaker's coherence and the interlocutor's coherence no
longer match, thus creating two different contexts of interpretation for the turns
to come. After the misunderstanding has occurred, the interlocutor's reaction to
the speaker's next turn will be coherent with her/his own interpretation of the
dialogue, which no longer matches the other's. In other words, the interlocutor
who has misunderstood will behave coherently with her/his misunderstanding,
but not in a manner appropriate to the speaker's intentions. Adopting a
geometrical metaphor, we could say that, if before the occurrence of a
misunderstanding the participants' views of coherence in conversation
coincided, afterwards they become like parallel lines, eventually diverging
perceptibly, as is the case in the following fragment which took place in a
context of doctor/patient interaction. In this episode, while the patient was
dressing behind a curtain, after an x-ray examination of her hip-bone, the doctor
was examining the x-ray outside, and noticed the presence of a metal paper clip
in the patient's abdomen. All at once, the doctor asks, without showing her the
x-ray:
(2)
1 M. scusi (-) ma lei mangia graffe T
2 P. ma (-) si (-) ogni tanto (-) se capita↓
3 M. ma come (-) cosa vuol dire ogni tanto (-) quando le capita f
guardi che le fanno male ↓
(-)
4 P. si (-) lo so che fanno male (-) ma infatti ne mangio una
volta ogni tanto ↓
5 M. si (-) ma sarebbe meglio se non ne mangiasse del tutto ↓
6 P. ma non è che tutti i giorni io mangio krapfen ↓
7 M. ma cosa dice t (-) io ho detto graffe↓ne ha una nell 'addome ↓
'1 D. excuse me (-) are you in the habit of eating gräffe1 ↑

2 P. well (-) yes (-) now and then (-) occasionally ↓
3 D. what do you mean occasionally↑ listen↓ they're bad for you↓

(-)
4 P. yes (-) I know they are (--) but as I said I only eat them now and then↓
5 D. yes (-) but it would be better if you didn't eat them at all↓
6 P. but I don't eat krapfens2 every day↓
7 D. what are you talking aboutt (-) I said graffe↓ you have one in your
abdomen↓
In this example, the misunderstanding of the word grqffe uttered by the

doctor lasts as many as five turns, ending only when the interlocutor (the
patient), in the sixth turn, reveals her own mishearing (i.e. krapferir= doughnut
vs. graffe2= staples). Due to the ambiguity raised by the use of the personal
pronouns "they" and "them" in turns 3, 4, 5, their discourse proceeds in parallel
(cf. Figure 1), without striking contradictions, until turn 6 reveals the 'wrong'
interpretation. The participants' effort to maintain conversational coherence is
evident here, since, although they appear more and more surprised by each
other's replies, they both succeed in finding a sense in what they are hearing5.
Despite the nonsensical answers he receives, the doctor goes on asking the
patient about the object she has swallowed. At the same time, the patient, as she
declared later, was increasingly surprised and offended by the doctor's reactions
to her answer, which she imagined was a consequence of some serious internal
disease appearing from the x-ray.
The symmetrical usage of repeated anaphors is particularly interesting
here, since it underlines both the cooperative effort and the different
'perspectives' of the interlocutors which apparently collide. Ambiguity and
misunderstanding become evident only with the explicit mention of krapfen in
line 6, and the self-repetition (cf., e.g., Bazzanella 1996) of graffe, which works
as a correction, in line 7.
In general, when a misunderstanding occurs, the speaker's intended
coherence is not recognised by the interlocutor, thus creating a misalignment
between the interpretations of the dialogue given by the speaker and the
interlocutor. The distance between the participants' interpretations of the
conversation will normally lead, later on, to one of them being unable to
integrate subsequent turns in her/his interpretation of the interaction in a
coherent manner. Since they have different contexts at their disposal for the task
of interpretation, they lack the basic condition for reconstructing, at any
subsequent turn, the other's intended coherence. At the same time, unless they
discover the failure in mutuality, they continue to make their contributions by
referring to different contexts.
realignment
clash
parallel
discourse
alignment
I interlocutor M interlocutor P
Figure 1. Diverging coherence in example 2
The 'wrong' coherence results from cohesion links, topical coherence, or

other mechanisms. Many examples taken from real data show that 'wrong'
interpretations often remain coherent with the context of the preceding
conversation, mostly on the basis of co-reference or lexical cohesion, as in (3),
which is taken from Mauro et al. 1993 (C1/MA16). The interlocutor, i.e. C,
misunderstands the referent of the pronoun 'it' in 6, identifying its referent as
'the cheque', instead of'the account', as intended by the speaker, B. Although
it is not the intended referent, 'the cheque' could be a possible subject of the
expression "to be in someone's name", since both current accounts and cheques
can be in someone's name. The only difference between the two objects that
helps the speaker in identifying the occurrence of an interpretation failure in turn
6, is that, while cheques can be blank, accounts must necessarily be in
someone's name6.
(3)
1 B. con un assegno no <??>
2 C. io ho il conto qua
3 B. ha il conto qua quattro otto sei? duecento lire di commissione
4 C. ha bisogno del numero del conto # quello gia' e'piu' difficile
[silenzio]
5 C. ecco # /'/ conto e' nove tre due otto barra uno
6 B. come e'intestato?
7 C. non e'intestato
8 B. no come e' intestato il conto
9 C. ah XYZ Francesca
'1 B. by cheque, right <??>

2 C . I have an account at this branch
3 C. you have an account at this branch four eight six? two hundred lire
commission
4 C. you need the account number # ah, that's more difficult
[silence]
5 C. here you are # the account is nine three two eight slash one
6 B. whose name is it in?
7 C. no name
8 B. no, whose name is the account in
9 C. oh, Francesca XYZ'
2.1.2 Awareness vs. non-awareness

Even in the absence of a misunderstanding, however, the ascription of
coherence and the consequent negotiation play an important part in producing
comprehension. Consider, for example, the great importance of clarification
sub-dialogues, repair turns, self and other corrections for leading the
participants towards mutual understanding7.
Many examples, taken from real data, show that when one participant fails
or finds difficulties in integrating a new turn in a coherent view of the ongoing
conversation she/he is aware of it and normally tries to overcome the obstacle
by means of many devices, including clarification requests, requests for
confirmation, discourse markers and so on8. Weigand's (forth.) notion of
coming to understanding fits perfectly here: "Dialogic action games are always
interactive action games negotiating on an interactive level the general purpose
of coming to understanding." By contrast, when a misunderstanding occurs, no
sign is available to the participants until a breakdown in coherence comes; in
Weigand's terms: "The interlocutor who misunderstands is not aware of it".
2.1.3 Causes
The role of ambiguities in generating misunderstanding has been underlined by
several scholars (cf., among others, Blum-Kulka and Weizman 1988; Zaefferer
1977), and seems to be confirmed by the examination of the data: although
there are instances of misunderstanding in which no ambiguity is present,
ambiguity causing a misunderstanding was found in approximately 70 % of the
instances considered. In example (4), again taken from our corpus, the
referential ambiguity of the subject of the verb phrase "they don't exist", in the
first turn, reflects on the understanding of the sentence topic. After the
misunderstanding occurs in line 3, the speaker makes it clear, in line 4, that he
was talking about limits and not products. Interestingly, it is difficult, here, for
the interlocutor to accept such an interpretation, as is shown by her surprised
reaction. Nevertheless, the 'wrong' interpretation displays cohesion with the
first sentence, since 'they' could also refer to 'the products' instead of 'the
limits', even though the latter is preferred as the nearer.
(4)
1 A. ci sono una serie di prodotti per i quali noi non diamo limiti perché
ufficialmente non ci
2 sono
3 B. come non ci sono↓(-) non ci sono i prodotti (-) ufflcialmente T
4 A. non i prodotti (-) i limiti↓
'1 A. there's a series of products for which we don't give limits as they don't
exist
2 officially
3 B. what do you mean don't exista↓(-) the products don't exist
(-) officially↑
4 A. not the products (-) the limits↓'
Ambiguities in a speaker's utterance (as in fragments 1 and 4) are just a

small part of the several factors that can cause a misunderstanding at any level
of a turn's meaning, setting the premises for a misalignment between the
speaker's and the interlocutor's interpretations. One or more of these factors -
which we propose to label triggers, rather than causes (to underline the non-
deterministic role they play in predisposing communication to a negative
outcome) - can appear at a particular point in an interaction, making the process
of understanding more difficult, but not necessarily resulting in a
misunderstanding. Furthermore, while the responsibility of the hearer and of
her/his interpretation choices has been underlined in the literature, nothing has
been said about the other component parts of communication: in addition to the
triggers on the hearer's side, different sets of triggers should be taken into
account, corresponding to the role of the speaker, the structural components of
communication, and the interaction itself (cf. Bazzanella and Damiano 1997).
2.2 Lack of coherence as a symptom of misunderstanding
How do speakers and interlocutors arrive at a diagnosis of what has occurred?

And what part does coherence play in it? As we have already said, if a
misunderstanding occurs, coherence as a cooperative achievement apparently
collapses, since a turn arises that cannot be integrated in one participant's
context of interpretation. But, if we assume that the partner is still cooperating
(cf Grice 1989), the lack of coherence must rather be seen as a symptom of
misunderstanding: the unexpected and non-coherent behaviour on the part of
the partner must be explained as coherent with an interpretation of the past
conversation that differs from her/his own. Consequently, the person who has
detected the presence of a misunderstanding has to reason about it, in order to
find the cause for the diverging coherence. The lack of conversational
coherence as a symptom of misunderstanding has been underlined, in different
forms, by many researchers, both in linguistics and in computational linguistics.
Schegloff (1992) adopts the expression 'sequentially coherent' about the turns
that fulfil the expectations established by the previous conversation, matching
with the interpretation context constructed until that point. Bublitz (1989)
argues that for the ascription of coherence to an utterance to be confirmed it has
to match with subsequent data from the same speaker. With regard to
computational linguistics, McRoy and Hirst (1995) have designed a default
system that exploits preferences based on adjacency pairs to evaluate the
coherence of new turns and to detect misunderstandings.
Real data evidences that the task of reconstructing what the 'wrong'
interpretation consists of is normally a simple one for human participants,
probably because they are aware of the weak spots (ambiguities, speaker's
failures, etc.) of their past contribution to the conversation and of their own
interpretation of their partner's turn. As speakers, in particular, they seem to
have an immediate understanding of what has gone wrong and their repair turns
are thus explicitly aimed at correcting the faults detected in their partners'
interpretation. Moreover, the participant who tries to reconstruct how the
other's interpretation differs from her/his own (no matter whose fault it is) is
guided in this task by the coherence constraints of both the turns that come
before the misunderstanding and the turns that follow it. In this sense, the whole
context that surrounds the misunderstanding is not only meaningful for the
interlocutor's beliefs and intention, but also because it provides the repairing
participant with a valuable guide for the task of reconstructing the alignment
between the participants.
2.3 Coherence and repair
We will deal here with four aspects of the process of repair which occurs after a
misunderstanding:
• the agent of repair
• the phases of negotiation
• the devices of repair
• the placement of repair.
When coherence can no longer be assigned, the participants must work
together to reconstruct the lost common ground essential to successful
communication, eventually abandoning their 'individual' visions of coherence.
In this sense, coherence in a conversational and interactional context can never
be said to be static or ascertained, and, even when it is treated as if it was, it can
suddenly turn to uncertainty and need further work for its re-establishment9.
The participant who realises that a misunderstanding has occurred,
whatever her/his role in the conversation, first has to signal the difficulty, so that
she/he can act together with her/his partner to re-establish the lost common
coherence. This result is obtained by means of a negotiation, open to different
endings. The form this negotiation normally assumes is the following (cf. Figure
2).
Figure 2. The interactional development of misunderstanding

Here, we are dealing with the second step: cDoes the participant who has
detected the misunderstanding make a repair?'10. The participant who has
detected the misunderstanding can make a repair or not (second step). If she/he
does not, a communication breakdown is likely to happen.
If the participant who carries out the repair turn proposes a diagnosis of
what has happened and a consequent solution, her/his partner can confirm or
disconfirm the proposed reading (third step): cf. line 9 in 3, where the discourse
marker oh underlines the interlocutor's uptake.
If the interlocutor refuses the repair turn, there can be a conversational
breakdown, or a shift in conversational topic if they decide to disregard it and
go on (see third step, b). By contrast, if she/he accepts it completely (see third
step, a), we have a 'fresh start'; if the acceptance is partial, participants enter a
cycle of further negotiation.
The effectiveness of repair turns in human conversation can be explained
as a consequence of the participants' intuitions about lost coherence (as we said
earlier, cf. 2.2). There is, in fact, a strong relationship between a participant's
intuition about the misunderstanding and her/his repair. The speakers, in
particular, after reconstructing what has triggered the interlocutor's 'wrong'
interpretation and coherence, usually try to indicate the intended coherence
more clearly, not only by simply correcting, as they sometimes do, but also by
stressing or reinforcing the relevant features in their misunderstood turn. This
aim is achieved in several ways, depending on the factors that have triggered the
misunderstanding: by a more or less literal repetition, when the speaker thinks
that the interlocutor did not catch her/his utterance properly, by specifying what
the correct alternative was if there was ambiguity, or by explicitly contrasting
the interlocutor's interpretation, when it is easily reconstructed, with her/his
intended meaning (cf. 4).
With regards to the placement of the repair, the interlocutor's incorrect
interpretation is in most cases detected immediately afterwards by the speaker,
thanks to her/his inappropriate reaction, and corrected by the speaker in her/his
next turn. This case corresponds to 'third turn repair' in Schegloff's (1992)
terms.
More rarely the interlocutor her/himself, thanks to the speaker's next
contributions to the conversation, realises she/he has misunderstood and
corrects her/himself (fourth turn repair).
These are the most common patterns, but they can be extended over
several turns, so that repair can take place virtually in any of the turns that
follow the misunderstood one, starting from the third one, as shown by our
data, where instances of repairs can be found at the twelfth, fifteenth, and
thirtieth turn.
3 Conclusions
We have briefly dealt with three aspects of the complex relation between
coherence and misunderstanding:
• the mismatching of speaker's and interlocutor's coherence as the source of
misunderstanding;
• the awareness of the lack of coherence as the start of a repair process;
• the repair of misunderstanding and the re-establishment of 'shared'
coherence.
The phases of this interactional cycle, based on negotiation and sequential
processes, have been analysed on the basis of real data, and some linguistic
devices (among the several involved in this process) have been stressed.
Within this framework it is evident that understanding itself, let alone
misunderstanding, is not a clear cut notion, but a product of coherence
ascription and, when necessary, of negotiation.
Notes
1. Our 'corpus' includes 63 instances of misunderstanding, partly extracted from existing
text corpora, partly collected for the specific purpose. Two of the existing corpora are
published and two smaller ones are unpublished. In the first group, Mauro, Mancini,
Vedovelli and Voghera 1993 is a collection of excerpts taken from both telephone and
face-to-face conversations, in different contexts that range from formality to informality,
while Gavioli and Mansfield 1990 consists of transcriptions of complete book-shop
encounters. The two unpublished corpora were collected by Cristina Ferrus and Orsola
Fornara for their theses in the Philosophy of Language (1994, University of Turin, Italy).
2. Usual transcription conventions are adopted in the corpus we collected; conventions of
other corpora are maintained.
3. See Bazzanella and Damiano 1997 for a proposal for classification of the levels of
misunderstanding.
4. The second feature, in particular, rules out instances of planned or intentional
misunderstanding and exploitation of misunderstanding (cf. Souza 1985).
5. It is worth noting that other interlocutors, who were listening to this conversation in the
adjacent room, realized immediately that a misunderstanding was occurring.
6. The reason why this kind of failure is not always immediately recognizcd is that thc
mechanism sketched above is not absolutely safe from failures itself, since every new turn
is open to a range of different interpretations, thus allowing for the misalignment
occasionally to remain covert for several turns.
7. In fact, if the interlocutor has to make an effort to come to the correct understanding, the
speaker also makes a complementary effort to show her/him the way. Preference
mechanisms, like adjacency pairs, topical coherence or different types of cohesion
(repetitions, collocations, discourse markers, etc.) have been indicated by researchers as
examples of the devices used by the speaker to obtain this aim, but, in spite of these
devices, misunderstanding occurs and is dealt with interactionally.
8. These devices are often accompanied by discourse markers, which serve to communicate
the interlocutor's uncertainty and to make it a mutual goal to establish the compre
hension the interlocutor is not able to reach on her/his own.
9. In monological texts, too, a re-interpretation is often necessary, cf. Conte this volume.
10. We are not interested here in step one, where misunderstanding is not detected, and the
interaction goes on without any changes.
References
Bazzanella, C. (ed). 1996. Repetition in Dialogue. Tübingen: Niemeyer.
Bazzanella, C. and Damiano, R. 1997. 'Il fraintendimento nelle interazioni quotidiane:
proposte di classificazione". Lingua e Stile XXXII: 173-200.
Bazzanella, C. and Damiano, R. (forthcoming). "The interactional handling of
misunderstanding in everyday conversation". Journal of Pragmatics (special
issue on Misunderstanding).
Blum-Kulka, S. and Weizman, E. 1988. "The inevitability of misunderstandings:
discourse ambiguities". Text 8: 219-241.
Bublitz, W. 1989. "Topical coherence in spoken discourse." Studia Anglica
Posnaniensia XXII: 31-51.
Conte, M.-E. 1988. Condizioni di Coerenza. Ricerche di Linguistica Testuale.
Firenze: La Nuova Italia.
Gavioli, L. and Mansfield, G. 1990. The PIX1 Corpora. Bologna: Cooperativa Libraria
Universitaria Editrice Bologna.
Grice, P. 1989. Studies in the Way of Words. Cambridge, MA: Harvard University
Press.
Mauro, T. de, Mancini, F., Vedovelli, M. and Voghera, M. 1993. Lessico di Frequenza
dell'Italiano Parlato. Milano: Etas.
McRoy, S. and Hirst, G. 1995. "The repair of speech acts by abductive inference".
Computational Linguistics 21: 435-478.
Reilly, R.G. (ed). 1987. Communication Failure in Dialogue and Discourse.
Amsterdam: Elsevier.
Schegloff, E.A. 1992. "Repair after next turn: the last structurally provided defense of
intersubjectivity in conversation". American Journal of Sociology 97: 1295-
1345.
Souza, F. de. 1985. "Dialogue breakdown". In M. Dascal (ed), Dialogue: an
Interdisciplinary Approach. Amsterdam: Benjamins, 415-426.
Weigand, E. (forthcoming). "Misunderstanding: the standard case". Journal of
Pragmatics (special issue on Misunderstanding).
Zaefferer, D. 1977. "Understanding misunderstanding: a proposal for an
explanation of reading choices". Journal of Pragmatics 1: 329-346.
The Effect of Context in the Definition and Negotiation of
Coherence
Anna Ciliberti
Università per stranieri, Perugia
1 Introduction and aims
My contribution is based on the analysis of a corpus of 60 tape-recorded oral

examinations, all in the field of the Humanities and Social Sciences, collected1 at
a major Italian university during the last couple of years. What I shall do in this
paper is focus on the way the definition and negotiation of coherence is affected
by certain aspects concerning the oral examination. More specifically, I shall
look at topical coherence in different exam prototypes and at students' strategic
use of incoherent (i.e. evasive) answers - another aspect of what Bublitz and
Lenk refer to as 'disturbed coherence' (this volume).
2 T h e oral examination in Italian universities
The oral examination in Italian universities is the last stage in a complex and
diversified didactic process during which the student experiences several
didactic modes and comes into contact with different types of texts. This
examination may assume different forms depending upon factors related to the
examiner, to the candidate, and to the subject taught.
As to factors related to the examiner (in an asymmetrical situation like the
oral examination), the examiner's individual traits of personality, age, gender,
geographical and social origin, together with the subject taught and her
pedagogic stance, play an important role in determining the way in which the
examination will be conducted. Among this set of variables, a special place is to
be ascribed to the examiner's "pedagogic voice" - as Bernstein (1990: 190) calls
it - i.e. to "the distinguishing features of specialised pedagogic communicative
practices".
190 Anna Ciliberti
Amongst the set of variables related to the student's psychological,

experiential and socio-cultural background, the strategies she adopts to cope
with the asymmetrical situation she finds herself in are of special importance.
Equally significant is the degree to which the student's strategies relate to those
of the examiner.
One of the consequences of this interplay of variables is that, although
from a linguistic point of view the oral examination takes the form of a
dialogue, it may at times be a dialogue only in the sense of talk between people,
of Gespräch, in terms of turn-taking. It cannot be considered a dialogue if by
this term we mean a mutual undertaking - in this case by educator and student -
a mutual commitment requiring, among other things, "argumentative treatment
of the propositional content" (Ehlich 1985: 384; Ciliberti, forthcoming).
3 The context-dependency of coherence

3. J Coherence and activity type
Researchers in educational psychology have shown how the situational context

in which linguistic action takes place influences the cognitive activity people
engage in. For instance, the situational context in which questioning takes place
- everyday vs. classroom interaction - prompts pupils to attribute different
meanings to the questioning act and thus to modify the kind of cognitive activity
they believe they are required to engage in (Schubauer-Leoni and Grossen
1993: 460). By conforming to the implicit rules valid in the two contexts pupils
will either blatantly refuse or readily attempt to solve absurd problems such as
"There are 200 goats and 25 sheep on a boat. How old is the captain?". In the
everyday interactional context logical consistency prevails and pupils tend to
discard the problem as nonsensical; in the classroom context, however,
contractual congruity prevails, and students try to solve the problem using the
available data (cf. above).
In order to assess whether a contribution is coherent or not, we thus have
to take into account the relevant contextual features of that contribution. This
has to do, first of all, with the activity type under consideration, since this
"incorporates sets of obligations and predispositions, which act as constraints
not only on which messages should be conveyed and how they should be
conveyed but also on how messages should be interpreted" (Coupland and
Coupland 1997: 7). Consider the following incoherent sequence taken from a
transcript involving a complainant and a police officer who has been called to
intervene in an incident of domestic violence (McElhinny 1992; Coates 1995:
54):
Coherence Negotiation in Context 191
(1)
Complainant: You know and all of a sudden he said 'old drunken bitch' I said
What' re you talking about' I said. We're together we gotta help
each other' you know
Police officer: —» What's your date of birth?
As Coates (1995: 55) comments upon the sequence: "The formal

procedures which police are obliged to follow militate against their responding
sympathetically [cooperatively] - and coherently - in such situations".
What is of interest here is that the conversational maxims operating in
cooperative, symmetrical situations are not present in conflict or asymmetrical
situations. The definition of coherence valid in cooperative conversations
between equals is not necessarily valid in conflict or asymmetrical situations as
the following excerpt from an oral examination shows. Here the interactants
uncooperatively and incoherently interrupt each other, trying to follow their
own line of reasoning and to reach their own goals. Of course, it is the examiner
who 'wins' in the end: 2
(2) (Contemporary history)

(The general topic under discussion is Enrico Berlinguer's 'historical compromise')
1. S: Erano iniziate le bom*be%.
2. E: *Con%temporaneamente erano iniziate le bombe neonaziste. E quindi: ecco,
il rischio per la democrazia, il rischio per lo stato, no?
+ + Ecco, que*sta è lapropo -%
3. S: *Queste bo%mbe in questi anni [3syllj avevano instaurato=
4. E: =Ed è una proposta che, tral'altro, trova dei consensi nella piccola e media
borghesia italiana.
' 1. S : Bombs had begun to make their appea*rance%.

2. E: *At the% same time neonazist bombs had appeared. And: here was theriskfor
democracy, theriskfor the state, ok? + + Yes, this *is the propos -%
3. S: *These bo%mbs during those years [3 syll] had established=
4. E: =And it is a proposal that, among other things, was approved by the Italian
petite and middle bourgeoisie.'
The interactional sequence is thus the following: (i) the student introduces
a topic (turn 1); (ii) the examiner interrupts the student and expands the topic
(turn 2); (iii) the student tries to go on with what she was saying disregarding
the last part of the examiner's turn (turn 3); (iv) the examiner takes no notice of
the student's topic resumption and brings talk to an end (turn 4).
192 Anna Ciliberti
3.2 Topical coherence and exam prototypes
I have maintained that the definition of coherence changes according to activity

type. But oral examinations may result in rather different activity types. In my
data I have distinguished two exam prototypes that, using a medical metaphor,
might be called: the diagnostic and the treatment types of examination.
In the diagnostic manner of conducting an examination, the examiner's
goal is not so much to discuss on equal grounds with the student, but to have
certain facts reported in a - often rigidly - pre-defined manner (Ciliberti 1998).
The examiner's institutional role of assessing the student's mastery of a certain
knowledge tends to prevail over other roles and other discursive positions. As
to students, the primary fiinction of their discourse is to report certain facts in
order to show they have acquired a savoir enseigné as opposed to the
examiner's savoir savant (Chevallard 1985). In other words, it is the intertext
that constitutes the real discursive object of interaction. The question-answer
scheme is the central dialogical structure; the linguistic event is fragmented by
frequent topic change, as the extracts below illustrate.
(3) (Language teaching methodology)

1st Q E: Mi parli della ricercazione
2nd Q E: Vabbè. + + Mmhh, mmhhh. Senti, mi parli un pochino di:: (o5)

teori e dell 'apprendimento di una lingua straniera + o: perlomeno,
non materna.
[...]
3rd Q E: Questa è stata una corrente, una scuola. Poi? + + Di piü recente
non ti ricordi nulla?
4th Q E: Senti. Lasciamo perdere + le ipotesi. + + Invece tu stavi parlando

di ipotesi descrittive. Mmhh? E hai nominato Austin. Chi è Austin?
[...]
5th Q E: Senti. Mi sapresti definire il sillabo rispetto al curricolo?
[...]
6th Q E: Okay, ehhh + c'è una parte del programma che ti è piaciuta di
meno?
(05) Qual è la parte del programma che ti è piaciuta di meno
(05). No, non te la chiedo. (ride)
7th Q E: Senti una cosa. Quindi del libro 'Le facce del parlare' ti sono + ti è
+ piaciuto molto il capitolo sui segnali discorsivi?
S: Si.
E: A cosa si riferisce con 'segnali discorsivi'?
' 1st Q E: Tell me something about action research.
2nd Q E: OK. + + Mmhhh. mmhh. Listen, will you talk a little bit abo::ut (05)
second language learning theories + or: at least, not mother tongue.
3rd Q E: This was an approach, a school. And later on? + + Don't you
remember something more recent?
4th Q E: Listen. Let's leave it at that + the hypothesis + + You were talking
instead about descriptive hypotheses. Mmhh? And you mentioned
Austin. Who is Austin?
5th Q E: Listen. Could you define syllabus as opposed to curriculum?
6th Q E: Okay, ehhh + is there apart of the programme you liked the least?
(05) Which is the part of the programme you liked the least? (05) No,
I won't
ask you about that. (laughs)
7th Q E: Listen. So in 'Le facce del parlare' you liked + you liked the chapter
on 'discourse signals' very much?
S: Yes.
E: What does the expression 'discourse signals' refer to?'
In the 25 minute long examination, the examiner changes topics seven

times. The disconnected question-answer structure here is mainly motivated by
the student's insufficient grasp of the course programme. But the diagnostic
way of conducting an examination may also be an examiner's stylistic trait, a
way of assessing the student's mastery of a certain body of knowledge as
thoroughly as possible. In both cases what prevails in this type of examination is
the examiner's institutional role of assessing and evaluating.
In the second way of conducting an oral examination, the treatment type
of examination, a much more peer-like, conversational style is adopted and
students themselves may ask questions. Interactants adapt and negotiate their
linguistic action. Theirs is more like a 'conversational duet' than the hierarchical
structure of T (examiner) ask questions, you (student) provide answers' -
typical of the first exam-prototype. The examiner may supplement her
institutional goal of assessing the student's mastery of a pre-determined body of
knowledge with the further goals of adding to the student's knowledge and of
holding a discussion with her. The examiner allows the student the freedom to
194 Anna Ciliberti
express her opinions and to pose questions to the examiner as well. The
outcome is a much more argumentative and coordinated kind of encounter.
From the point of view of topic, the treatment type of examination tends
to concentrate on fewer, broader and more related topics. One topic leads to
another by contiguity or similarity. The examiner links together the topics under
discussion by way of analogy, comparison, expansion, contrast, etc., much the
way people do in face-to-face conversation. When topic change occurs, it is
normally explicitly declared, justified or explained. The following is an example
of the second type of examination. It is 35 minutes long and organised around
four interrelated questions.
(4) (Italian literature)

(The exam is about Ariosto's Orlando furioso)
1st Q. E: Adesso prendiamo il nostro Orlando e: + vabbé. Cominci pure lei
da un momento, da un punto, da un Canto, da un episodio che le è
particolarmente piaciuto.
2nd Q E: Senta. Vuol prendere il Canto 23 dove c 'è un 'altra variazione +

visto che stiamo parlando di questo povero Petrarca che viene usato
e cambiato + vediamo se: + si, perché nella variazione di cui si
parlava
(...) Si parlava prima di petrarchismo cambiato.
[...]
3 nd Q E: Si ri corda un altro caso in cui (...)
4th Q E: Cambiamo + cambiamo discorso sennò questo (4syll) la mette in

difficolta.
'1st Q E: Now let's take our Orlando an: + well. You can start from a moment,
a point, a Canto, an episode you particularly liked.
2nd Q E: Listen. Would you mind going to Canto 23 where there is another
variation + since we're talking about this poor Petrarca who is used
and changed + let's see i:f + yes, because in the variation we were
talking about before (...) We were talking about changed petrarchism.
3 rd Q E: Do you remember another case in which (...)
4th Q E: Let's change + let's change topic since this one (4 syll) is causing you
problems'
Topicality is maintained throughout the exam, connections are established

by way of analogy and comparison. The only question that is thematically

unconnected is the last one; however the examiner explicitly makes out a case
for choosing an unrelated topic, thus concluding by connecting even this last
question to the previous ones: 'sennò questo la mette in difficoltà' ('since this is
causing you problems').
As I pointed out before, the examiner's choice of a 'diagnostic' rather
than a 'treatment' way of conducting an examination often depends on the
student's insufficient knowledge of the course programme and on the way she
answers the examiner's questions. But such choice may also reflect an
idiosyncratic stylistic trait: examiners tend to adopt the same standard format in
all examinations. Examples (4) and (5) are a case in point; the examiner is the
same, but in (4) the candidate is a poor student while in (5) the examinee is a
very good student. In both examinations topical coherence is pursued and
maintained throughout, as is evident in example (5):
(5) (Italian literature)

(The exam is about Ariosto's Orlando furioso)
1 st Q: E: Si vuol riguardare un attimo + anche attraverso appunti o schemi +
tutto quello che vuole, il secondo Canto e mi dice come è fatto, che
episodi ci sono, e perché è fatto in questo modo. Insomma + tutto
quello che le potrebbe venire in mente per spiegare a un altro
questo Canto del poema.
t...]
2nd Q: E: Vabbene, proprio questo -proprio in questo secondo canto, mi
sembra, c 'è la variazione tra le due, tra Rinaldo e Bradamante.
3rd Q: E: Allora, andiamo un po ' al trove. Ecco + la faccio rimanere più o

meno nello stesso tema. (...) Ma allora, questa dichiarazione del
canto 32 è paragonabile a quella del secondo canto?
4th Q: E: Va bene, senta, le faccio leggere l'ultima ottava e poi la rimando

via eh? + Vedo che c 'ha molto ben chiare le cose. + + Senta, qui io
- il tema che c 'è è evidente ma non mi importa tanto il tema
inquadrato nella (...) Io vorrei proprio la spiegazione letterale delia
prima ottava del Canto terzo.
'1st Q: E: Would you mind looking at + the second Canto, you can look at your
notes, outline, or anything you like, and tell me how it is constructed,
what episodes there are, and why it is constructed this way. In short,
+ anything that may come into your mind in order to explain this
Canto of the Poem to somebody else.
196 Anna Ciliberti
[••■J.
2nd Q: E: Well, exactly this - exactly in this second Canto, I think, is the
variation between the two, between Rinaldo and Bradamante.
3rd Q: E: Well, let's go somewhere else. Here + I'll have you stay more or less
on the same topic (...) But then, this declaration in Canto 32 is
comparable to the one in the second Canto?
4th Q: E: Well, listen, I'll have you read the last octave and then you can go,
ok? + I see that you have things very clear. + + Listen, here I - the
topic is quite evident, but I'm not interested in having a thematic
framing in (...)I'd like really a literal explanation of the first octave of
the third Canto.'
Again, the last question has no thematic ties with the previous ones. Yet
the new topic is legitimated by the declared wish to contrast it with the previous
ones, and these are all resumed under a superordinate topic: 'qui io ... non mi
importa tanto il tema inquadrato' ('here, I am not so much interested in the
thematic framing').
An extreme example of a 'treatment' examination is one in which the
student is approximately 35, sophisticated and unusually well-read. What is
striking about this exam is its conversational character. The examination
becomes a symmetrical event in which the student is on a par with the examiner.
Matching assessments and matching stories contribute to the attainment of
coherence both on a local and on a more global level. The whole examination
deals with one single topic: dead languages and how they can signify. In
example (6), the examiner is commenting on a Latin quotation in the candidate's
written assignment:
(6) (General linguistics)

E: Perché una lingua morta? e qual è il fascino di una lingua morta? Ow ro,
l 'uso della lingua morta sta nell 'ordine di una necrofilia? (ride)
S: Dio + se potessi dare una risposta a questa domanda, ehh, devo citare un film,
devo citare Tn nome della rosa', che dice, appunto, se lo sapessi insegnerei
teologia alla Sorbona.
E: (ride) Adesso quelli della Sorbona dicono se io lo sapessi insegnerei negli Stati
Uniti perché pagano di piu.
(E & S ridono)
'E:Why a dead language? And what is the fascination of a dead language? Rather,
the use of a dead language is a kind of necrophilia? (laughs)
S: God + if I could answer this question, ehh, I have to quote one film, I have to
quote 'In the name of the rose', that says, in fact, if I knew I would teach
theology at the Sorbonne.
E: (laughs) Nowadays those who teach at the Sorbonne say if I knew it I would
teach in the United States because they pay more.
(both laugh)'
The student's witty response is taken up and expanded by the professor.

In this way the examiner establishes a sort of polite form of solidarity with the
candidate.
4 Negotiating coherence
Activity type influences not only the establishment of topical coherence patterns
but also the way coherence is negotiated. People belonging to a common
universe of discourse and sharing the same social goals will try to make their
contributions coherent and will go to any length to detect coherence in what
they hear. This is not necessarily the case when the universe of discourse is not
a common one - i.e. when interactants do not share the same social goals - or do
not attune to one another's individual goals. In the latter case the interaction
may turn out to be non-cooperative and discursively incoherent.
In the exam situation, the student's goal, 'to get through the examination',
is re-defined by the examiner's as 'applying for eligibility to a 'pass". These two
goals may at times (at some level at least) be mutually exclusive and this "may
account for discrepancies in the participants' verbal behaviour" (Sarangi and
Slembrouck 1992: 128). These 'discrepancies' often result in incoherent
sequences.
In the exam situation, in order to reach their goals, students may
consciously - i.e. strategically - produce inconsistent, incoherent answers to the
examiner's questions.
When students do not know the answer to a question, they may take one
of the following two paths: (i) they may openly declare they don't know the
answer; (ii) they may give an evasive answer. Evading a question, or getting
around it, is a typical incoherent way of answering.3
According to Bull and Mayer (1988) question evasion involves the
following strategies:
• ignoring the question asked

• questioning the question
• attacking the question
198 Anna Ciliberti
• apologising
• stating that the question being asked has already been answered
• declining to answer the question
• repeating an answer to a previous question and making a political point
Among this set of strategies, only 'questioning the question' - in the sense
of asking for clarification - is adopted by students during an examination.
Students cannot risk employing other strategies such as 'attacking the question',
or 'repeating an answer to a previous question and making a political point', or
'stating that the question being asked has already been answered'. The most
feasible course of action left to the student would appear to be the adoption of
yet two other strategies:
1. producing a totally unrelated, topically incoherent, response, in the hopes

that the examiner will let the inconsistency slip by or will not notice it since
she takes the internal coherence of the adjacency pair question-answer for
granted;
2. expanding or narrowing the scope of the examiner's question without really
providing the desired answer.
What is of interest is that, in both cases, students use cohesive devices to

give their answer a 'coherent look'. Consider the following:
(7) (German literature)

(The topic under discussion is the 'Junges Wien' group)
E: No. + + Vediamo + qual è il manifesto di Bahr e quali sono quest e opere?
+ + se non altro quelle di Schnitzler e di Hofmannsthal.
S:→ Allo:ra. + + (ride) C'ho un sacco di vuoto in testa. + + Per quanto
riguarda + eh + le opere di questo periodo, cioè, fondamentalmente il
superamento del - dell 'estetismo come + + un semplice + eh, eh [?3 syll]
E: No. Diciamo qual è il manifesto letterario (...)
'E: No. + + Let's see + what is Bahr's manifesto and which are these works?
+ + at least those by Schnitzler and Hofmannsthal.
S:→ We:ll. + + (laughs) My head is completely empty. + + As far as + eh + the
literary works of this period, that is, fundamentally the overcoming of- of
aestheticism as a + + si:mple + eh eh [3 syll]
E: No, let's say, which is the literary manifesto (...)'
(8) (General Linguistics)

(Student and examiner are discussing a picture featuring Saint Brandano and a girl
kneeling in front of him)
1. E: E che tipo di attainte è la fanciulla?
2. S: → L'atteinte? + Du.nque. + Attante, *intanto%
3. E: *Perché% anch 'io sono d'accordo **con te%%-
4. S: **è il%% personaggio che compie un 'azione, è l'eroe che fa qualcosa.
5. E: Non necessariamente, perd, FA qualcosa.
'1.E: And what type of actant is the girl?

2. S: → The actant? + We:l1 + Actant, * first of all,%,
3. E: *Because% I agree with **you too%%~
4. S: **is the%% character who performs an action, it is the hero who does
something.
5. E: He doesn't necessarily DO something, though'
In both examples (7) and (8), the students do not know the answers to the
examiners' questions and decide to make use of evasive strategies. In (7) the
student answers in a totally nonsensical way; in (8) the student expands the
scope of the examiner's question. The advance organiser with metatextual
function "intanto" (Tor one thing'/'first of all') announces an insertion
sequence, thus establishing a "hold" (Levinson 1983: 348) in the production of
the second part of the adjacency pair. In this way it functions as a "coherence-
securing means" (Bublitz 1996). The evasive answer goes unnoticed and the
examiner intervenes to correct the student's side sequence. The original
question has been forgotten (turn 5).
The following is the same conversation as it continues:
(9)
6. E: (...) Non necessariamente. E' un ruolo logico, non psicologico, è un
ruolo logico. E' per questo che la Unguistica offre processi di
formalizzazione, eh?
7. S:→Difatti questa ragazza è inginocchiata, forse sta pregando, forse
Brandano, questo non si sa.
8. E: Si, eh eh, per*ché%?
9. S :→*Comun%que, dimostra di essere pentita della vita passata, quindi -
10.E: Sí, ma aspetta, aspetta. Io + qui + tu stai - questo è il limite della tua
cosa. Ricadi sempre nei contenuti. Ennô. Invece, ecco (...)
'6. E: (...) Not necessarily! It is a logical role, not a psychological one, it's a
logical role. It is for this reason that linguistics offers formalisation
200 Anna Ciliberti
processes, right?
7. S:→In fact this girl is kneeling, perhaps she's praying, perhaps Brandano, we
do not know.
8. E: Yes, eh eh, *why%?
9. S:→*Any%way, she shows she has repented for her previous life, therefore -
10. E: Yes, but wait, wait. I + here + you are - this is your limit. You always go
back to content. This is no good. Instead, here (...)'
In turn 7, the student links her contribution to the examiner's by the

cohesive device "difatti" ('exactly', 'in fact'), but what she says does not cohere
with the examiner's statement at all. Once again, the examiner lets the
inconsistency slip by. In turn 9, instead of answering the examiner's question
"Si, eh eh, perché?" ('Yes, eh eh, why?'), the student introduces her
contribution with an utterance-initial "comunque" ('anyway') marking return to
prior topic. However, this time the examiner stops her: "Si, ma aspetta, aspetta"
('Yes, but wait, wait').
5 Conclusion
The values that go together with certain pragmatic principles - cooperativeness,
discursive rights/obligations, and the like - and that are critical in relation to
what counts as coherent in interaction depend on the activity type taking place.
Participants in interaction do not have the same rights and obligations, and not
all forms of verbal interaction are dominated by the interactants' wish to be
cooperative. Most importantly, communicators often do not share the same
communicative and/or social goals. But in order for discourse coherence to
apply, interactants must (i) wish to cooperate (in the sense Grice uses the word
cooperation), (ii) have compatible goals, or mutually accepted different goals4,
(iii) communicate within compatible universes of discourse.
The oral examination is an asymmetrical type of encounter in which one
interactant has the institutional role of evaluating the other interactant's
possession of a certain body of knowledge in a certain field. As an activity type,
though, it may serve as a site for multiple and mixed social relations. Various
types of local contexts are intertwined and act upon each other. The examiner
may assume different role identities, thus determining differing levels of
discourse: institutional, professional, interindividual. Coherence - as I tried to
show - will apply differently according to the position - or to the dominant
position - from which the examiner speaks. In what I have referred to as the
'diagnostic' type of examination, the examiner's predominant role is
institutional, i.e. evaluative. She wants the student to report upon a given topic,
usually in a rigidly pre-determined manner. Topic discontinuity and fragmented

discourse typify this type of examination. In the 'treatment' type of examination,
however, the outcome is much more argumentative and coordinated. The
examiner connects the topics under discussion by analogy or contrast, and when
topic change occurs, she declares, explains or justifies it. Excerpt (6) constitutes
a good example of a 'treatment' examination located at the far end of the
spectrum of discursively coherent examinations: it is maximally coherent. As we
have seen, the entire examination revolves around one single topic, and
examiner and examinee discuss on equal grounds. In fact the examiner almost
regrets that the oral examination is over (out of a total of 60 examinations,
however, we have only three such examples):
(10)
E: No, no La volevo trattenere di più, perché Lei ha lavorato bene + +ma anche
perché, no? + mi pareva di chiaccherare con Lei volentieri (...)
'E: No, no, I wanted to keep you longer, because you have worked well but also I
was enjoying chatting with you'
As for students' negotiation of coherence when they do not know the

answer to a question, I have shown some of the strategies they may adopt: they
may produce topically unrelated replies as if they were coherent answers or they
may expand or narrow the scope of the examiner's question. The interesting
thing is that in both cases their answers are linked to the examiner's questions
by cohesive devices, e.g. by discursive signals with metatextual function like
'intanto' ('first of all') or by reformulation signals like 'cioè' ('that is'). This
strategic - one might want to say 'cosmetic' - use of cohesion to mask
incoherence is, paradoxically, yet another display of how important coherence is
for communicators: if your turn of talk is incoherent you should at least make it
appear coherent.
Notes
1. The examinations were collected by myself and a colleague of mine, Laurie Anderson
2. Transcription conventions
E examiner
S student
+ short pause
++ longer pause
(n) long pause (approximate duration n second)
202 Anna Ciliberti
*text% or
**text%% spoken in overlap with next/previous *text% or **text%%
= latched to previous turn
text - tone group interrupted
text- syllable cut short
text: syllable lengthened (number of colons indicates extent)
.,?! punctuation gives a rough guide to intonation
(?text) text unclear
(?nsyll) tape untranscribable: n = approximate length in syllables
(comment) non-verbal behaviour or context information)
3. See Wilson (1990) for a discussion on the controversial issue of what is and what is not
an answer to a question and, consequently, what counts as an evasive answer.
4. As Sarangi and Slembrouck (1992: 125) note, social equality is what we strive for, not
what we encounter in the social world. But social inequality does not exclude the
possibility to relate with our interlocutors on an equal basis and therefore to produce
coherent discourse.
References
Bergvall, V. and Remlinger, K. 1996. "Reproduction, resistance and gender in
educational discourse: the role of critical discourse analysis". Discourse and
Society 7: 453-479.
Bernstein, B. 1990. The Structuring of Pedagogic Discourse: Class, Codes and
Control, Vol VI. London: Routledge and Kegan Paul.
Bublitz, W. 1989. 'Topical coherence in spoken discourse". Studia Anglica
Posnaniensia XXII: 31-51.
Bublitz, W. 1996. "'I bought a freezer - you've bought a freezer - they've bought a
freezer': Repetition as a text-building device". In C. Bazzanella (ed), Repetition
in Dialogue. Tübingen: Niemeyer, 16-28.
Bull, P. and Mayer, K. 1988. "How Margaret Thatcher and Neil Kinnock avoid
answering questions in political interviews". Paper presented to the British
Psychological Association. London. (Quoted in Obeng 1997).
Chevallard, Y. 1985. La Transposition Didactique: du Savoir Savant au Savoir
Enseigné. Grenoble: La Pensée Sauvage.
Christie, F. 1995. "Pedagogic discourse in the primary school". Linguistics and
Education 7: 221-242.
Ciliberti, A. 1998. "Egocentric vs. self-denying communication in an asymmetric
event". In S. Cmeirkovà et al. (eds), Dialogue Analysis VI. Tübingen: Niemeyer,
29-38.
Coates, J. 1995. "The negotiation of coherence in face-to-face interaction: Some
examples from the extreme bounds". In M. Gernsbacher and T. Givón (eds),
Coherence in Spontaneous Text. Amsterdam: Benjamins, 41-58.
Coupland, N. and Coupland, J. 1997. "Bodies, beaches and burn-times:
'Environmentalism and its discursive competitors'". Discourse and Society 8: 7-

25.
Ehlich, K. 1985. "School discourse as dialogue?" In M. Dascai (ed), Dialogue.
Amsterdam: Benjamins, 383-411.
Fairclough, N. 1989. Language and Power. London: Longman.
Giora, R. 1988. "On the informativeness requirement". Journal of Pragmatics 12: 547-
565.
Giora, R. 1996. "Review article to Morton Gernsbacher, Language Comprehension as
Structure Building. Hillsdale: Erlbaum. 1990". Journal of Pragmatics 26: 417-
436.
Goodwin, C. 1995. 'The negotiation of coherence within conversation". In M.
Gernsbacher and T. Givón (eds), Coherence in Spontaneous Text. Amsterdam:
Benjamins, 117-138.
Hodge, R. and Kress, G. 1988. Social Semiotics. Cambridge: Polity Press.
McElhinny, B. 1992. '"I don't smile much anymore': affect, gender, and the discourse
of Pittsburgh police officers". In K. Hall, M. Bucholtz and B. Moomwomon
(eds), Locating Power. Berkeley: Berkeley Women and Language Group, 386-
403.
Obeng, S. 1997. "Language and politics: indirectness in political discourse". Discourse
and Society 8: 49-83.
Sarangi, S. 1988. Selection interviews as activity type: An intercultural perspective.
Lancaster: University of Lancaster.
Sarangi, S. and Slembrouck, S. 1992. "Non-cooperation in conversation: A
reassessment of Gricean pragmatics". Journal of Pragmatics 17: 117-154.
Sarangi, S. and Slembrouck, S. 1996. Language, Bureaucracy and Social Control.
London: Longman.
Schubauer-Leoni, M-L. and Grossen, M. 1993. "Negotiating the meaning of questions
in didactic and experimental contracts". European Journal of Psychology of
Education 8: 451-471.
Sperber, D. and Wilson D. 2 1995. Relevance: Communication and Cognition. Oxford:
Blackwell.
Trabasso, T., Suh, S. and Payton, P. 1995. "Explanatory coherence in understanding
and talking about events". In M. Gernsbacher and T. Givón (eds), Coherence in
Spontaneous Text. Amsterdam: Benjamins, 189-214.
Wilson, P. 1990. Politically Speaking: The Pragmatic Analysis of Political Language.
Oxford: Blackwell.
Young, R. 1992. Critical Theory and Classroom Talk. Clevedon: Multilingual
Matters.
Coherence in Summary:
The Contexts of Appropriate Discourse
Barbara Seidlhofer and Henry Widdowson

Universität Wien and University of London
1 Introduction
This paper is an enquiry into applied linguistics and language pedagogy. We

want to consider how far the conditions for the naturally occurring use of texts
in written communication relate to those that need to be taken into account in
getting students to learn from texts. In the first part of the paper, we sketch out
a theoretical framework in which to locate the concept of coherence, as we
understand it. In the second part, we then show how this can be brought to bear
on questions of practical language teaching.
2 Definitions
To begin with, we want to propose a clear distinction between text and
discourse, and correspondingly between cohesion and coherence. These four
concepts crop up continually in the literature, but their relationship generally
remains uncertain and ambiguous. Thus in the work of Michael Halliday, the
terms text and discourse are used in more or less free variation (see Halliday
1992). Wallace Chafe conflates them, too. In his entry in the Oxford
International Dictionary of Linguistics, he has this to say:
The term discourse is used in somewhat different ways by different scholars, but
underlying the differences is a common concern for language beyond the boundaries
of isolated sentences. The term TEXT is used in similar ways. Both terms may refer
to a unit of language larger than the sentence: one may speak of a 'discourse' or a
'text'.' (Chafe 1992: 356)
Both terms may indeed be used to refer to units larger than a sentence,
and many scholars have spoken of either text or discourse as if they were
206 Barbara Seidlhofer and Henry Widdowson
synonymous. We would argue that it is misleading to do so, and that a failure to

make a clear distinction has created its own incoherence. Reasons for keeping
the two concepts apart have been given elsewhere (e.g. Seidlhofer 1995;
Widdowson 1973, 1979, 1995), but it might be helpful to rehearse them briefly
here.
2.1 Text and discourse
As we define it, text is the linguistic product of a discourse process. As a

product it can be treated simply as data manifesting language usage whereby
you can identify patterns of collocational occurrence. This is text analysis, and
this is what corpus linguistics does so effectively. But text can also be taken as
evidence of how language is realised as use, in other words as the trace of a
discourse, of the pragmatic process of a communicative interaction. The
difficulty is that whereas data is fairly straightforwardly factual and can be
readily agreed on, evidence is a matter of interpreting the data from different
points of view, and can be argued about. Evidence is not factual at all.
On the other hand, text is the only evidence we have of this discourse
process, so, unreliable though it is, that is all we have to go on. A central
problem of discourse analysis is to establish the grounds for interpreting textual
data selectively as valid evidence of aspects of discourse (see also Edmondson,
this volume).
Discourse, as we define it, is the process of conceptual formulation
whereby we draw on our linguistic resources to make sense of reality. It is the
continual consultation and modification of the contexts which constitute our
world and therefore a continual renewal of our identity by interacting with the
third person world and second person others. This sounds rather vague, so let
us try to elaborate a little. Each person, each ego, is at the centre of an
individual reality: a schematic matrix of ideas, values, beliefs which are partially
shared because they are socially derived. This, for us, is what context is: an
internal conceptual construct. But such contexts are continually being engaged,
modified, re-aligned as individual first persons relate ideationally with the third
person world and interpersonally with other (second) persons, as they think,
learn, make sense of things, communicate. This is what we mean by the
discourse process. Communication occurs when certain contextual constructs of
first and second persons are activated by this process and brought into
approximate convergence, usually by means of language. Thus the resulting text
is a kind of symptom of communicative interaction. Here, diagrammatically, is
the model of communication we have in mind:
Coherence in Summary 207
Text
( 1 st Person) (2nd Person)
This view of text and discourse is, of course, entirely at variance with that
which holds that text (or equivalently discourse) is a quantitative unit: language
beyond the sentence. Text, in our view, has nothing to do with linguistic extent,
and indeed nothing to do with sentences. Texts come in all shapes and sizes.
They may indeed take the linguistic form of sentences in combination, but that is
incidental. They may equally take the form of isolated sentences (KEEP OFF
THE GRASS) or phrases (WET PAINT) or single words (PRIVATE) or even
letters (P). All such public notices are texts and when we recognise them as
such we engage our contextual knowledge to derive discourses from them, and
read into them what we assume to be the intended reference and force (for
further discussion see Widdowson 1995).
2.2 Cohesion and coherence
When texts do take the form of larger linguistic units, we can, of course,
consider what links the parts together, and talk about cohesion. Cohesion, then,
is a textual property and has to do with how linguistic elements relate by virtue
of their lexical or grammatical features. But again the occurrence of cohesive
devices are data which have to be interpreted as discourse evidence before you
can make coherent sense of them. This means that they have to be referred to
context. Cohesion is simply the textualisation of those contextual connections
which it is assumed need to be made explicit, but coherence is the discourse
function of realising those connections. It follows that you might derive a
coherent discourse from a text with no cohesion in it at all. Equally, of course,
textual cohesion provides no guarantee of discourse coherence. A simple
example. If you find a plural noun phrase in a text (say, autumn leaves), then
the very copying of the feature of plurality in the following pronoun they
constitutes a cohesive link and this might be read as evidence that the two are
related to a common referent. But consider text (1):
(1)
Unfortunately in the days to come autumn leaves will become a hazard to the elderly
as they fall and become a wet soggy mess on the pavement.
We find this amusing, cruelly ambiguous, because the possibility of the

elderly becoming a wet soggy mess on the pavement is within the scope of our
contextual fancy, and it is generally the case that when a text is referable to two
different sets of contextual conditions we will realise an ambiguity. But it is
important to note that this is a matter of pragmatic interpretation. There are
innumerable cases of cohesive indeterminacy which remain unnoticed quite
simply because they are edited out by the discourse process and have no effect
on coherence whatever. And this brings up an important if obvious point about
this process: it depends on a selective attention to textual features and a
disregard of those which are deemed irrelevant. Discourse depends on co
operation. This is particularly the case with written language use, where there is
no co-presence of interlocutors, and where, therefore, there can be no
reciprocal negotiation of common ground.
3 Written language use

Written texts are generally unilaterally designed and transferable, and are
consequently open to a range of different discoursal interpretations. Unlike
spoken conversation, where meanings are jointly constructed by first and second
person participants, written language is one-sided and non-negotiable in the
sense that its composition is a first person responsibility, and the second person
has no say in the matter. It follows that the conditions of production are
dissociated from the conditions of reception and there can be no certainty that
meanings that are written into a text correspond with those that are read into it.
In these circumstances, convergence becomes difficult to achieve.
Written text is of its nature only a partial record of the discourse the
writer enacts with the supposed second person, and in this respect written text
is crucially different from spoken text recorded as transcription. What we
typically get is only the first person part recorded with the second person's
anticipated responses presupposed and left untextualised. It is this partiality of
written text, of course, that allows the reader room for manoeuvre.
What is produced on a page is a text. It is the linguistic trace of a
discourse process enacted by the writer, and this process is one in which the
writer interactively takes on both first and second person roles, projecting
thoughts of his/her own, and also making assumptions about what the readers
know and believe, what their ideas and values are. Readers then take up the text
and activate it by creating their own discourse from it. They identify the text as
soon as they assign it intentionality. They interpret it as discourse only when
they relate it to the context of their own reality outside language.
4 Co-operation and interpretation

In this view, then, a written text is the trace of a discourse enacted by a first
person assuming both participant roles on the understanding that the actual
second person reader will engage with the text on these terms, thus ratifying the
role assigned by the writer. It is in this respect that there is so much reliance on
the co-operative principle. The writer envisages readers and makes assumptions
about their reaction. That is to say the writer adopts a first person position and
assigns a second person position to readers on the assumption, or in the hope,
that they will co-operatively assume that position. But what if they do not?
Since there is no reciprocal management of the discourse, readers are co-
opted without their overt co-operation, and so the writer can, of course, get it
wrong and assign a position which the reader cannot ratify because it is based
on an assumption of shared contextual knowledge, attitude, belief and so on,
which is unwarranted. In this case, the reader cannot co-operate, and so the text
cannot mediate a discourse convergence. Equally, the reader may resist the
assignment and, though capable of co-operating, choose not to do so. Or the
reader may not be the intended recipient at all, but is assuming the role of
outsider, the reader over the shoulder, the third person taking on a second
person role not envisaged by the writer. Reading somebody else's private
correspondence would be a case in point.
The general point we want to make is that where, by accident or design,
readers do not co-operate, do not, therefore, meet the conditions of discourse
convergence because there is an absence or withholding of common contextual
assumptions, then they will take up independent positions and make a coherent
discourse on their own terms. So it is that (so it is said) a man avoided taking an
escalator on the London Underground because a notice said DOGS MUST BE
CARRIED and he had no dog. Or you might perversely interpret the notice DO
NOT USE LIFT IN CASE OF FIRE to mean that the lift was never to be used,
because a fire might break out as a result. These examples, trivial though they
may seem to be, illustrate this fundamental point about the interpretation of
written texts, namely that it is dependent on the co-operative convergence of
contextual assumptions.
5 Coherence and appropriation of meaning

The general point we are making here, and it is obvious enough, is that
meanings cannot be read from the text but only into the text. Writers generally
design texts with particular discourse communities in mind and on the
assumption that the co-operative principle is in force, and that readers,
therefore, will seek to bring their interpretation in line with the writers'
intentions as they understand them. If there is a disparity between the contextual
assumptions of first and second persons, then the conditions for co-operation
are undermined, and interpretation will diverge from intention, and readers will
make texts coherent on their own contextual terms.
But making sense of a text always involves making it coherent on your
own contextual terms. That is what understanding means: you appropriate
meaning, and make it your own. And here we come to an issue of particular
pedagogic relevance. We mentioned earlier that readers might either be unable
to co-operate (even though willing), or unwilling (even though able). It would
seem obvious that we need to acquire the ability first, and then decide the extent
to which we are willing to exercise it in particular cases. Understanding is
always a matter of negotiating a convergence of different contextual realities, of
regulating the discourse process, to the degree consonant with your purpose.
But if communication is to take place at all, readers must surely be ready (in
both senses of that term) to take in what the other person is saying and allow it
as much contextual access as possible. The question then arises as to how to
make them ready. At this point we turn our attention to the pedagogic
implications of this view of coherence. But first, it might be co-operative on our
part to provide a brief interim summary of our position.
Coherence is a relative quality. It has to be because it is a matter of the
extent to which readers make sense of a text as it relates to their own contexts
and purposes. You can measure cohesion because it is textual and so takes
overt linguistic form, and this is why it is comfortable to equate it with
coherence. But coherence is a function of the discourse derived from a text and
this will necessarily always be covert and variable. So it is not a matter of asking
whether a text is coherent or not, but how coherent you can make it. Different
readers, bringing different purposes and contextual assumptions to bear will
read different meanings into the same text: they will, in other words, make it
coherent in different ways.
6 Appropriation and summarisation

The notion of appropriation is particularly significant in (foreign) language
pedagogy, where learners are offered a (new) language in order to make it their
own, to use it in contexts relevant to them, for their own purposes - in other
words, to create coherence for themselves. The standard way of presenting
learners with new language is through texts: we learn new words and structures
from texts we encounter, make these texts coherent by deriving from them
discourses which are relevant to us, and then go on to produce texts of our
own. In this sense, language learning is essentially intertextual.
In order to observe learners' coherence-making in action, it therefore
seems particularly appropriate to look at an overtly intertextual activity, such as
summarising. We see summarising primarily as making sense of an input text, as
making things coherent for oneself before trying to also make them coherent for
the readers of the summary. In that sense, all learning is summarising for
oneself: we identify what is salient and important to us, and incorporate what is
new into what is familiar.
We said above that coherence is bound to be relative; in written
summarisation it is thus a matter of the fashion, and the extent, to which readers
make sense of a text as it relates to their own realities. It seems inevitable, then,
that any summary will always have an element of individual interpretation to it,
there will always be some textual evidence of how readers account for the
meaning a text has for them personally, of how they regulate their co-operation
in relation to the writer's intentions as they understand them. And due to the
lack of overt negotiation of meaning in written discourse, there is likely to be
considerable variation in the way readers comply, or do not comply, with the
role assigned to them by the writer.
There exists, of course, a notion of an ideal summary which is as faithful
as possible to writer intention, and we encounter it daily in all walks of life both
outside and inside the school, in news summaries, film reviews, journal abstracts
and précis tasks, to mention but a few. But it is important to realise that
however powerful and useful the notion of an objective, faithful summary may
be, it is always an idealisation, and it is never absolute but always approximate,
relative to person and purpose. Bearing this in mind may be particularly salutary
in the domain of foreign language pedagogy, where there is often a tendency to
expect students to get more 'complete' meaning out of a text than was ever, or
ever can be, in the text in the first place.
In order to address these matters directly and empirically, then, an
experiment was set up to elicit summary responses from students to an input
text. The aim was to explore how respondents would derive a coherent
discourse from the original text, and how they would regulate their co-operation
according to different instructions. The respondents were Austrian university
students majoring in English, but the fact that they were L2 learners is not of
central importance here, since the educational implications of our observations
apply to all language teaching, and in fact probably to teaching and learning in
any subject.
7 Summaries vs. accounts

As part of a larger experiment (whose details are not relevant here), 133
students were split in two groups of roughly equal size, groups A and B. Both
groups were given the same text, an article from Time magazine of about 1000
words, entitled "The Dilemmas of Childlessness" (Smiglis 1988). Both groups
were required to 'reduce' the content of the input text and so to record a
coherent response to it, but under different priming conditions. Group A was
asked to write a faithful summary, whereas group B was asked to give a
personal account. The instructions were as follows:
A. SUMMARY INSTRUCTION
Please write a summary (in no more than 60 words) of the following text, capturing
as faithfully as you can the main points of the author's intended meaning.
B. ACCOUNT INSTRUCTION
Please give a brief account (in no more than 60 words) of what strikes you
personally as of particular interest in the following article. Give your account a title.
The summary instruction thus asked respondents to keep close to the text,
it did not invite them to relate the text to their own reality: they were to make it
coherent on the writer's terms. The account instruction, on the other hand, did
invite the students to engage with the text and make sense of it on their own
terms. These students, then, accessed whatever they found relevant to their own
world. With these different kinds of tasks a mapping-out of the whole range of
responses was targeted, from 'submissive' to author's intention to 'assertive' of
own conditions of relevance. And it was exactly this range of responses that
emerged. To illustrate what the responses were like, here are a few examples:
7.1 Group A (Summary)
(1)
In the 195 O's, 9% of women of childbearing age had no children; now 25% remain
childless, mostly well educated women from urban areas, who married late and work
outside the home. They either deliberately decide not to have children or postpone it,
preferring their freedom. Later many regret their childless state and try to satisfy
their nurturing instincts at others.
(2)
Those Americans who choose not to have children usually come from urban areas,
are well-educated and marry late. Some of them make the choice deliberately and
some postpone the decision until nature decides for them.
The childless often satisfy their nurturing instincts with nieces and nephews.
Some childless women think they have violated a biological law but most of
them enjoy their freedom.
(3)
In today's America, childlessness is spreading. The childless tend to be well
educated, live in urban areas, marry late and work outside the home. Basically, there
are two groups: the deliberate types and the postponers.
Nonetheless, babies seem to regain their important role. The birth rate among
college-educated women 20 to 24 years old is beginning to climb.
All three of these Group A protocols are essentially extractions of

information, corresponding quite closely in respect to content. And they are not
only summaries on the writer's terms, but actually in the writer's terms, with
many expressions lifted verbatim, or with minimal variation, from the original
texts {come from urban areas, work outside the home, satisfy their nurturing
instincts, and so on). They have something of the appearance of a textual
collage.
7.2 Group B. (Account)
(4) Is 'The Family' a dead issue?

A singular feeling of egocentricity pervades this article. No mention is made of
the aspects of family life and its influence on the mental and emotional well-being of
family members, nor of the adjustments necessary to achieve positive relationships.
The main objectives of the women interviewed seem to be self-realisation and
(instant) gratification, without considering responsibility towards others. The
opinions express the current glorification of the individual and the drive to make
changes at will.
It is interesting to note the desire for vicarious parenting, gratification of the
need for human relationships without total commitment.
(5) The Final End of the Beat Generation?

In the '50s artists like: Jack Kerouac, Neal Cassady, Allen Ginsberg a.m.o.
gave birth to the Beat Generation. They rebelled against the U.S. society, its
conformity, stereotypes and productiveness which caused depersonalisation,
unsatisfied parents and children. The family image has always been one of the most
important supports for economy and for keeping conformity. But through 'social-
well-fare' Americans became more selfish than the U.S. society wanted them to be.
No children - to fulfil ones own life. Now, birth-rate is climbing again - will that be
the final end of the Beat Generation?
(6) The childless - injured and abnormal

The article's approach to childlessness is comprehensive, including social,
feminist, philosophical dimensions of the phenomenon. The latter being the most
interesting aspect, I would have been delighted to learn more about the deliberate,
'uncompromising' childless, who are so without being subject to social
circumstances, alcohol problems or feminist aspirations. This group is hinted at by a
summing-up of characteristics, 'educated, urban areas, not religious,...', but further
investigations only bring to light women who regret their childlessness, who spoil
child-substitutes, who are disheartened by their mothers' example and question their
female 'normality'. The tragic tenor, that the childless are unfortunate after all,
remains.
(7) Structural analysis

The information provided by this article is understandably conveyed and
extremely well structured:
After an introductory passage the author makes the disposition evident; thus the
article is structured according to the criterion of deliberately chosen childlessness
(first group of women) and postponed motherhood (second group of women). Each
of these motivations is illustrated by examples, so that the reader is able to trace the
reasons for this far-reaching decision in question.
(8) Statement:
Some people want to have children, others don't. As for the latter, they may
have good reasons for their choice, or they may have none: some just will not spend
part of their lives on raising children, others turn out to regret their hesitating later
on. Personally, I don't care about people's having children or not, and I won't care
about their justification either.
These protocols are most obviously different from those of Group A in

that they are evaluations of the extracted information. The students take up
positions of their own and in effect derive their own discourses from the original
text. They are, therefore, not only different from Group A responses as a whole,
but different from each other.
Obviously it is impossible to do justice to students' responses by just
looking at so few examples (for the full data set and detailed discussion, see
Seidlhofer 1995). But what should be clear even from this tiny sample is that, in
accordance with the instructions, respondents created coherence by negotiating
a convergence of different contextual realities in relation to the task, and
responses ranged from 'willing and able to co-operate' to 'unwilling (though
clearly able)'. The set-up of this particular experiment meant that the responses
were distributed over different individuals, but it is easy to imagine the whole
spectrum being covered by any one respondent when given both kinds of
instructions.
8 Participant roles
This is also what Goffman (1981: 144f) seems to have in mind when he talks
about the "production format of an utterance", and there are distinctions
introduced by him which help conceptualise these different roles quite
powerfully. Goffman points out that a speaker/writer can fulfil three different
kinds of role: the animator is somebody who lends his or her voice to the
expression of somebody else's ideas, acting as a "sounding box", as Goffman
puts it. The one responsible for the actual wording of the text is the author,
"someone who has selected the sentiments that are being expressed and the
words in which they are encoded" (Goffman 1981: 144).
Behind these two, however, there is the principal, "someone whose
position is established by the words that are spoken, someone whose beliefs
have been told, someone who is committed to what the words say" (ibid.).
Schiffrin rephrases the distinction as follows: "an animator produces talk, an
author creates talk, [...] and a principal is responsible for talk" (Schiffrin 1994:
104). Schiffrin also points out that these positions can be filled by different
people, but also by a single individual.
Goffman goes on to say that "[t]he notions of animator, author, and
principal, taken together, can be said to tell us about the 'production format' of
an utterance" (op.cit.: 145; see also Couper-Kuhlen, this volume). As
mentioned above, then, Goffman talks about these roles in terms of producers
of language, speakers or writers. But as reading can be seen as creating
coherence for oneself, these distinctions are just as useful for thinking about the
reception side of communication. They also tell us about the 'reception format'
of an utterance.
Of course, as is illustrated by the examples from students' protocols
above, many accounts went far beyond an interpretation of the text: not only did
they make the input text their own and express their personal reaction to it, they
actually made the entire communicative event their own and, to use Goffman's
words, "staked out [their] own position" in it - in short, they acted as
principals.
It is important to remember that the distinction between participant roles
is not clear-cut any more than is that between summaries and accounts. It is
actually impossible, and nonsensical, to envisage a summary which does not
involve selection, which after all is something an author, and not an animator
would do. Likewise, it is hard to imagine an absolutely pure instance of
'principal account', without any 'contamination' whatsoever by ideas which are
not entirely the account writer's own. In fact, it would seem that the very nature
of the task, which is intertextual in the case of summaries as well as in the case
of accounts, necessitates that some ideas from the original text be embodied in
the response, whether they become manifest as overt textual features on the
surface or not.
It is clear, then, that the three roles interrelate in complex ways. Just as it
is hard to imagine animating without some degree of authoring, so it is equally
hard to imagine authoring without some expression of attitude, point of view,
etc., i.e. the principal role. In most normal communication, all three are
implicated. The question is the degree of involvement on different
communicative occasions.
For summaries and accounts, we might represent the continuum like this:
ANIMATOR AUTHOR PRINCIPAL

SUMMARY ACCOUNT
This diagram suggests that in order to be competent summarisers, writers

need to be able to act out mainly the roles of animator and author. And in order
to be competent account-givers, they need to be able to act out mainly the roles
of author and principal. But how might students be helped in acquiring this role
repertoire, how might they learn to be in control of the whole spectrum of

regulating their co-operation?
As for the animator role, there is a sense in which, within the limits we
have been discussing here, meaning is in the text to be discovered by identifying
textual clues. Thus developing the submissive animator role in the student is
necessary and pedagogically legitimate. One could in fact argue that such
development is logically prior: you have to first learn the conventions of the
craft in order to become artistically creative. By the same token, you first have
to be able to submissively understand a text on the writer's terms, by reference
to linguistic conventions, then you can assertively appropriate it on your own
terms. And indeed, a criticism that may be, and has been, levelled against some
contemporary, 'student-centred' approaches in foreign language teaching is that
they fail to offer learners opportunities for this necessary first step if they simply
ask them to 'be creative', or only equip them with a performance repertoire of
idiomatic phrases modelled on native-speaker contexts of use.
However, developing the co-operative, non-creative animator role is only
the first step, and it is important not to stop there - as indeed older language
teaching methods tended to do. Authorship is obviously crucial, and indeed
there is little justification for instructing students in the animator role unless it
prepares the way for more effective authoring. However, assuming that this
authoring stage comes, as we have suggested, when students have already
acquired a repertoire of animator skills, this transition should create no great
difficulties. And indeed, our respondents' reaction to the invitation to give a
personal response to the article demonstrates that they embarked on this
enterprise with enthusiasm. In effect, we were inviting them to assume the
principal role - to express their own ideas, views and values. It seems clear that
in asking for accounts in these terms, we were inevitably eliciting an authoring
of their role as principal, that is to say, the one role was implicated in the other.
They were being asked to be not only responsive but, in Schiffrin's terms,
"responsible for their talk" (Schiffrin 1994: 104).
9 Conclusion: the partiality of coherence

If one limits the scope of 'legitimate' summarisation to producing abbreviated
versions rather than brief accounts, then of course one could object that our
group B students, who gave their personal reactions, only captured part of the
meaning of the input text: by invitation, they picked up on something that was
particularly meaningful and significant to them, something they could relate to,
and gave their reactions to that while ignoring a host of other issues. What they
got out of the input text was very partial.
But our point is that meaning is partial by definition, and that in language
teaching, and especially foreign language teaching, teachers often create
bafflement by requiring students to derive more complete meaning from a text
than is actually intended to be derived from it. And this is, furthermore, totally
at variance with people's first language experience.
So when talking about coherence and appropriation of meaning we should
think not only about how we can maximise access in the sense of animating, but
also about how we can maximise learners' flexibility and room for manoeuvre,
how we can foster in our students an understanding of the limits of animating.
That is to say that our students need to understand that it is in the nature of
discourse, and of the way people achieve relevance, that meaning is intrinsically
partial and imprecise, that texts are necessarily indeterminate, and that all
readings are approximate. One of the educationally most valuable things we can
do for our students is to make them aware of, and help them cope with, this
necessary indeterminacy and insecurity. Foreign language teachers in particular
tend to be very language-fixated, or code-fixated, and to a certain degree this
has to be so, for otherwise no learning of new language would take place. But it
is too common in foreign language teaching to expect learners to 'understand
everything', and to regard this 'complete understanding' as the most desirable
outcome of any reading or listening activity. This is to say that reformulation in the
form of renderings of a source text very much tends to veer to the 'summary' side on the
left of our diagram. But it is important for learners to explore the whole spectrum, and to
include the other extreme, where they can recast any text in, and on, their own terms. In
other words, teachers would do well to appreciate the fact that coherence cannot be a
fixed, absolute objective, but is always relative to purpose, and we access meaning to the
extent that it is relevant to our purpose. So really any act of comprehension, and any
learning in general, takes place as a personal account. In this sense, then, reformulation is
a very useful activity for enabling learners to make texts coherent for themselves.
References
Chafe, W. 1992. "Discourse: an overview". In W. Bright (ed), International
Encyclopedia of Linguistics, Vol. 1. New York: Oxford University Press, 355-
358.
Goffman, E. 1981. Forms of Talk. Oxford: Blackwell.
Halliday, M.A.K. 21992. An Introduction to Functional Grammar. London: Edward
Arnold.
Schiffrin, D. 1994. Approaches to Discourse. Oxford: Blackwell.
Seidlhofer, B. 1995. Approaches to Summarization. Discourse Analysis and Language

Education. Tübingen: Narr.
Smiglis, M. 1988. 'The dilemmas of childlessness ". Time, May 2.
Widdowson, H. 1973. An Applied Linguistic Approach to Discourse Analysis. Unpubl.
Ph.D. Thesis, University of Edinburgh.
Widdowson, H. 1975. Stylistics and the Teaching of Literature. London: Longman.
Widdowson, H. 1977. "Approaches to discourse". In Widdowson, H., Explorations in
Applied Linguistics 1. Oxford: Oxford University Press, 112-140.
Widdowson, H. 1978. Teaching Language as Communication. Oxford: Oxford
University Press.
Widdowson, H. 1979. Explorations in Applied Linguistics 1. Oxford: Oxford
University Press.
Widdowson, H. 1983. Learning Purpose and Language Use. Oxford: Oxford
University Press.
Widdowson, H. 1984. Explorations in Applied Linguistics 2. Oxford: Oxford
University Press.
Widdowson, H. 1995. "Discourse analysis: a critical view". Language and Literature
4: 157-172.
Widdowson, H. 1998. 'The conditions of textual meaning". In Malmkjaer K. & J.
Williams (eds), Context in Language Learning and Language Understanding.
Cambridge: Cambridge University Press.
Coherence in Hypertext
Gerd Fritz
Justus-Lie big- Universität Gießen
1 Introduction
At first sight hypertext does not look like a good subject for research on
coherence. Hypertext is non-linear text, and coherence is typically defined for
linear text. So coherence does not seem to be involved in hypertext at all. But
on closer inspection it emerges that some of the basic structural problems with
hypertexts are classical problems of coherence.
My central question for this paper is: What does hypertext show us about
coherence? But of course the direction of enquiry could easily be reversed by
asking: What does the theory of coherence teach us about the properties of
hypertext? It is interesting to see that until recently much of the literature on
hypertext has been written by computer specialists, specialists in technical
documentation and educational psychologists (e.g. Shneiderman 1992; Kuhlen
1991; Horton 1990; Hofmann and Simon 1995; contributions to Jonassen and
Mandl 1990). Many of these authors have little contact with pragmatics and
text linguistics. As a consequence, some of the aspects which are central to a
pragmatic view of language use still seem to be under-represented in research
on hypertext. It might therefore be useful to see what the hypertext world looks
like from the vantage point of a pragmatic theory of texts. The following
remarks are intended to take a few steps in this direction.
For my theoretical framework I shall assume an action-theoretic concept
of text and dialogue (cf. Fritz 1982).1 In this framework coherence is regarded
as a guiding principle for text production and as the basis for understanding
texts. Understanding a text consists in seeing the relevant internal and external
connections of textual elements. Interpreting a text consists in searching for
and pointing out its relevant connections. According to this theory, coherence
is based on the interaction of different organising principles of texts or
dialogues. In the prototypical case, authors and readers make use of the whole
bundle of organising principles to produce a use or a reading of a text which
has strong coherence properties. Such a reading is functionally coherent,
topically coherent, it is coherent in its knowledge management, and its
222 Gerd Fritz
coherence is signalled by the appropriate use of cohesive ties. In accordance

with these different organising principles we can distinguish different aspects
of coherence and coherence failure, and we can diagnose different degrees of
coherence.
2 What is hypertext?
The difference between text and hypertext mainly concerns the following
structural and pragmatic properties: A text is a sequence of textual elements T1
... Tn (i.e. sentences or sequences of sentences) which can be used to perform a
sequence of linguistic acts, including the development of a topic. From the
point of view of its author a text can be represented as a pruned tree, as shown
in diagram (1).
(1)
At every node the author has to make up his mind which textual element
to position at this particular decision point. From the point of view of the
reader the textual elements simply form a fixed sequence.
Hypertext, on the other hand, is basically a network of textual elements,
sometimes with hierarchical structures built in. These elements are usually
called topics and the connections provided between the topics are called links.
Diagram (2) shows the structure of a hypertext consisting of the textual
elements T1 ... T8 connected by various links.
Generally, a reader may go through the network in any direction he or
she chooses. For the user hypertext is therefore not really non-linear but multi
linear. A sequence produced by travelling through such a network is called a
path. In the context of this paper it is paths that I am particularly interested in,
for a path is something like a text and therefore it can be evaluated for its
coherence properties.
Incidentally, it is revealing that the basic building blocks of a hypertext

should be called topics. This terminology lacks the sophistication that was
reached in studies on coherence some 20 years ago. It is common knowledge in
text linguistics that individual segments of text - sentences or paragraphs - can
be used not only to present a topic but that they are also used in a function. In
other words, a certain paragraph is not only about, say, matters of environment
but it also functions as a description, as a narrative or as an explanation. Recent
work in hypertexts has caught up in this respect by introducing the concept of
"typed topic". A typed topic is something like a topic with a functional tag like
"argument" or "explanation" or "example" etc. In the following I shall not use
the term topics to refer to the basic building blocks of texts. Instead, I shall use
the expression textual elements.
3 Paths in hypertext
In discussing the construction of paths through a network I shall mainly
address the following question: How do users of a hypertext make sense of the
path they are following? Generally speaking, within the chosen framework the
answer to this question is: Users make sense of a path or a segment of a path by
seeing sequences of textual elements as realisations of sequencing patterns and
by drawing inferences on the basis of their local and general knowledge. But I
should like to be somewhat more specific than that.
Basically, paths are produced in two different ways, either as pre-defined
paths, which the author presents to the user, or as self-selected paths, which the
user himself chooses from different options available at the individual nodes
within the chosen network. In a way, self-selected paths are the real raison
d'être for hypertext as an interactive medium. Here the responsibility for
making sense of paths is largely shifted to the user. The fact that the user
224 Gerd Fritz
himself chooses from alternatives reminds one of the activities of a speaker in

dialogue. Therefore the structure of dialogues obviously provides a useful
object of comparison for the interactive aspect of hypertext.
In most hypertexts for instructional purposes, like online help-systems or
teaching materials, we find a mixture of self-selected and pre-defined paths. As
a rule these systems combine search procedures for reference information, a
network of individual instruction elements and so-called guided tours or
tutorials as the following contents of "Word Help" show:
(3)
Word Help Contents
To learn how to use Help, press F1.
Pre-defined paths are closely related to normal linear text. Therefore their
conditions of coherence are quite similar to those of ordinary text. I shall give
two examples, both from the Windows 3.1 online help system, the first one
quite successful, the second one much less so.
Example 1: If you want to know how to create a table in Winword 6.0,
you go to the index of the help system and click your way through to the
following overview "creating a table".
(4)
Creating a table
®ExarnplesandDemas
Use tables to organize information and create interesting page layouts with side-by-side
columns of text and graphics. In a table, as on a spreadsheet you work with rows and
columns of calls,
To create a table
1. Position the insertion point where you want to create a table.
2. On the Standard toolbar,Clickthe Insert Table button.
Insert Table button

A grid appears below the button.
Cancal Table grid

3. Drag over the grid until you've selected the number of rows and columns you want and
then release the mouse button.
Word positions the insertion point in the first cell of the table. You can immediately type
text in the table. For more information, seeTypingordeletingtextinatable.
From this element links are provided to various other textual elements.
The links are marked by underlined expressions.2 If you choose "examples and
demos" you move to a sequence of small two-part units, in which a step-by-
step instruction is regularly followed by a demonstration of these steps. You go
from instruction to demonstration by clicking on the button "next" and from
there you go to the next instruction by clicking again on "next" and so on.
What we have here is a functional sequence of the basic type: overview
followed by instruction followed by demonstration. This is a type of functional
sequence with which we are familiar from all kinds of teaching, for example in
sports, but also from paper versions of technical documentation. So in this case
the global coherence is grounded on a fiinctional sequencing pattern and on the
continuity of topic.
At this point I should like to digress to briefly discuss an interesting
minor type of sequence which is well known in dialogue analysis. In (4) you
find the word cells underlined in the fourth line. If, as a user, you happen not to
know what a cell is you can click on the word cells and a little pop-up window
226 Gerd Fritz
will tell you: "A cell is the basic unit of a table. In a table, the intersection of a
row and a column forms one cell". If you already knew what rows and columns
are, you now know what a cell is, if not, you will have to move back to the
index and get the necessary information there. From the point of view of
coherence these small explanatory elements are very similar to footnotes or
parenthetical remarks in written text or to so-called side sequences in dialogue
(cf. Jefferson 1972). Sequences of this type interrupt the ongoing dialogue for a
clarification request, followed by a clarification, and lead directly back to the
point of departure. As a kind of question-answer sequence they are themselves
strictly coherent, and as a regular type of insertion they do not disturb the
coherence of the ongoing dialogue either - unless they occur too frequently.
A second type of pre-defined path is a sequence of related topics that can
be accessed by repeatedly clicking the "forward" button. This kind of path is of
course subject to strict conditions of topic coherence. And if anything goes
wrong there - which it easily does - the reader is justifiably upset. My second
example comes from the introduction to Windows Help. This section of
hypertext has a typical hierarchical structure which is, however, not actually
shown to the user. In order to demonstrate what happens to the unsuspecting
user I shall give a reconstruction of the respective hierarchy in the following
tree diagram.
(5)
PROCEDURES
DEFINING INSERTING MOVING

BOOKMARKS FOOTNOTES IN HELP
BROWSING MOVING
THROUGH BACK TO CHOOSING
RELATED PREVIOUS HOT-TEXT
TOPICS TOPIC
At a certain point in your path you reach the textual element

"procedures", which contains subtopics like "defining bookmarks", "inserting
footnotes in a help topic", "moving in help". Now let us assume you would like
to know more about "moving in help". You click the button "moving in help"
and this leads you to another button "browsing through related help topics".
This sounds interesting, so you move to this topic. There it says that if you
want to reach the nearest related topic you have to click on the "forward"
button. So that is what you do. And what you get is the topic "Inserting a
footnote" Unfortunately, this is not a closely related topic at all. It is far away
on the next higher rung of the hierarchy and seems to have got into the pre
defined path by mistake. As a novice user you will either think that "Inserting a
footnote" is a strange method of moving in the help system, which is a
misguided hypothesis, or you will suspect that this is a blatant case of topical
incoherence, which in fact it is. It is pleasing to know that this flaw is no longer
found in later versions of the relevant software.
4 Coherence in self-selected paths

I shall now continue with a few remarks on coherence in self-selected paths,
concentrating on two problems: 1. What difference does it make if one arrives
at a textual element from different directions? 2. How does the forward-looking
construction of coherence in search procedures work? At this point it is
necessary to mention the most powerful factor that guides the user's search for
coherence relations in the first place: "the simple fact that users expect
purposeful, important relationships between linked materials" (Landow 1991:
83).
On account of the formal properties of networks, we can arrive at a
certain textual element of a hypertext in the course of different paths. Therefore
the same textual element may play different roles in different paths. From the
point of view of the author this means that he must formulate the respective
chunk of text in a way that is neutral in respect of different directions of access.
In practice this means that he has to refrain from using anaphoric devices like
pronouns, as the anchor for a backwards-looking cross-reference may not be
available. From the point of view of the user it means that in constructing the
coherence of the respective path he may have to attribute different functions to
the textual element, depending on the preceding history of his path of reading.
How is this possible? This is, in fact, quite easy to explain, because it is only a
special case of a very general phenomenon, and a particularly interesting one.
It is a well-known fact that a portion of text can be used in different ways,
depending on the sequential position and the respective knowledge available to
the addressee at a particular point in the history of a particular communication.
In action-theoretic terms one could describe an example of this kind of
situation as follows: By describing a certain procedure to someone who is not
228 Gerd Fritz
familiar with it, one can explain it to him. Or: By describing this procedure to
someone who has read the description before, one can remind him of the
content of this description. Both patterns are very frequent in instructional
discourse. This kind of structure - explaining by describing - was called level-
generation in Goldman's "Theory of Human Action" (1970) and has been an
important element in other theories of action as well, e.g. in Heringer's
"Practical Semantics" (1978). It is obvious that level-generation plays a very
powerful role in the creation of coherent paths in hypertext. Level-generation
also works for topics. Presupposing appropriate knowledge one can talk about
X by talking about Y, e.g. one can talk about environmental problems by
talking about heating systems.
Due to the hierarchical structuring of many hypertext topics, a very
frequent difference of topic interpretation exists between a top-down or a
bottom-up interpretation. If you move down a hierarchy in a sequence A-B-C,
B will be interpreted as more specific than A, and C as more specific than B.
Alternatively, if you move up the same hierarchy, B will be interpreted as more
general than C, and A as more general than B. The following is a very simple
example which is modelled on structures we frequently find in instructional
hypertexts, e.g. in the teaching programme "Hyperlinguistics" (cf. Ansel and
Jucker 1992; Suter 1995). A short paragraph containing the main aspects of a
theory of grammar (e.g. phonology, morphology, syntax and lexicon) can be
approached from two different directions, i.e. with two different questions in
mind. If we read this paragraph as a partial answer to the question "What
aspects of language does modern linguistics deal with?", its function is to
specify these aspects. If, however, we read the paragraph as an answer to the
question "Where does syntax belong in an overall theory of grammar?", it is
embedded in a different topic. In the first case it belongs to the topic "aspects
of language", in the second case it belongs to the topic "syntax".
All this is theoretically perfectly straightforward. In practice, however,
the user must permanently monitor where he is moving in the hierarchical
structure. And at times this is a very difficult task to perform.
The fact that one can approach textual elements from different directions
also poses interesting problems for knowledge management. In writing good
linear text we try to arrange information in such a fashion that one building
block of knowledge is placed before the next. If understanding block B
presupposes knowledge from block A, we position A in front of B and so on.
Of course, basic conditions of understanding apply to hypertext in the same
way as they apply to linear text. Therefore, as users of hypertext, we must often
be prepared to compensate for the lack of systematic knowledge management.
One of the strategies for this purpose is what David Lewis (1979) called
"accommodation". If, for example, a bit of text begins with the statement "The
(6) MODERN LINGUISTICS
GRAMMAR,
SEMANTICS,
PRAGMATICS
ASPECTS OF A
THEORY OF GRAMMAR
B PHONOLOGY,
MORPHOLOGY,
SYNTAX
SYNTAX
(GRAMMAR)
three major components of a grammar are syntax, morphology and phonology

..." we are entitled to infer that there are also minor components of grammar,
which are however not dealt with in this particular paragraph. So we continue
with this assumption, and perhaps at some point we will find out if the
assumption is correct or not. Accommodation is one important type of
inference for the construction of coherence.
Now I turn to forward-looking strategies of coherence construction. If
you can navigate wherever you like in a network of textual elements, you have
to provide coherence for yourself. This concerns local coherence as well as
global coherence, including the particular problem of orientation in
"hyperspace". Problems of coherence in free navigation are particularly
interesting, because they lead us to the less prototypical forms of coherence
where reliance on explicit indicators and standard sequences is reduced.
A case in point are the strategies we use in searching for information on a
certain topic. If a user browses through a network in search of information, he
has to employ a forward-looking strategy. He has to decide which textual
230 Gerd Fritz
elements look like hopeful candidates for a useful continuation of his search.
This is somewhat like bringing your interlocutor to make a relevant
contribution to a conversation. Once the chosen chunk of text is presented on
the screen the user has to decide whether it can be incorporated into his
knowledge base as a useful contribution or not. To give an example: If you
search the Internet for information on coherence, you will come across entries
like "The role of coherence in ultrafast chemical reactions" or "coherence
modulating reactive rates" in physics. Maybe your interest in coherence
includes the term coherence and its different uses. In this case you might look
at these topics. If not, you will look somewhere else. You might go to the
homepage of a colleague and find a useful reference to her papers on
"discourse markers". From there you could move on to information on the
"Purdue University On-line Writing Lab". There you might get side-tracked a
bit - but of course you realise that you are being side-tracked - and after a while
you move back to other items on the list. Maybe you will modify your search
topic as you go along and in so doing you will modify your criteria of
coherence. And maybe you will learn new factual connections, and this may
also change your criteria of coherence. If you document your search path you
will probably be able to justify each individual move as a relevant step and
therefore you will classify the whole path as coherent.3 I realise that there are
many open questions at this point. But I shall leave it at that.
5 Conclusions
The main results of my enquiry can be summarised as follows: In hypertext we
get everything from very strong prototypical coherence in guided tours to
minimal coherence in browsing. In self-selected paths forward-looking
coherence construction plays an important role, whereas the role of classical
cohesive ties between textual elements is minimised. As for the concept of
coherence, my observations on hypertext seem to confirm the following
picture: In creating coherence we standardly draw on a whole bundle of
organising principles, but it is possible to deviate from this kind of prototype in
various ways:
(i) It is obviously possible to reduce the amount of cohesive ties like
pronouns, conjunctions and adverbs without losing too much coherence
between textual elements. This loss in explicit marking of coherence relations
is compensated by implicit factors, i.e. by the reader's knowledge of standard
sequencing patterns like functional sequencing, topical progression and so on.
Where this is not the case, lack of explicit marking will often be made up for
by means of inferences.
(ii) A second type of deviation from the prototype consists in a reduction

of the prominence of the functional aspect, which is effected by relying mainly
or exclusively on the connection of topics. In such a case we can still
understand a sequence of utterances as connected text. This kind of functional
vagueness is quite common in the internal sequencing of descriptive texts. As
many hypertexts are basically descriptive, it is not surprising that we should
frequently find this property in them.
(iii) Within the field of topic-connectedness we find a gradient from strict
topical coherence to fairly loose topical connections. In extreme cases we may
see a connection between paragraphs on the basis of a simple reference to the
same object, even if we would not accept this object as the actual topic of
either of these paragraphs. This is a very weak connection but it may be
enough for us to make sense of a sequence of textual elements. And that, of
course, is what coherence is all about.
(iv) In many cases the use of criteria of relevance and coherence is
dynamic. What we consider relevant and in which respect we consider it
relevant may change as we go along, and therefore the way in which we
interpret a sequence as coherent may also change as we go along. This is not to
say that a judgement of coherence is a matter of arbitrary decision. If we had to
justify such a judgement, we would have to explain our interpretation in terms
of the organising principles of texts and dialogues.
Notes
1. Recent developments of this framework can be found in Fritz (1991), (1994), (1997).
2. Names of links are important cohesive elements in hypertext. As opposed to most
cohesive ties in linear text, they are forward-looking.
3. For the connection between the concepts of relevance and coherence cf. Carlson (1983:
45f.) and Hintikka (1986).
References
Ansel, B. and Jucker, A.H. 1992. "Learning linguistics with computers: hypertext as a
key to linguistic networks". Literary and Linguistic Computing 7: 124-131.
Carlson, L. 1983. Dialogue Games, An Approach to Discourse Analysis. Dordrecht:
Reidel.
Fritz, G. 1982. Kohärenz. Grundfragen der Linguistischen Kommunikationsanalyse.
Tübingen: Narr.
Fritz, G. 1991. "Comprehensibility and the basic structures of dialogue". In S. Stati, E.
Weigand and F. Hundsnurscher (eds), Dialoganalyse III, Vol. 1. Tübingen:
Niemeyer, 3-24.
232 Gerd Fritz
Fritz, G. 1994. "Grundlagen der Dialogorganisation". In G. Fritz and F.

Hundsnurscher (eds), Handbuch der Dialoganalyse. Tübingen: Niemeyer, 177-
201.
Fritz, G. 1997. "Coreference in dialogue". In E. Weigand (ed), Dialogue Analysis:
Units, Relations and Strategies beyond the Sentence. Tübingen: Niemeyer, 75-
88.
Goldman, A.I. 1970. A Theory of Human Action. Englewood Cliffs: Prentice-Hall.
Heringer, H.J. 1978. Practical Semantics. The Hague: de Gruyter.
Hintikka, J. 1986. "Logic of conversation as logic of dialogue". In R.E. Grandy and R.
Warner (eds), Philosophical Grounds of Rationality. Intentions, Categories,
Ends. Oxford: Clarendon Press, 259-276.
Hofmann, M. and Simon, L. 1995. Problemlösung Hypertext. Grundlagen -
Entwicklung - Anwendung. München: Carl Hanser.
Horton, W.K. 1990. Designing and Writing Online Documentation. Help Files to
Hypertext. New York & Chichester: John Wiley & Sons.
Jefferson, G. 1972. "Side sequences". In D. Sudnow (ed), Studies in Social
Interaction. New York: The Free Press, 294-338.
Jonassen, D.H. and Mandl, H. (eds). 1990. Designing Hypermedia for Learning.
Berlin: Springer.
Kuhlen, R. 1991. Hypertext: ein Nicht-Lineares Medium zwischen Buch und
Wissensbank. Berlin: Springer.
Landow, G.P. 1991. "The rhetoric of hypermedia: some rules for authors". In P.
Delany and G.P. Landow (eds), Hypermedia and Literary Studies. Cambridge,
MA: MIT Press, 81-104.
Lewis, D. 1979. "Scorekeeping in a language game". Journal of Philosophical Logic
8: 339-359.
Shneiderman, B. 21992. Designing the User Interface. Strategies for Effective Human-
Computer Interaction. Reading: Addison-Wesley.
Suter, B.A. 1995. Hyperlinguistics. Hypertext-Lehrumgebungen im Akademischen
Kontext: eine Fallstudie. Unpubl. dissertation, University of Zürich.
Part III
How to Describe Coherence: Views of Coherence
Communicative Intentions and Coherence Relations1
Ted Sanders and Wilbert Spooren

Universiteit Utrecht and Katholieke Universiteit Brabant
1 The coherence of discourse

A fundamental trait of discourse as opposed to a random set of sentences is its
coherence.2 Despite its fondamental character, the notion of coherence has
proven to be very difficult to characterise formally. What is clear is that
coherence has to be defined in terms of the cognitive representation people have
or make of a discourse, and not so much in terms of the explicit linguistic
characteristics of the discourse itself (a stance recently defended anew in several
contributions to Gernsbacher and Givón 1995 and also in the majority of the
contributions to this volume). Even if explicit linguistic cues like anaphors and
connectives are missing, language users still have no trouble in interpreting a
discourse as a coherent whole, see the pair of sentences in (1).
(1) The winter of 1963 was very cold. Many barn owls died.
Coherence is that which makes a discourse more than the sum of the
interpretations of the individual utterances. A set of sentences is coherent if and
only if all of the segments in the discourse structure are connected to each other
(cf. Mann and Thompson 1988). Whatever the merits of this definition, it makes
clear that coherence is a notion that applies to the level of the discourse
representation.
How can we account for the coherence of discourse? This is an important
question in both linguistic and cognitive studies of discourse. At least two
different leading approaches can be distinguished in recent research on
coherence, the relational approach and the intentional one.3 In the first
approach, represented prominently by Hobbs (1979) and Mann and Thompson
(1988), coherence is modelled in terms of rhetorical or coherence relations like
Result and Concession that exist between discourse segments (the basic
elements of connected discourse, which correspond minimally to clauses).
236 Ted Sanders and Wilbert Spooren
In the second approach, the notion of discourse purpose or intention is

pivotal. Researchers try to describe discourse coherence in terms of the
intention of each segment in the discourse. Grosz and Sidner (1986, 1990) are
the strongest advocates of the intentional approach to discourse structure: "[...]
each segment is engaged in for the purpose of satisfying a particular intention"
(Grosz and Sidner 1990: 417).
The relation between the two approaches is problematic. Reductionists try
to eliminate one notion in favour of the other. Grosz and Sidner, for instance,
state that "a discourse can be understood at a basic level even if [the reader]
never does or can construct [...] such rhetorical relationships" (Grosz and
Sidner 1986: 202).
An in-between position is that advocated for instance by Moore and
Pollack (1992). They acknowledge the relevance of both coherence relations
and intentions for the coherence of discourse by claiming that more than one
relation can hold between two discourse segments simultaneously.
In this paper we argue, along with Moore and Pollack, that intentions and
relations should both be present in an adequate theory of discourse coherence.
But unlike Moore and Pollack, we also argue that this can only be achieved
adequately if the notion of intention is conceived of in a different way than it is
often defined in the intentional approach. We claim that at the level of discourse
structure there is no need for intentional notions.
2 A popular position: the multi-level thesis

In recent accounts of coherence, the so-called multi-level thesis is widely
adhered to in some version or another: Given two discourse segments there are
several coherence relations holding between the segments simultaneously, rather
than just one relation, see Bateman and Rondhuis (1997), Andriessen, De Smedt
and Zock (1996), Kroon (1995), Moore and Pollack (1992), Redeker (1990)
and many contributions to Rambow (1993). A typical multi-level claim is the
following one, where RST refers to Rhetorical Structure Theory (Mann and
Thompson 1988): "A [...] problem with RST is its assumption that between two
segments only one relation may hold. This is clearly wrong: many of them can
signal more than one relation and can do so in a single token [...]." (Andriessen
et al. 1996:261)
Moore and Pollack (1992), computational linguists working from the
point-of-view of Natural Language Generation, hold a similar view: "RST
presumes that, in general, there will be a single, preferred rhetorical relation
holding between consecutive discourse elements. In fact, [...] discourse
Coherence Relations 237
elements are related simultaneously on multiple levels." (Moore and Pollack

1992: 537)
They proceed by distinguishing between two levels of connection, the
intentional and the informational level. 'Ordinary' causal relations like (3), in
which one event causes another, should thus be taken to hold at the
informational level, whereas Evidence relations like (2) hold at the intentional
level, because one utterance (part) is intended to increase the likelihood that the
hearer will come to believe the previous or subsequent utterance (part).
(2) I am sure it was a sparrow-hawk. It had a grey back and brown stripes
on its chest.
(3) The sparrow was attacked by a sparrow-hawk. It was killed.
The crucial point in Moore and Pollack's proposal is their claim that a complete
computational model of discourse structure cannot depend upon analyses in
which the informational and intentional levels of relation are in competition. [...]
In RST, and, indeed in any viable theory of discourse structure, analyses at the
informational and the intentional level must coexist. (1992: 538)
This claim can be taken to imply that each time a rhetorical relation can be
identified between two consecutive discourse segments, there are in fact two
relations, or rather two levels at which the relation can hold, the intentional and
the informational level. Although it is not clear whether Moore and Pollack
would agree with this position (they formulate a somewhat different view
elsewhere in the very same paper)4, we want to explore its implications for
discourse structure theory. Let us assume this position to be the strong version
of the multi-level thesis. It incorporates the following two claims:
a. two relations hold simultaneously between two discourse segments (an
informational one and an intentional one)
b. the speaker's intentions help to account for the coherence of discourse.
Moore and Pollack's argument for the claim that informational relations
and intentional relations co-exist simultaneously rests on the following example.
(4)
a George Bush supports big business.
b He's sure to veto House Bill 1711.
They analyse this example as an instance of an Evidence relation (at the

intentional level). A paraphrase is "George Bush supports big business and that
is why I can claim that he's sure to veto House Bill 1711". At the same time, it
is an instance of a Volitional Cause relation (at the informational level). In this

reading it can be paraphrased as "The fact that George Bush is sure to veto
House Bill 1711 is caused by the fact that he supports big business".
3 Why the (strong) multi-level thesis is wrong

Our first argument against the strong version of the multi-level thesis is that it is
not the case that all relations can be interpreted at both an informational and an
intentional level. There are clear cases of exclusively informational and
exclusively intentional relations.5 To our knowledge no one has yet proposed an
intentional relation to hold between the two clauses in (5):
(5)
De weg was geblokkeerd doordat er een lawine
The road was blocked because there an avalanche
was geweest op Roger's Pass.
was been on Roger's Pass.
'The road was blocked as a result of there being an avalanche at Roger's

Pass.'
Second, the notion of intention is problematic and it fails to account for

the coherence of discourse. 'Intentions' are frequently used in the literature, but
hardly ever defined. Those definitions that do exist are often too abstract to be
used in the analysis of discourse structure (cf. Bratman 1990; for a notable
exception, which is not referred to in the multi-level debate, see Searle 1983,
especially chapter 6).
The crucial question is: Can intentions account for the coherence of
discourse? Grosz and Sidner (1990) typically present examples of intentions like
the following, in which two participants develop a shared plan to lift a piano (p.
432):
(6)
a S1: I want to lift the piano.
b S2: ok.
c I will pick up this end.
One of the things Grosz and Sidner argue is that both participants infer
something like (7) from segment (6)c:
(7)
MB(S1,S2,INT(Si,lift(foot-end)))
(where MB=Mutual Belief, INT=Intention)
But that does not tell us how such an intention can account for the
coherence of the discourse representation of the fragment, i.e. in what way the
segments are connected to each other.
Should it come as a surprise that it is so difficult to align the concepts
'intention' and 'coherence relation'? We believe the answer is no, because
intentions and coherence relations have a very different ontological status.
This is our third argument against the strong multi-level thesis: intentions
and relations are different types of entities (see Asher and Lascarides 1994 for
additional reasons)6. Intentions are basically unary in that they are functions, the
arguments of which are the speaker, the hearer and a single proposition. By
contrast, coherence relations are minimally binary, in that they connect at least
two propositions. A typical discourse intention is "the speaker wants to
persuade the hearer that a certain state of affairs is the case/worth persuing/..."
(WANT(S,(BELIEVE(H,P)))). This is ontologically very different from
connecting two states of affairs or events in, for instance, a Cause-relation
CAUSE(P,Q).
4 Toward a cognitive account of relations and intentions

We want to give a blueprint of an alternative for the multi-level thesis. This
cognitive account consists of two basic concepts: coherence relations and
communicative intentions. The distinguishing trait of this alternative is that these
two concepts are separated.
4.1 An organised set of coherence relations
The approach advocated here is typically relational in the sense that the
coherence of discourse is accounted for in terms of coherence relations that
exist between segments. These relations are conceptual by nature because they
are not part of the discourse, but of the representation language users have or
make of a discourse. A number of theories have made use of relations in
explaining coherence (Hobbs 1990b; Longacre 1983; Mann and Thompson
1986, 1988; Martin 1992; Meyer 1975). Yet, there is no consensus about a
single set of relations (see, for instance, Hovy 1990), and the alternative sets
that have been put forward are very different. Furthermore, many sets of
relations are presented as plain lists, unorganised and extendable ad infinitum.
This is unsatisfactory for several reasons. The first and most important
one is that in this situation it is difficult to account for the way in which a
coherent cognitive representation can be constructed. For instance, how does a
reader arrive at the interpretation of a particular coherence relation such as
Evidence? If all relations are considered to be on the same basic level, it must be
assumed that readers use their knowledge of all these relations (30, 100, 1000?),
in order to interpret a stretch of discourse; it must further be assumed that
Evidence, and Solutionhood, and Frustrated Expectation as well as many other
relations which scholars have identified, but about which they do not agree, are
all cognitively basic. A second reason to feel discomfort about such lists is that
students of coherence relations seem to agree about some basic notions, which
occur in every proposal. Examples are the causal and contrastive nature of
relations. A third point to consider is that the way in which coherence relations
are expressed linguistically reflects the relationship that exists between the
different coherence relations. There are restrictions on the type of relation a
connective can express. For example, and can express additive and causal
relations but not concessive relations. And however can express contrastive and
concessive relations, but not causal relations (see, for instance, Degand 1996;
Knott and Dale 1994; Knott and Sanders 1998 and Pander Maat 1998 for some
systematic studies). These restrictions imply an organisation of the relations that
connectives can express, and an adequate theory of discourse structure should
be able to express these restrictions.
Hence, an adequate theory of discourse structure will have to explain the
fact that the similarity between coherence relations varies. To that end, the set
of relations should be categorised, and it can then be regarded as a list which is
in principle finite.
In Sanders, Spooren and Noordman (1992, 1993) we have proposed a
classification scheme which accounts for the 'relations among relations'. The set
of coherence relations is described in terms of four primitives, shared by all
relations: Basic operation (relations are causal or additive), Polarity (positive or
negative), Source of coherence (semantic or pragmatic) and Order of segments
(basic or non-basic order). In these terms, the examples (1) and (3) would be
categorised as causal, positive, semantic, basic order {Cause-Consequence) and
sequence (2) would end up as causal, positive, pragmatic and non-basic order
{Claim-Argument).
The four primitives are important cognitive categories, prominent in
research on language and language behaviour. Predictions based on this
relational classification theory have been tested in several experiments. In
Sanders et al. (1992) it was shown that the coherence relations that are similar
in terms of these primitives were confused more often than relations that are less
similar. In another experiment (Sanders et al. 1993) subjects were explicitly

asked to make direct comparisons between different relations. In general, the
similarity judgements conformed to the categorising principles.
One of the primitives is particularly important with respect to the multi
level discussion: the source of coherence; compare (8) and (9).
(8) Theo was exhausted because he had to run to the university.

(9) Theo was exhausted because I saw him gasping for breath.
In (8) but not in (9) the second clause can be taken to be a cause for the
effect in the first clause. This Cause relation is a relation between states of
affairs. In (9) there is not such a relation between states of affairs. Here the
second clause gives a justification for uttering the first clause. Following the
terminology of van Dijk (1977) we use the terms semantic and pragmatic
relations to describe the difference. In view of the rich and laden history of these
phrases it is perhaps wiser to use more neutral terminology: 'propositional' for
relations between states of affairs and 'illocutionary' for relations between
speech acts. 7 All relations that are called intentional in the literature, for
instance by Moore and Pollack (1992), are of the illocutionary type, whereas
the informational ones are propositional.
The relevance of a distinction between propositional and illocutionary
relations comes from the fact that several researchers have suggested that
connectives exist across languages, which can be considered 'domain-specific',
i.e. they can only be used to express illocutionary relations or only to express
propositional relations. This has been suggested for Japanese kara versus node
(Takahara 1990), for French car versus puisque (Bentolila 1986) and for
German denn versus weil (e.g. Günthner 1993; Keller 1995); Németh (1995)
suggests that Hungarian hat and mert are pragmatic connectives, Sweetser
(1990) suggests that English since prefers non-propositional relations, and
Knott (1996) and Knott and Sanders (1998) show that in both English and
Dutch domain-specific cue phrases exist.8
In sum, we agree with Moore and Pollack (1992) that coherence relations
come in two sorts, but we also think that this property of relations should not be
confused with the entirely different notion of intention.
4.2 Communicative intentions
In multi-level accounts of coherence, the speaker's and hearer's intention is

placed at the level of coherence relations, therefore the term 'intentional
relations'. We have already argued why we think this wrong. Nevertheless, it is
obvious that any account of discourse must deal somehow with communicative
intentions and their link to discourse structure. The question is how to account
for this connection.
From a cognitive point of view it seems very well possible to combine the
two concepts, for instance on the basis of one of the leading theories of
discourse production, namely Levelt's (1989) model of speech production. In
this model there is a planning component, the Conceptualiser, in which
communicative intentions are mapped on speech acts, which in turn are mapped
on the preverbal message. The latter is input to the subsequent components of
the model, such as the Formulator, in which lexical items are selected and
syntactic structures are built. In this model the preverbal message is the
(preverbal) realisation of the communicative intentions, and we suggest that
coherence relations should be located exactly there: They are a means of
realising complex communicative intentions.
5 The Bush example: propositional and illocutionary relations

in context
If this cognitive account is an alternative to the multi-level thesis, how can the
Bush-example be analysed in these terms? We agree with Moore and Pollack's
(1992) claim that there is more than one possible relation that can be identified
to connect fragment (4). But we disagree with the stance (implied by the strong
interpretation of the multi-level thesis) that it is a systematic property of
connected discourse that whenever there is a relation at the informational level
there is also a relation at the intentional level, and vice versa. In fact, we think
that example (4) is simply a case of vagueness: the example is unspecific
concerning the intended relation. If one adds contextual information, then the
vagueness disappears. And then it turns out that much more interpretations of
one and the same stretch of discourse are available. Below we have listed a set
of other plausible interpretations for this fragment.
(10) List
a. George Bush is the 53rd president of the US.
b. He's from Texas.
c. He is bound to sign the economical treaty with Canada this month.
d. He supports big business.
e. He's sure to veto House Bill 1711.
f. He has a keen interest in foreign politics.
g. And he is the president of the Washington Rotary club.
PLANNING COMPONENT
Communicative intentions
DISCOURSE STRUCTURE
a. Semantic information
b. Coherence relations
(i) 'propositional'
(ii) 'illocutionary'
DISCOURSE
Figure 1: A schematic representation of the relationship between

communicative intentions and coherence relations.
(11) Claim-Argument
a. Last week's Newsweek contained an article
b. revealing that behind the screen George Bush is a fervent supporter of
the Texan oil billionaires.
c. This has been suggested previously,
d. but this is the first time that a reporter gave a conclusive argument for
that claim.
e. George Bush supports big business,
f He's sure to veto House Bill 1711.
(12) Enumeration
a. Republican Presidents have always been strongly opposed to the
environmental lobby:
b. George Bush supports big business.
c. He's sure to veto House Bill 1711.
d. On top of that, he has always refused to talk to activist groups like
Greenpeace.
(13) Contrast
a. You never know what to think of this guy Bush.
b. He supports big business.
c. He's sure to veto House Bill 1711.
d. So his politics are totally inconsistent.
(14) Contrastive Argument-Claim

a. Usually presidents who support the country's business giants are not
very favourable towards environmental laws.
b. There are exceptions though.
c. George Bush supports big business.
d. He's sure to veto House Bill 1711.
The multi-level thesis originates in the observation that in, for instance, a
Claim-Argument relation (as in (11)), the argumentation relies on certain states
of affairs in the world. For instance in (11), House Bill 1711 must have a
character that is detrimental to big business. But we think that it is misleading to
say that in (11) there is a Cause relation between e and f, because the link
between the two segments is argumentative and not causal. Evidence for this
claim is that the Claim-Argument relation can co-vary with a whole range of
states of affairs in the world: time relations (John's the murderer, because
directly after the murder I saw him at the scene of the crime), categorisation
relations (It is a singing bird, because it is a robin), listing (or even contrast)
relations {It must have been my mother who gave me the tennis balls for my
birthday, because my father gave me the tennis racket), something that one
might want to call source relations (John isn 't coming, because he told me so),
and 'hypothetical world' relationships (You must borrow me ten dollars,
because otherwise I'm not speaking to you anymore).
More generally, the fact that out of context relations can be vague does
not imply that there is always more than one relation, let alone that when there
is a relation at the informational (propositional) level there is also a relation at
the intentional (illocutionary) level.
The conclusion is that for example (4) there are many more possible
interpretations than just the two identified by Moore and Pollack. Furthermore,
the examples (10)-(14) show that, within context, it is perfectly clear which
relation connects the segments: they can be classified in terms of the primitives
given in Sanders et al. (1992). The relations in examples (10) and (13) exist only
at an ideational/propositional level (as in Moore and Pollack's original
Volitional Cause reading), and in examples (11), (12) and (14) at an
intentional/illocutionary level10 (as in Moore and Pollack's original Evidence
reading).
6 Conclusion: intentions and relations in a cognitive theory of

coherence
So far we have argued that intentions and coherence relations are different
concepts. Therefore it is not surprising to find that communicative intentions
and coherence relations frequently diverge. Many texts have a communicative
intention that does not run parallel with the discourse structure. For instance,
there is a well-known Dutch hortatory text (a non-commercial advertisement)
that tries to warn the reader of the dangers of firecrackers. For persuasive
reasons (the intended audience being Dutch adolescents), the authors have
chosen an ironic tone. The text, which is accompanied by a picture of a blind
boy accompanied by a guide dog, reads (with a liberal English translation):
(15)
Dankzij dat veel te kleine lontje heb ik nu eindelijk een hondje.
'Thanks to that fuse that was much too short, I finally have a doggy for my
support.'
Whatever analysis one makes of this example, it seems very difficult to

find a coherence relation that is in accordance with the optimistic tone of the
connective thanks to and the adverb finally and that at the same time captures
the advisory nature of the authors' intention. To us it seems obvious that at the
level of discourse coherence, the authors express a Volitional Result relation,
whereas at the level of discourse intentions they try to warn the readers. Hence,
although it is clearly possible to detect an intentional message in the
utterance - for instance, the linguistic form strongly signals 'gratitude' - the
connectedness of the discourse cannot be characterised in terms of its intention.
The reason is that there is a clear mismatch between the linking of the clauses
and the communicative message that the authors try to convey.
There is a more general point to be made here: does connected discourse
cohere because the author wants to convince the reader or does it cohere
because the author gives evidence for a claim? We think the latter is the case.
To convince someone you do not even need words, you could just as well point
a gun at him. And if you use words, why would you need more than a single
proposition? In other words, the level of intentions is not suited to account for
the coherence of a discourse.
In short, our proposal is to keep communicative intentions and coherence
relations apart and to look upon coherence relations as realisations of intentions.
Such a cognitive account predicts close relationships between intentions and
relations: When a writer intends to inform the reader about some event in the
world, (s)he will typically choose propositional relations and when the intention
is persuasive, a writer will often use illocutionary relations. Such a prediction is
corroborated by the finding that there are strong correlations between
coherence relation type and discourse type (cf. text type in Virtanen 1992). In a
corpus of argumentative and descriptive texts it was found that relatively more
illocutionary relations occur in argumentative texts than in descriptive texts.
This relative predominance of illocutionary relations also holds in terms of the
amount of relations at the highest text levels (Sanders 1997). And in an
experimental study on the acquisition of connectives and relations we found that
the coherence relations between utterances produced by primary school children
were strongly determined by the discourse task in which the children were
involved: a picture description task resulted mainly in propositional relations;
when asked for an opinion the children mainly produced illocutionary relations
(Spooren, Tates and Sanders 1996).
The picture that emerges is the following: intentions and discourse
coherence are to be separated; coherence relations are realisations of
communicative intentions. Some intentions are realised in a preverbal message
without any coherence relation, others are realised in a preverbal message
containing an illocutionary relation, and still others are realised as a
propositional relation. For instance, if I were to have the communicative
intention to convince you that George Bush supports big business I might try to
realise this intention by stating that George Bush supports big business. But I
may also believe that merely uttering this will not do the job. In that case I may
add an argument to my utterance, namely that Bush is sure to veto House Bill
1711 (which is the analysis of example (11); in this reading the bill is detrimental
to big business). Alternatively, I may have the communicative intention to
convince you that Bush's politics are totally inconsistent, and again, I also
believe that merely uttering this will not convince you and therefore I present as
an argument the contrast between Bush's support of big business and his
upcoming veto on House Bill 1711 (example (13); in this reading the bill is
favourable toward big business). Finally, I might want to describe to you the
main characteristics of Bush's political life among which his support of big
business and his veto on House Bill 1711 (example (10)).
Such a view is 'multi-level' in the sense that speakers and hearers keep
track of their (alleged) intentions and the discourse structure simultaneously.
But it does not imply a discourse structure which is multi-level in itself, in that it
consists of an intentional discourse structure separated from and coexisting with
an informational discourse structure.
Notes
1. Authorship of this paper is shared equally. Earlier versions of this paper were presented
at the 4th Conference of the International Cognitive Linguistics Association in
Albuquerque, NM, July 1995 and at an International Workshop on Discourse Markers in
Duisburg, Germany, October 1995. We would like to thank several people for comments
on an earlier draft, especially Ed Hovy and Alistair Knott. Needless to say that all
remaining errors are ours.
2. Strictly speaking, the coherence is not a property of the discourse but of the
representation language users make of the discourse.
3. We leave aside approaches dealing with referential coherence, i.e. repeated reference to
the same set(s) of entities. See Sanders, Spooren, and Noordman (1992), section 1.1 for a
discussion and relevant literature.
4. "We are not claiming that interpretation always depends on the recognition of relations
at both levels, but rather that there are obvious cases where it does. An interpretation
system therefore needs the capability of maintaining both levels of relation" (Moore and
Pollack 1992: 540).
5. Empirical evidence for this claim comes from the finding that when exclusive cases are
presented to judges, they intuitively agree with the classification of items in terms of
'informational' or 'intentional' (or, in the terminology to be introduced later,
propositional and illocutionary, respectively), independently of the context in which the
items are presented (Sanders et. al. 1993; Sanders 1997).
6. Asher and Lascarides (1994) outline a formalisation of the relation between intentional
structure and discourse structure using a system of non-monotonic reasoning and argue
for a distinction among the two on different grounds, namely that the rules governing the
two types of structures behave differently. For instance, intentional structure allows for
abductive reasoning, and discourse structure does not.
7. We are aware of the fact that each segment has a locutionary and an illocutionary
meaning (Pander Maat 1994), but that does not imply that the link between discourse
segments is to be located at the locutionary and the illocutionary level. Coherence
relations like Cause connect segments at a propositional level, whereas Justification and
Evidence relations connect segments at an illocutionary level.
8. At the same time, corpus-analytic research shows that the distribution of many common
connectives, for instance Dutch causal conjunctions, cannot be explained in terms of
these 'domains' only, see for instance Pander Maat and Sanders (1995), Pit, Pander Maat
and Sanders (1997).
9. The model suggests that language production proceeds serially. This is only partially
correct, in that the model works incrementally. as soon as a minimal unit is composed by
one component, it is passed on to the next component. In this way the various
components can work in parallel, be it on different parts of the message.
10. This is confirmed in experiments in which language users are asked to judge relations
(Sanders 1997).
References
Andriessen, J., de Smedt, K. and Zock, M. 1996. "Discourse planning: empirical
research and computer models". In T. Dijkstra and K. de Smedt (eds),
Computational Psycholinguistics. London: Taylor and Francis, 247-278.
Asher, N. and Lascarides, A. 1994. "Intentions and information in discourse".
Proceedings of the 32nd Annual Meeting of the Association of Computational
Linguistics June, 27-30, 1994. New Mexico.
Bateman, J.A. and Rondhuis, K.J. 1997. "'Coherence relations': towards a general
specification". Discourse Processes 24: 3-49.
Bentolila, F. 1986. "CAR en français écrit". La Linguistique 22: 95-115.
Bratman, M.E. 1990. "What is intention?" In P.R. Cohen, J. Morgan and M.E. Pollack
(eds), Intentions in Communication. Cambridge, MA: MIT Press, 15-32.
Degand, L. 1996. A Situation-Based Approach to Causation in Dutch with some
Implications for Text Generation. Dissertation, Université Catholique de
Lou vain, Belgium.
Dijk, T. van. 1977. Text and Context. Explorations in the Semantics and Pragmatics
of Discourse. London: Longman.
Gernsbacher, M. and Givón, T. (eds). 1995. Coherence in Spontaneous Text.
Grosz, B.J. and Sidner, C.L. 1986. "Attention, intentions and the structure of
discourse". Computational Linguistics 12: 175-204.
Grosz, B.J. and Sidner, C.L. 1990. "Plans for discourse". In PR. Cohen, J. Morgan
and M.E. Pollack (eds), Intentions in Communication. Cambridge, MA: MIT
Press, 417-444.
Guenthner, S. 1993. "'...weil - man kann es ja wissenschaftlich untersuchen' -
Diskurspragmatische Aspekte der Wortstellung in WEIL-Sätzen". Linguistische
Berichte 143: 37-59.
Hobbs, J. 1979. "Coherence and coreference". Cognitive Science 3: 67-90.
Hobbs, J. 1990. Literature and Cognition. Menlo Park: Center for the Study of
Language and Information.
Hovy, E. 1990. "Parsimonious and profligate approaches to the question of discourse
structure relations". Proceedings of the 5th International Workshop on Natural
Language Generation: 128-136.
Keller, R. 1995. "Das epistemische WEIL". In R. Keller (ed), Zeichentheorie.
Tübingen: Francke, 239-252.
Knott, A. 1995. A Data-Driven Methodology for Motivating a Set of Coherence
Relations. Ph.D. Thesis, University of Edinburgh.
Knott, A. and Dale, R. 1994. "Using linguistic phenomena to motivate a set of
coherence relations". Discourse Processes 18: 35-62.
Knott, A. and Sanders, T. 1998. "The classification of coherence relations and their
linguistic markers: an exploration of two languages". Journal of Pragmatics 30:
135-175.
Kroon, C.H.M. 1995. Discourse Particles in Latin: A Study of NAM, ENIM, AUTEM,
VERO and AT. Amsterdam: J.C. Gieben.
Levelt, W.J.M. 1989. Speaking. From Intention to Articulation. Cambridge, MA: MIT
Press.
Longacre, R.E. 1983. The Grammar of Discourse. New York: Plenum Press.
Mann, W. and Thompson, S.A. 1986. "Relational propositions in discourse".
Mann, W. and Thompson, S.A. 1988. "Rhetorical structure theory: toward a functional
theory of text organization". Text 8: 243-281.
Martin, J.R. 1992. English Text: Systems and Structure. Amsterdam: Benjamins.
Meyer, B.J.F. 1975. The Organization of Prose and its Effects on Memory.
Amsterdam: North-Holland.
Moore, J.D. and Pollack, M.E. 1992. "A problem for RST: The need for multi-level
discourse analysis". Computational Linguistics 18: 537-544.
Németh T., E. 1995. "On the role of pragmatic connectives in Hungarian spoken
discourse". In B. Wårvik, S.-K. Tanskanen and R. Hiltungen (eds), Organization
in Discourse. Proceedings from the Turku Conference. Turku: University of
Turku, 393-402.
Pander Maat, H. 1994. "Coherentie-relaties hebben zowel semantische als pragmatische
aspecten". [Coherence relations have both semantic and pragmatic aspects]. In A.
Maes, P. van Hauwermeiren and L. van Waes (eds), Perspectieven in
Taalbeheersings onderzoek. Dordrecht: ICG, 118-129,
Pander Maat, H. 1998. "Classifying negative coherence relations on the basis of

linguistic evidence". Journal of Pragmatics 30: 177-204.
Pander Maat, H. and Sanders, T. 1995. "Nederlandse causale connectieven en het
onderscheid tussen inhoudelijke en epistemische coherentie-relaties". [Dutch
causal connectives and the distinction between content and epistemic coherence
relations]. Leuvense Bijdragen 84: 349-374.
Pit, M., Pander Maat, H. and Sanders, T. 1997. "Doordat, omdat en want.
Perspectieven op hun gebruik". [Doordat, omdat and want. Perspectives on their
use]. Taalbeheersing 19: 238-251.
Rambow, O. (ed). 1993. Intentionality and Structure in Discourse Relations.
Proceedings of the ACL-Workshop, Ohio State University, Columbus, Ohio.
Redeker, G. 1990. "Ideational and pragmatic markers of discourse structure". Journal
of Pragmatics 14: 367-381.
Sanders, T. 1997. "Semantic and pragmatic sources of coherence: on the categorization
of coherence relations in context". Discourse Processes 24: 117-145.
Sanders, T., Spooren, W. and Noordman, L. 1992. "Toward a taxonomy of coherence
Sanders, T., Spooren, W. and Noordman, L. 1993. "Coherence relations in a cognitive
theory of discourse representation". Cognitive Linguistics 4: 93-133.
Searle, J.R. 1983. Intentionality. An Essay in the Philosophy of Mind. Cambridge:
Cambridge University Press.
Spooren, W., Tates, T. and Sanders, T. 1995. "Coherentierelaties en connectieven bij
basisschoolkinderen". [Coherence relations and connectives as used by primary
school children.]. Nederlandse Taalkunde l: 26-51.
Sweetser, E.V. 1990. From Etymology to Pragmatics. Metaphorical and Cultural
Aspects of Semantic Structure. Cambridge: Cambridge University Press.
Takahara, P.O. 1990. "Semantic and pragmatic connectives in English and Japanese".
Paper presented at the International Pragmatics Association Conference,
Barcelona, July 1990.
Virtanen, T. 1992. "Issues of text typology: Narrative - a 'basic' type of text?" Text 12:
293-310.
If Coherence is Achieved, Then Where Doth Meaning
Lie?
Willis J. Edmondson
Universität Hamburg
I wish to focus on the scientific status of the results of discourse analysis. My

concern is validation - when a given analysis is offered in print in one of the
scientific journals devoted to such issues, how can it and divergent analyses of
the same data be discussed, compared, validated? Is discourse analysis itself
neutral regarding the analytic results it enables? Or are there criteria which we
can use in evaluating the appropriateness of a specific analysis? In order to
address such basic issues, it is particularly necessary that one clarifies one's
terms. This then will be my first concern.
1 Discourse, coherence, and the mind

There are almost as many different definitions of discourse analysis as there are
professionals in various fields willing to offer one1. The following attempt at
delineation seeks to be uncontroversial.
As I understand it, a discourse is a social event encoded - at least in part
- in language: a football-match is probably not to be considered a discourse, a
broadcast commentary on that football match is. These two events may be co-
present in real time, if the match is viewed on a television-screen, and this
transmission is accompanied by the commentary. More commonly, of course, a
discourse does not accompany or provide a commentary on a social event, it
constitutes a social event itself, and involves therefore interaction, most
obviously between at least two humans, who create discourse out of the stuff of
which sentences and texts are made. The relationship between sentences and
utterances is paralleled by the relationship between texts and discourses.
The distinction between text and discourse is however not always
transparent. For example, the notion of transcribing a discourse is an
interesting one. Is the result a 'text' or a 'discourse'? While a transcription is
itself more 'text' than 'discourse', we can, on the basis of our transcription, seek
to re-create the discourse ourselves. It follows, then, that the distinction
252 Willis J. Edmondson
between discourse and text - like the distinction between utterance and
sentence - is not one of substance, but one of use. How far in reading a
transcription (or, indeed, viewing a video-recording of a discourse event), we
are in fact creating a discourse distinct from the one that was transcribed and/or
recorded, is, in a sense, precisely the question I wish to raise in this paper.
Let me now briefly gloss the term coherence. To refer to the coherence of
a discourse is to refer to the ways in which its parts constitute a whole. We are
talking about whether and how a discourse makes sense, and this making-sense
is constituted by the behaviours realised or represented in the words, sentences
and texts used in the discourse, and not by these linguistic tokens themselves.
By contrast, the way or ways in which elements in a sentence or in a text fit
together is a matter of grammar and of cohesion. Now if a discourse is a social
event, and if, as is surely the case, a social event only makes sense for members
of the society in which it has a purpose, then it follows that coherence is a
matter of interpretability. The coherence of an instance of language in use is
therefore a reflection of the coherence-making skills of the participants and/or
the researcher. A discourse cannot therefore be said to be coherent without the
mediation of a human mind.
If these rather commonplace views on the nature of discourse and of
coherence are accepted, what, then, can we say of the term discourse analysis
itself? I shall assume that discourse analysis is a principled undertaking,
aspiring to scientific status, and centrally concerned with explicating the nature
of the coherence residing in a discourse. Other, additional goals might also
hold, but they will be premised on the discernment of coherence. If, I wish to
claim, you are not interested in asking how the participants "make sense" of the
piece of language you are handling, then you are not doing discourse analysis,
on my understanding, but some other thing.
Returning to the question raised above concerning the status of the
analyst's interpretation of a transcription or of a video-recording, where, we
may now ask, does then the discourse analyst fit in, in terms of the coherence-
creating capacities of the human mind? Does he or she establish coherence
additional to, or independently of the discourse participants themselves? What
validation criteria hold for a particular analysis of a given piece of data? If
coherence is achieved, as my title rather quaintly puts it, then where doth
meaning lie?
Coherence and Meaning 253
2 Some features of discourse in the light of this view of

coherence
Before I attempt an answer, let me underscore the problem by pointing out
some consequences which follow from the view of coherence as an
achievement of a human mind, and which put the question of analytic
validation (where doth meaning lie?) into sharp relief. At the risk of
belabouring the obvious, let me briefly mention six features of discourse which
seem to raise the problem of 'objective' and verifiable interpretative procedures.
Firstly, a discourse can be interpreted in many ways. Edmondson (1990)
refers to discourse behaviours as infinitely-interpretable. This is doubtless an
exaggeration, but the observation that multi-interpretations are possible for a
given discourse segment is so banal that I refrain from elaborating upon it.
Secondly, it is relevant to observe that discourse meanings vary over
time. This is true in two related senses. Firstly, persons living at widely
separated times may view a specific discourse differently. This is why some
people can still make a living writing books about the plays of Shakespeare.
Secondly, individuals can react differently to one and the same discourse over
a time interval. To put this another way, humans develop discourse
interpretative skills over time - it's called education. In this sense, then,
coherence not only requires the intervention of a human mind. The nature of
the coherence arrived at, derived, discovered, or created will be determined -
at least in part - by which mind was operating at which point in time.
A third point follows. If the mind constructs discourse out of text, then it
follows that what is constructed is in part a function of what is stocked in the
construing mind. In other words, expectations play a role in discourse
processing. This is the ground for pragmatic normalisation.
Fourthly, in any ongoing interaction, a hearer-knows-best principle
operates. This means that in conversation, for example, the significance of the
speaker's concluded turn-at-talk is determined at that point in time by its
reception by the hearer. So there is no difference, for example, between your
being irritated, and my irritating you. In the business of discourse,
perlocutionary effect, as it were overrules illocutionary intent (on the hearer-
knows-best principle, cf. for example Edmondson 1989a).
It follows, fifthly, that discourse meaning is negotiable. This is a
consequence of the hearer-knows-best principle, as turns are taken in ongoing
talk, and speaker/hearer roles alternate in two-party discourse. This process of
negotiation also occurs in written discourse, but the process is more complex,
and can assume different forms.
Finally, the distinction between coherent and non-coherent discourse is
not clear-cut. This follows from the multiple-interpretation hypothesis, and
from the hearer-knows-best principle. In a nutshell, one woman's coherence is

another man's chaos. In other words, the issue of coherence is itself negotiable.
The issue has been discussed in the literature from two perspectives. The first
is the case in which human agents produce discourse segments which,
linguistically, offer interpretive difficulties. Schizophrenic discourse or the
language of persons suffering from various kinds of aphasia falls into this
category (cf. Roberts and Kreuz 1993). Sometimes the speech of non-native-
speakers is also treated with suspicion. Secondly, the [+/- coherent] issue has
been raised in the opposite context - i.e. when segments of discourse are
produced which are, to all intents and purposes, interpretable (given good will),
but for which some normally presupposed intentional stance contained in the
term speaker-meaning2 seems to be absent. In this category, we have dialogic
sequences resulting from interactions between computer programs and naive
humans, or the Washoe-syndrome, i.e. the creative construction processes
evidenced by researchers convinced that apes can be taught to talk. I know in
fact of no interesting attempts to specify what criteria might determine how
coherent and non-coherent discourse can be sharply distinguished, and
conclude that such attempts are misguided (but cf. Bublitz and Lenk, this
volume). Coherence is not an either-or issue, nor is it a more-or-less issue: it is
a relational, and not a quantitative category.
3 The status of discourse analysis

The six features listed above do not tell the whole story, however. There are
limits to how a given discourse segment can be read. We do not normally enjoy
the freedom to let things mean what we want them to mean. While negotiation
is the essence of discourse participation, it is, of course, not the case that
everything has to be negotiated anew each time two social members meet.
Negotiation is only possible precisely because of the fixed points of reference
provided by rituals, conventions, situational constraints and what are to all
intents and purposes non-negotiable linguistic elements. The role of routines in
everyday talk confirms the point here (cf. for example Edmondson 1989b).
However, that having been said, the six points given above are
characteristic of human communication. The central issue they raise is that of
normalisation. What counts as 'normal', or 'standard' in discourse interpretation,
and how do we know? If fuzziness, negotiation, multiple meanings, and so on
characterise discourse in social life, what status has the analysis produced by a
professional calling himself or herself a discourse analyst?
The answer proposed here is neither new nor sophisticated. It simply says
that scientific rigour is achieved, if an analysis derives from an explicit theory
of discourse. There are, however, alternatives to this view, which seem to me
to be misguided, as they seek to bypass the issue of validation, and in doing so

redefine the notion of discourse analysis. I shall mention two such current
views.
One radical answer to the question of discourse interpretation is to say
that all interpretations are equally valid (or equally invalid, which comes to the
same thing). This view constitutes the ultimate relativistic stance. It is logically
consistent with the nature of discourse and nature of coherence, as
characterised above. Further, it is of course true that the social status of the
analyst - as academic linguist, or of native-speaker, for example - is not
acceptable as a validation criterion.
The ultimate relativistic stance can, moreover, invoke a metatheoretical
corollary, associated with for example the name of Foucault. This metatheory
claims, roughly, that 'anything goes', in the sense that there is no objective
ground for preferring one discourse interpretation over another. Discourse
manifests the ideological framework inside which it occurs. As there are no
acultural or apolitical norms holding for discourse interpretation, there is no
neutral ground for the validation of discourse meaning. On this argument, then,
the proper subject of scientific enquiry is the social, psychological and
ideological factors that determine the discourse behaviours that occur.
This is as far as I can see a perfectly valid logical extension of the
anything-goes position. However, it undermines discourse analysis in the sense
I wish to use that term. An overriding concern with discourse itself loses any
scientific ground it might have, as all discourse analysis is ideologically and/or
culturally determined: we are all apparently indulging political posturing or
cultural closeting, in aspiring to concern ourselves with discourse analysis.
For my purposes here, we can accept that this might well be true, but it is
only part of the truth about human behaviour, just as the six points I made
above about coherence constituted only part of the truth about discourse. It
seems to me that the whole thrust of the so-called 'soft' sciences is to take us
beyond the notion of cultural/ideological determinacy. However, it is not my
purpose to evaluate such relativistic discourse philosophies in absolute terms: it
is sufficient to point out that a relativistic stance is perfectly acceptable
philosophically, but by its very nature cannot be valid as a theory of discourse
analysis. The problem is that all too often the anything-goes syndrome and the
discourse-as-ideology stance are accommodated under the umbrella term
'discourse analysis', thereby robbing that term of substance. The irony is, of
course, that the discourse-as-ideology position is inherently flawed as a
scientific theory, precisely because its validity disappears when its argument is
applied reflexively.
Such relativistic views, which essentially sidestep the notion of
validation, seem nonetheless to be both attractive and widespread. They are
implicitly or explicitly present in areas such as Critical Discourse Analysis (cf.

Fairclough 1992, and the discussion initiated by Widdowson 1995, and carried
further in Widdowson 1996), Intercultural German Studies as established in
Germany ("die interkulturelle Germanistik"), which seek to combine the
anything-goes syndrome with an authoritarian stance (cf. Wierlacher 1992;
Zimmermann 1989), "neo-hermeneutic" approaches to translation and
translation quality assessment (cf. e.g. Stolze 1992), issues of textual
canonisation inside university literary departments - is it acceptable for
example to obtain an MA in English Studies, after introductory courses in
Linguistics and Literature, followed by four selected seminars dealing with
American comics? - and queries concerning the culture-specificity of
academic norms, insofar as these are linked with a particular language, namely
English. This issue has recently been discussed in the context of the teaching of
specialised registers of English to foreign students wishing to study in an
English-speaking university, or function inside some English-speaking work
context. The discussion raises the issue as to how far such teaching is justified,
how far it is merely a form of linguistic and/or cultural imperialism (Santos
1992; Pennycock 1994).
It is not my purpose to dispute the legitimacy of these topics and
questions, or their potential sociological, political, philosophical or practical
importance. My purpose is simply to illustrate some current developments in
different fields which link up with a relativistic view of discourse coherence,
interpretation and judgement, and which implicitly or in some cases explicitly
(e.g. Fairclough, Pennycock) challenge the very foundations of discourse
analysis as I wish to understand this term. Such ultimately relativistic views
rob discourse analysis of any rationale save that of establishing its own
relativity.
A second, related, and currently fashionable validation procedure inside
discourse analysis is, in my opinion, to be taken less seriously, but is no less
widespread, though it is, to the best of my knowledge, never stated explicitly.
It is the Machiavellian view that the inherent interest, appropriateness, cultural
value, or indeed political correctness of the analytic results guarantees the
validity of the analytic procedures whereby they were obtained. We may refer
to this view as the resultative stance. Whereas the discourse-as-ideology
position can be linked to the view that 'anything goes' in terms of the outcomes
of discourse analysis, the resultative stance can be linked with the view that
anything goes methodologically, if the results satisfy various social or
ideological criteria. There are, in fact, links between the discourse-as-ideology
and the resultative positions. The philosophical stance concerning discourse
analysis adopted by Fairclough and Pennycock is consistent with the first view,
but the ways in which this philosophy is implemented in its proponents' own
analyses seem to be validated only in terms of the results their analyses lead to
- results which are themselves predictable, given the philosophy behind them.
Thus a resultative stance is taken, in order to support a relativistic view.
The resultative stance can further be seen to operate when the interest of
the data constitutes the interest of its analysis. Consider for example Peräkylä
(1993). The paper is rich in data, but remains, as far as I can see, descriptive.
The major point that emerges from the analysis of counsellor behaviour is as
follows: "As a summary, the recurrent management of the epistemological
framework of the counsellors' description of the clients' future seems to display
the counsellors' orientation to such descriptions as sensitive and delicate
matters". The empirical basis for this conclusion is essentially the observation
that in these counselling sessions, AIDS is mentioned, if at all, only as a
hypothetical possibility: death is mentioned, if at all, as something that will
inevitably confront Everyman. The author suggests finally that the analysis
attains a certain poignancy because behind the hypothetical a real possibility is
lurking: in other words, AIDS and/or death may not be far away.
Voss et al (1996) claim that in a "constructivist" framework, the
assumption is that "the reader or listener constructs a mental representation that
is a fonction of the discourse contents and characteristics of the writer". For
these authors, race and attitudes influence discourse processing (p. 104). This is
an extraordinary claim. What the authors seem to mean is that race and attitude
influence one's perceptions and social values. This is certainly true, but this has
little to do with discourse processing, I venture to suggest.
Consider too Kurzon (1996). The violation of the maxim of quantity in
terms of hyponymy is illustrated via an analysis of a broadcast interview given
by Princess Diana. The sole point of the article seems to be to document the
fact that the interviewee referred often enough in this interview to "'people in
my environment'", instead of saying, perhaps, "the Queen", "Mum", "Liz", or
whatever.
One does not wish to deny the validity, and indeed the inherent interest,
of such data, and of such analytic results. I do however want to cast doubt on
the claim that such analyses are necessary because of their plausibility and
interest instances of discourse analysis in a technical sense.
4 Discourse analysis and discourse theory
My simple position is that discourse analysis should be based on a theory of

discourse, inside which coherence will be explicitly handled. I do not find it
appropriate when seeking to overview our field to loosely define "a discourse"
- for example as a suprasentential stretch of language - and then say that any
attempt to analyse such objects counts as discourse analysis (cf. for example
Tannen 1990). If an analysis of discourse is to count as discourse analysis, it

presumably has to analyse certain types of things. Although I may paint my
fingernails every evening, I do not thereby lay claim to be a painter. Discourse
analysis will, on my view, seek to explicate certain discourse-specific aspects
of that data, and the most obviously relevant aspects are the structural, the
interactional and the pragmatic. In other words, a discourse is a discourse by
virtue of the type or types of structure whereby it exhibits coherence, its
essential interactional nature, and the fact that it constitutes an instance of
language in use, i.e. relates to language users in the sense of Morris (1964).
It is further useful to distinguish between an instance of discourse
analysis and an instance of discourse interpretation. An interpretation, on my
understanding, explicitly involves the views of the interpreter - the term is
being used here non-technically. So discourse analysis can be used in the
service of discourse interpretation, but an interpretation of a discourse segment
may be proposed independently of an analysis (in the relevant technical sense)
of that discourse segment. It is, for example, perfectly possible to examine any
discourse in terms of the extent to which the actants accept the basic tenets of
Marxism, Christianity, or any other creed. You can inspect its content in terms
of its implicit or explicit chauvinistic or ideological stance, and in myriad other
ways. We all interpret texts and discourses as relevant to our own life concerns
all the time. All human actants are interpreting discourse segments continually
in discourse participation. Further, of course, interpreting discourse segments
as a non-participant is a widespread activity. For example, historians, literary
critics, psychotherapists and criminal lawyers examine transcripts of discourse
for various totally valid reasons, and will, more often than not, in their analyses
take a "discourse perspective" in that mentalistic states, intentional stances, and
philosophical leanings might be reconstructed in the course of such
interpretations. Does the exercise of such interpretive skills necessarily count
as discourse analysis? I would say not. This claim is not, of course, a value
judgement: it is an attempt to delineate the scope of a technical term.
On my understanding, then, it is not what you handle (you analyse
'discourse', therefore you are doing discourse analysis), nor what comes out of
the transaction (I called this the resultative stance a moment ago) that defines
discourse analysis. Instead, it is both the goals that guide the undertaking, and
the nature of the procedures employed while pursuing them that define the
field. The issue of goals I have already mentioned - it is essentially a matter of
discovering what it is that makes a given discourse work as such, and
structural, interactional, and pragmatic aspects seem to me to be paramount. As
to how you do it, discourse analysis will make use of an explicit set of
principles - a discourse theory. Which theory you hold and expound is not the
point at issue here. It is not my purpose to be prescriptive. My claim is simply

that explicitness is at least a necessary step towards validation.
I want now in the second half of this paper to elaborate some aspects of a
theory of discourse. There are two topics I wish to touch on. The first step will
be a sketch of a view of discourse which distinguishes between universal and
culture-specific discourse features. The goal is to relativise the notion of
cultural and/or ideological relativity. Secondly, the notion of co-existent
discourse worlds will be illustrated, as one way of capturing the multiple-
interpretability hypothesis descriptively, without succumbing to an ultimately
relativistic stance. In addressing both these topics, I want to sketch a
framework for discourse analysis inside which both system and insight, both
the nature of discourse as displayed above, and the requirement of explicitness
developed here, can be reconciled.
5 Towards a universal theory of discourse
The relativistic stance - the view that nothing goes because anything goes (or
vice versa) - implies that there are no discourse universals. Therefore the
degree to which we can argue the opposite may determine our success in
demonstrating that the discourse analyses we propound and practice have a
valid grounding outside of our own discourse interpretive skills, intuitions, or
prejudices. A four-levelled model might be proposed, as follows.
Level 1 is biological, and focuses on the deep-seated tension between
individual and social drives. Noli me tangere and Come together! Self versus
Other as opposed to Self via Other. It can be shown that interaction rituals have
been developed to reconcile these drives. At this level, then, some basic
interactional structures can be posited, as they can be discovered in many
animal species, and in other semiotic systems than human language. The tied-
pair of ethnomethodological infamy is the most simplistic example.
Level 2 we may call sociolinguistic. Here the assumption operates that,
with humans language has further been developed and adopted - amongst
other things - to serve and reconcile these biological drives operating at level
1. When language comes into the picture, further interactional structures
become possible, for two reasons. Firstly, linguistic symbols can themselves
function as social tender - thus an apology both stands for and constitutes an
act of redress. Secondly, language itself can be topicalised. We can therefore
talk about talk. This has structural consequences. For example, if I solicit your
bread, and you comply with this request, the transaction is potentially closed:
if, however, I solicit your opinion, and you comply with this request, you
thereby open an exchange, rather than closing one (for details of such
structures, cf. Edmondson 1981a).
But the main thrust of the argument at level 2 is that certain outcomes
have to be possible via talk if human society is to function. Further, various
'speaker meanings', in my terms, have to be possible, if such outcomes are to be
reached via discourse. For example, agreement on a cooperative undertaking is
surely an outcome without which no social planning, task-sharing or
cooperative achievement would be possible. If such an outcome is to be
reached, and if speech is to play a decisive role to this end, then an utterance
whereby a speaker indicates the desirability of some future collective action is
necessary, as is an utterance whereby a speaker indicates his or her willingness
to participate. At this broad sociohnguistic level, the claim is not that things
like suggestions, offers, requests, apologies and so on are universal categories
of "speech acts" or speaker meanings. The claim is simply that such speaker
meanings, which enable discourse outcomes which fulfil biologically-based
social needs, will be of interactional significance in any speech community.
The third level I want to call interactional. At this level very broadly-
based constraints on human interaction operate. Roughly, the idea is that level
1 gives us biological constraints and goals. Level 2 gives us some requisite
discourse outcomes, together with some communicative categories needed in
order to reach them, while level 3 concerns general rules which govern the
exchange of such speaker stances in discourse interaction. It is a question of
general performance constraints. This is then the level at which conversational
maxims, theories of politeness or of relevance operate. It is very broadly-based,
and concerns essentially the ways in which the universally-relevant categories
and/or outcomes at level 2 can be realised linguistically (or indeed by other
communicative means).
The fourth level is that of culturally-institutionalised realisation. It is the
level of empirical observation. Different discourse conventions clearly hold in
different institutionalised contexts inside and outside one culture, whether
national, linguistic or professional. It is all too often at this level that cross-
cultural comparisons are made, and a search for universals is implemented.
General constraints may, of course, be found to operate at level 4 for all
languages, but it is unlikely that they will, except when such constraints are
formulated at an exceedingly high level of generality, i.e. at level 3 above, or
higher. Thus, for example, Kasper (1995) worries about the bases of
contrastive pragmatics; Rose (1994) has claimed that the discourse completion
tests used for various language groups inside CCSARP - a multi-cultural
research project (cf. Blum-Kulka, House and Kasper 1989) - may simply not
apply in some cultures at all, a worry voiced already in Wolfson et al (1989).
At this level, too, of course, changes occur over time: cultural discourse norms
are not static.
Clearly, the above constitutes at best the bare outlines of a multi-levelled
theory. It suggests, though, that relativistic views of discourse underplay the
universal core that underpins all human verbal interaction. Following this
approach, then, specific types of discourse behaviour could be related to
various kinds of discourse universals. It is an interesting fact, I think, that
inside linguistics, the a priori evidence for a universal theory of discourse is
much more present and biologically warranted than the evidence for a
universal theory of grammar, though universal bases for linguistic competence
have been investigated in great detail, while a universal base for discourse or
communicative competence has received very little attention.
6 The concept of co-existing discourse worlds

Finally, I would like to invoke the notion of co-existent discourse worlds. The
goal is to suggest that the concept allows of rich analytic insight, without
condemning us to unbridled relativity. Originally, a discourse world was
defined as a function from locutionary act (the act of performing a locution) to
discourse act (Edmondson 1981a). It is clear, however, that a discourse world
does not consist of a set of external contextual features, but rather a mental
representation of a set of contextual features: in other words, discourse worlds
are conceptual spaces, and can be readily interpreted in terms of frame theory,
for example. Descriptively, the concept is relevant for at least three classes of
discourse phenomena:
6.1 World-switching
Firstly, world-shifting and switching can occur quite rapidly inside one flow of
text - Tristam Shandy is an early literary exploitation of this fact:
My mother, you must know.. but I have fifty things more necessary to let you know
first..I have a hundred difficulties which I have promised to clear up, and a
thousand distresses and domestic misadventures crowding in upon me, and
threefold, one upon the neck of another. A cow broke in (tomorrow morning) to my
uncle Toby's fortifications, and ate up two rations and a half of dried grass.. Trim
insists on being tried by a court-martial- the cow to be shot- Slop to be crucified..
I want swaddling., but there is no time to be lost in exclamations.. I have left my
father lying across his bed and promised I would go back in half an hour; and five-
and-thirty minutes are lapsed already.
(Tristam Shandy, p. 170)
The foreign language classroom offers rich data on world-switching, and

may lead to confusion on the part of the learning participants, as pupils may be
uncertain as to which world they are operating in at a specific point in time.
Two variables at least contribute to the multi-world views co-existing in this
setting. The first is the embedding of a role-play scenario inside the classroom
scenario, while the second concerns the status of the foreign language - it is
the object and the means of learning, and a means of communication as well
(cf. Edmondson 1985).
6.2 Communicative networking
Secondly, not only do speakers switch frame, as it were, inside one and the
same discourse context, but they may simultaneously address different
audiences at the same time, transmitting different speaker meanings to them via
the same utterance or set of utterances. So the notion of co-existent discourse
world encompasses here superimposed communicative networks. For example,
remarks addressed to small children in the presence of their parents are
commonly primarily addressed to those parents. A teacher may give positive
feedback to an individual pupil, and at the same time tell the rest of the class
that the answer was not quite right.
6.3 Exploiting co-present discourse meanings
Thirdly, we have not only world-shifting in contiguous segments of speech,

and not only simultaneously-operating communicative networks, we also have
co-existent discourse worlds activated via one utterance addressed to one and
the same interactant.
That we entertain interpretive options, and indeed sometimes fail to
resolve them, is I suggest a cliché of human experience. In his classic work on
ambiguity (Richards 1930), I.A. Richards claimed that all potential ambiguities
in a text are co-present to the reader, and enrich the text's texture and meaning.
Humour also operates with co-present, multiple meanings.
The concept of co-existent discourse worlds is, I suggest, psychologically
grounded, cognitively interpretable, and can be incorporated into an explicit
theory of discourse. It is also substantiated experientially and empirically. It is
a concept, therefore, that may be invoked to retain systematicity, while
allowing for multiple data analysis. Discourse data is immensely rich: it seems
desirable that the discourse framework we operate with should expand rather
than contract our perception of that richness.
7 Summary
In this theoretical paper, I have put forward a view of discourse and of
discourse coherence which necessarily invokes the perceptual and interpretive
skills of human minds. I went on to suggest that the nature of discourse raises
problems of validation inside discourse analysis. To resort to relativistic
approaches is however to bypass the issue of validation, and is inconsistent
with viewing discourse analysis as a principled, scientific undertaking. In an
attempt to go some way towards avoiding relativity, but respecting discourse
data, I then presumed to sketch some elements which together constitute a
rudimentary universal theoretical base for discourse undertakings. Finally, I
went on to look at the notion of co-existing frames of reference or "discourse
worlds", as one conceptual attempt to reconcile the implications of the
multiple-meanings hypothesis with the requirements of theoretical and
structural explicitness.
As said at the beginning of the paper, the term "discourse analysis" is
used with very many senses, and I do not wish to claim that the view of
discourse analysis I have taken in this paper is in some sense primary. I do
wish to claim, however, that if discourse analysis is seen as a principled
approach to handling instances of language in use, then it has to be possible to
argue about a given piece of data analysis. This means, as far as I can see, that
the issue of analytic validation cannot be avoided.
Notes
1. The four-volumed Handbook of Discourse Analysis (van Dijk 1985) is still good evidence
for this observation. More recently, Schiffrin (1994) has devoted a book to different
concepts of discourse. On different concepts of coherence, cf. for example Sanders et al
(1992).
2. I want to refer to 'speaker-meaning' as the stance adopted by a speaker at a specific point in
the discourse via a discourse move. More abstractly, I shall refer to discourse meaning to
indicate the significance of any unit of discourse inside the whole. Inside the theory of
discourse that I uphold, speaker meaning and discourse meaning are not the same thing.
The latter is roughly made up of the former in the light of its structural placing. So the
meaning of a discourse as such derives from its constituting a coherently structured
sequence of speaker meanings. Cf. for example Edmondson (1981a) for details.
References
Blum-Kulka S., House, J. and Kasper, G. (eds). 1989. Cross-Cultural Pragmatics:
Requests and Apologies. Norwood: Ablex.
Dijk, T. van. (ed). 1985. Handbook of Discourse Analysis. London: Academic Press.
Edmondson, W.J. 1981a. Spoken Discourse: A Model for Analysis. London: Longman.
Edmondson, W.J. 1981b. "Illocutionary verbs, illocutionary acts, and conversational
behaviour". In H.-J. Eikmeyer and H. Rieser (eds), Words, Worlds and
Contexts. Berlin: de Gruyter, 485-499.
Edmondson, W.J. 1985. "Discourse worlds in the classroom and in foreign language
learning". Studies in Second Language Acquisition 7: 159-168.
Edmondson, W.J. 1989a. "Principles of cooperation in interaction". In R. Dirven (ed),
A User's Grammar of English: Word, Sentence, Text, Interaction. Frankfurt:
Lang, 809-840.
Edmondson, W.J. 1989b. "Discourse production, routines and language learning". In
B. Kettemann et al. (eds), Englisch als Zweitsprache. Tübingen: Narr, 287-302.
Edmondson, W.J. 1990. "Can one usefully do discourse analysis without investigating
discourse processing?" In H. Nyyssónen et al. (eds), Proceedings from the 2nd
Finnish Seminar on Discourse Analysis. Oulu: University of Oulu, 27-42.
Fairclough, N. (ed). 1992. Critical Language Awareness. London: Longman.
Kasper, G. 1995. "Wessen Pragmatik? Für eine Neubestimmung fremdsprachlicher
Handlungskompetenz". Zeitschrift für Fremdsprachenforschung 6: 69-94.
Kurzon, D. 1996. "The maxim of quantity, hyponymy and Princess Diana".
Pragmatics 6: 217-227.
Morris, C. 1964. Signification and Significance. Cambridge, MA: MIT Press.
Pennycock, A. 1994. "Incommensurable discourses". Applied Linguistics 15: 115-138.
Peräkylä, A. 1993. "Invoking a hostile world: Discussing the patient's future in AIDS
counselling". Text 13: 291-316.
Richards, I.A. 1930. Seven Types of Ambiguity. London: Chatto & Windus (Revised
edition 1961, Harmondsworth: Penguin).
Roberts, R.M. and Kreuz, R.J. 1993. "Nonstandard discourse and its coherence".
Rose, K.R. 1994. "On the validity of discourse completion tests in non-western
contexts". Applied Linguistics 15: 1-14.
Sanders, T., Spooren, W. and Noordman, L. 1992. "Toward a taxonomy of coherence
Santos, T. 1992. "Ideology in composition: L1 and ESL". Journal of Second
Language Writing 1: 1-15.
Stolze, R. 1992. Hermeneutisches Übersetzen. Tübingen: Narr.
Tannen, D. 1990. "Discourse analysis: the excitement of diversity". Text 10: 109-111.
Voss, J.F., Wiley, J., Ciarrochi, J., Foltz, P. and Silfies, L. 1996. "Race and the
representation of discourse: fictitious scenarios and the O.J. Simpson case".
Widdowson, H. 1995. "Discourse analysis: a critical view". Language and Literature
4: 157-172.
Widdowson, H. 1996. "Reply to Fairclough: Discourse and interpretation - conjectures
and refutations". Language and Literature 5: 57-69.
Wierlacher, A. 1992. "Toleranzforschung. Zur Forschungsplanung interkultureller
Germanistik". Jahrbuch Deutsch als Fremdsprache 19: 13-29.
Wolfson, N., Marmor, T. and Jones, S. 1989. "Problems in the comparison of speech
acts across cultures". In S. Blum-Kulka, J. House and G. Kasper (eds), Cross-
Cultural Pragmatics: Requests and Apologies. Norwood: Ablex, 174-196.
Zimmermann, P. (ed) 1989. Interkulturelle Germanistik: Dialog der Kulturen auf
Deutsch? Frankfurt: Lang.
A Bibliography of Coherence and Cohesion
Uta Lenk, Sarah Gietl and Wolfram Bublitz
The compilation of this extensive bibliography began as part of our preparation

for the International Workshop on Coherence in Discourse at the University of
Augsburg in 1997. We soon learned that the accessible literature on coherence
and cohesion is vast and continues to grow rapidly. Originally intended to serve
as reference material for the workshop participants, the bibliography began to
grow quickly and developed into A Select and Far From Comprehensive
Bibliography of Coherence and Cohesion (which was the original title of the
booklet we distributed to the workshop participants).
The present updated and revised version of the bibliography would not
have been possible without the help from Heinz Auernhamer, Nils Engel, Thilo
Jörgl and Gudrun Nelle and all of the contributors to this reader, who
supplemented a considerable number of references.
Originally, we intended to have one single list of references for the entire
volume and none for each individual contribution. But noting that most
contributors also refer to publications that are not directly related to coherence
or cohesion and are therefore not included in the general bibliography, we
decided to allow for a certain amount of overlap and to also include each
author's reference section. In order to be as comprehensive as possible, we have
attempted to list - to the best of our knowledge - all major publications on
coherence and cohesion. (We apologise to those whose articles or monographs
we have overlooked and would greatly appreciate notification of omissions
which we could use for future updates.) Due to limited space we found it
necessary to omit readers and textbooks in which coherence and cohesion are
only side issues, as well as titles concerned mainly with related issues such as
discourse topics and discourse (or pragmatic) markers.
The bibliography lists more than 400 titles, and we hope that they may
serve as the basis for further research on coherence, which we found a highly
gratifying issue to talk about during our workshop and which still warrants
many more interesting studies.
268 Uta Lenk, Sarah Gietl, and Wolfram Bublitz
Abadi, A. 1995. "Coherence/incoherence in stories by S.Y. Agnon". In B.

Wårvik, S.-K. Tanskanen and R. Hiltungen (eds), Organization in
Discourse. Proceedings from the Turku Conference. Turku: University of
Turku, 131-137.
Abraham, W., Givón, T. and Thompson, S. (eds). 1994. Discourse Grammar
and Typology. Papers in Honor of John W.M. Verhaar. Amsterdam:
Benjamins.
Ackerman, B.P. 1986. "Referential and causal coherence in story
comprehension of children and adults". Journal of Experimental Child
Psychology 41: 336-366.
Agar, M. and Hobbs, J.R. 1982. "Interpreting discourse: coherence and the
analysis of ethnographic interviews". Discourse Processes 5: 1-32.
Agricola, E. 1976. "Vom Text zum Thema". Probleme der Textgrammatik.
Studia Grammatica XL Berlin: Akademie-Verlag, 13-27.
Aijmer, K. 1986. "Why is actually so popular in spoken English?" In G. Tottie
and I. Bäcklund (eds), English in Speech and Writing. A Symposium.
Uppsala: Almqvist and Wiksell, 119-129.
Aijmer, K. 1987. "Oh and ah in English conversation". In W. Meijs (ed),
Corpus Linguistics and Beyond. Proceedings of the Seventh International
Conference on English Language Research on Computerized Corpora.
Amsterdam: Rodopi, 61-86.
Aijmer, K. 1988. "'Now may we have a word on this': the use of 'now' as a
discourse particle". In M. Kytö, O. Ihalainen and M. Rissanen (eds),
Corpus Linguistics, Hard and Soft. Proceedings of the Eighth
International Conference on English Language Research on
Computerized Corpora. Amsterdam: Rodopi, 15-33.
Allwood, J. 1992. On Dialogue Cohesion. Göteborg: University of Göteborg.
Altenberg, B. 1986. "Contrastive linking in spoken and written English". In G.
Tottie and I. Bäcklund (eds), English in Speech and Writing. A
Symposium. Uppsala: Almqvist and Wiksell, 13-40.
Alterman, R. 1985. "A dictionary based on concept coherence". Artificial
Intelligence 25: 153-186.
Alterman, R. 1989. "Event concept coherence". In D.L. Waltz (ed), Semantic
Structures: Advances in Natural Language Processing. Hillsdale:
Erlbaum, 57-87.
Anderson, A. 1995. "Negotiating coherence in dialogue". In M. Gernsbacher
and T. Givón (eds), Coherence in Spontaneous Text. Amsterdam:
Benjamins, 1-40.
Bibliography 269
Anderson, R.C., Reynolds, R.E., Schallert, D.L. and Goetz, E.T. 1977.
"Frameworks for comprehending discourse". American Educational
Research Journal 14: 367-381.
Andersson, E. 1978. "An analysis of textual cohesion in a passage from Maria
Gripe's 'Hugo och Josefin'". In J.-O. Östman (ed), Cohesion and
Semantics. Åbo: Akademi, 131-143.
Andriessen, J., de Smedt, K. and Zock, M. 1996. "Discourse planning:
Empirical research and computer models". In T. Dijkstra and K. de Smedt
(eds), Computational Psycholinguistics. London etc. : Taylor and Francis,
247-278.
Aronsson, K. and Rundström, B. 1989. "Cats, dogs, and sweets in the clinical
negotiation of reality: on politeness and coherence in pediatric discourse".
Language in Society 18: 483-504.
Asher, N. and Lascarides, A. 1994. "Intentions and information in discourse".
Proceedings of the 32nd Annual Meeting of the Association of
Computational Linguistics June, 27-30, 1994. New Mexico.
Auer, P. 1986. "Kontextualisierung". Studium Linguistik 19: 22-47.
Auer, P. 1995. "Context and contextualization". In J. Verschueren, J.-O.
Östman and J. Blommaert (eds), Handbook of Pragmatics, 1995
Installment. Amsterdam: Benjamins. (19pp.)
Auer, P., Couper-Kuhlen, E. and Mueller, F. 1999. Language in Time: The
Rhythm and Tempo of Spoken Interaction. New York: Oxford University
Press.
Bäcklund, I. 1988. "To begin with, this is the problem, for example. On some
reader-oriented structural markers in English expository text". Studia
Linguistica 42: 60-68.
Bäcklund, I. 1990. "To sum up. Initial infinitives as cues to the reader". In G.
Caie et al. (eds), Proceedings from the Fourth Nordic Conference for
English Studies. Helsingør, May 11-13, 1989. Copenhagen: University of
Copenhagen, 289-302.
Ball, W.J. 1986. Dictionary of Link Words in English Discourse. London:
Macmillan.
Baltaxe, C.A. and D'Angiola, N. 1992. "Cohesion in the disclosure interaction
of autistic, specifically language-impaired, and normal children". Journal
of A utism and Developmental Disorders 22: 1-21.
Bamberg, M. and Marchman, V. 1990. "What holds a narrative together? The
linguistic encoding of episode boundaries". Papers in Pragmatics 4: 58-
121.
Banfield, A. 1978. "The formal coherence of represented speech and thought".
A Journal for Descriptive Poetics and Theory of Literature 3: 289-314.
Barberis, J.M. 1991. "Cohérences discursives". Cahiers de Praxématique 16: 3-

153.
Bateman, J.A. and Rondhuis, K.J. 1997. "'Coherence relations': towards a
general specification". Discourse Processes 24: 3-49.
Bazzanella, C. and Called, D. 1991. "Tense coherence and grounding in
children's narrative". Text 11: 175-187.
Bazzanella, C. and Damiano, R. 1997. "I1 fraintendimento nelle interazioni
quotidiane: proposte di classificazione". Lingua e Stile 32: 173-200.
Bazzanella, C. and Damiano, R. This volume. "Coherence and misunderstanding
in everyday conversations".
Bazzanella, C. (ed). 1996. Repetition in Dialogue. Tübingen: Niemeyer.
Beach, W.A. 1983. "Background understandings and the situated
accomplishment of conversational telling-expansions". In R. Craig and K.
Tracy (eds), Conversational Coherence: Form, Structure, and Strategy.
Beverly Hills: Sage, 196-221.
Beaugrande, R. de. 1980. Text, Discourse and Process. Toward a
Multidisciplinary Science of Texts. Norwood: Ablex.
Beaugrande, R. de and W.U. Dressier. 1981. Einführung in die Textlinguistik.
Tübingen: Niemeyer.
Beaugrande, R. de and W.U. Dressler. 1981. Introduction to Text Linguistics.
London: Longman.
Behre, F. 1961. "On the principle of connecting elements of speech in
contemporary English". In G.A. Bonnard (ed), English Studies Today.
Lectures and Papers Read at the 4. Conference of the International
Association of University Professors of English Held at Lausanne and
Berne, August 1959. Bern: Francke, 303-316.
Bellert, I. 1970. "On a condition of the coherence of texts". Semiotica 2: 335-
363.
Benes, E. 1973. "Thema-Rhema-Gliederung und Textlinguistik". In H. Sitta and
K. Brinker (eds), Studien zur Texttheorie und zur Deutschen Grammatik.
Düsseldorf: Schwann, 42-62.
Bennett-Kastor, T. 1983. "Noun phrases and coherence in child narratives".
Journal of Child Language 10: 135-149.
Benson, J.D. and Greaves, W.S. 1981. "Field of discourse: theory and
application". Applied Linguistics II: 45-55.
Black, A. 1987. "A method for measuring conversational coherence".
University of Victoria, Working Papers of the Linguistics Circle 6: 17-24.
Black, A. 1988. "The syntax of conversational coherence". Discourse Processes
11:433-455.
Bibliography 271
Black, J.B. and Bern, H. 1981. "Casual coherence and memory for events in
narratives". Journal of Verbal Learning and Verbal Behavior 20: 267-
275.
Blakemore, D. 1988. "The organization of discourse". In F. Newmeyer (ed),
Linguistics: the Cambridge Survey, Vol IV, Cambridge: Cambridge
Blakemore, D. 1992. Understanding Utterances. Oxford: Blackwell.
Bosch, P. 1989. "Coherence and cohesion: comments on Roger G. van de
Velde's paper 'Man, verbal text, inferencing, and coherence'". In W.
Heydrich, F. Neubauer, J. Petöfi and E. Sözer (eds), Connexity and
Coherence: Analysis of Text and Discourse. Berlin: de Gruyter, 218-227.
Braecke, C , Geluykens, R. and Pelsmaekers, K. 1997. "Clause ordering as a
text-building device in written L2". In M. Pütz. (ed), The Cultural
Context in Foreign Language Teaching. Amsterdam: Benjamins, 35-52.
Broek, P. van den. 1990. "Causal inferences and the comprehension of narrative
texts". In A.C. Graesser and G.H. Bower (eds), Inferences and Text
Comprehension. San Diego: Academic Press, 175-196.
Brown, E.K. 1980. "Grammatical incoherence". In H. Dechert and M. Raupach
(eds), Temporal Variables in Speech. The Hague: Mouton de Gruyter,
27-37.
Brown, G. and Yule, G 1983. Discourse Analysis. Cambridge: Cambridge
University Press.
Bublitz, W. 1989. "Topical coherence in spoken discourse". Studia Anglica
Posnaniensia 22: 31-51.
Bublitz, W. 1994. "In the eye of the beholder: 'the rather mystical notion of
coherence'". In K. Carlon, K. Davidse and B. Rudzka-Ostyn (eds),
Perspectives on English. Studies in Honour of Professor Emma Vorlat.
Leuven, Paris: Peeters, 213-230.
Bublitz, W. 1996a. "'I bought a freezer - you've bought a freezer - they've
bought a freezer': repetition as a text-building device". In C. Bazzanella
(ed), Repetition in Dialogue. Tübingen: Niemeyer, 16-28.
Bublitz, W. 1996b. "Semantic prosody and cohesive company: somewhat
predictable". Leuvense Bijdragen 85: 1-32.
Bublitz, W. (forthcoming). "Cohesion and coherence". In J. Verschueren, J.-O.
Östman, J. Blommaert and C. Bulcaen (eds), Handbook of Pragmatics.
Bublitz, W. and Lenk, U. This volume. "Disturbed coherence: 'Fill me in'",
Campbell, K.S. 1995. Coherence, Continuity, and Cohesion. Theoretical

Foundations for Document Design. Hillsdale: Erlbaum.
Carlson, L. 1983. Dialogue Games. An Approach to Discourse Analysis.
Dordrecht: Reidel.
Carrell, P. 1982. "Cohesion is not coherence". TESOL Quarterly 16: 479-488.
Carter, R. and McCarthy, M. 1988. Vocabulary and Language Teaching.
London: Longman.
Chafe, W. 1972. "Discourse structure and human knowledge". In J.S. Carroll
and R.O. Freedle (eds), Language Comprehension and the Acquisition of
Knowledge. Washington: Wiley, 41-69.
Chafe, W. 1974. "Language and consciousness". Language 50: 111-133.
Chafe, W. 1987. "Cognitive constraints on information flow". In R. Tomlin
(ed), Coherence and Grounding in Discourse. Outcome of a Symposium,
Eugene, Oregon, June 1984. Amsterdam: Benjamins, 21-51.
Chafe, W. 1994. Discourse, Consciousness, and Time. The Flow and
Displacement of Conscious Experience in Speaking and Writing.
Chicago: Chicago University Press.
Chafe, W. (ed). 1980. The Pear Stories. Cognitive, Cultural, and Linguistic
Aspects of Narrative Production. Norwood: Ablex.
Charolles, M. 1983a. "Coherence as a principle in the interpretation of
discourse". Text 3: 71-97.
Charolles, M. 1983b. "Towards a heuristic approach to text-coherence
problems". In F. Neubauer (ed), Coherence in Natural-Language Texts.
Hamburg: Buske, 1-16.
Charolles, M. 1985. "Text connexity, text coherence and text interpretation
processes". In E. Sözer (ed), Text Connexity, Text Coherence. Aspects,
Methods, Results. Hamburg: Buske, 1-15.
Charolles, M. 1989. "Coherence as a principle in the regulation of discourse
production". In W. Heydrich, F. Neubauer, J. Petöfi and E. Sözer (eds),
Connexity and Coherence: Analysis of Text and Discourse. Berlin: de
Gruyter, 3-15.
Charolles, M., J.S. Petöfi and E. Sözer (eds) 1986. Research in text connexity
and text coherence. Hamburg: Buske.
Cherchi, L. 1978. "L'ellipse comme facteur de cohérence". Langue Française
38: 118-128.
Ciliberti, A. 1998. "Egocentric vs. self-denying communication in an asymmetric
event". In S. Cmejrkovâ et al. (eds), Dialogue Analysis VI. Tübingen:
Niemeyer, 29-38.
Ciliberti, A. This volume. "The effect of context in the definition and
negotiation of coherence".
Bibliography 273
Cirilo, R.K. 1981. "Referential coherence and text structure in story

comprehension". Journal of Verbal Learning and Verbal Behavior 20:
358-367.
Clark, H.H. and Schaeffer, E. 1987. "Collaborating on contributions to
conversations". Language and Cognitive Processes 2: 19-41.
Clark, H.H. and Schaeffer, E. 1989. "Contributing to discourse". Cognitive
Science 13:259-294.
Coates, J. 1995. "The negotiation of coherence in face-to-face interaction: some
examples from the extreme bounds". In M. Gernsbacher and T. Givon
(eds), Coherence in Spontaneous Text. Amsterdam: Benjamins, 41-58.
Colby, B.N. 1987. "Coherence in language and culture". In R. Steele and T.
Threadgold (eds), Language Topics. Essays in Honour of Michael
Halliday, Vol. I. Amsterdam: Benjamins, 451-460.
Connor, U. 1984. "A study of cohesion and coherence in English as a second
language students' writing". Papers in Linguistics 17: 301-16.
Connor, U. and Johns, A.M. (eds). 1990. Coherence in Writing. Research and
Pedagogical Perspectives. Alexandria: TESOL.
Conte, M.-E. 1980. "Coerenza testuale". Lingua e Stile 15: 135-154.
Conte, M.-E. 1981. "Textdeixis und Anapher". Kodikas/ Code 3: 121-132.
Conte, M.-E. 1986a. "Textreferenten und Typen anaphorischer
Wiederaufnahme". In W. Heydrich and J. Petöfi (eds), Aspekte der
Konnexität und Kohärenz von Texten. Hamburg: Buske, 1-15.
Conte, M.-E. 1986b. "Coerenza, interpretazione, reinterpretazione". Lingua e
Stile 21: 357-372.
Conte, M.-E. 1988. Condizioni di Coerenza. Ricerche di Linguistica Testuale.
Firenze: La Nuova Italia.
Conte, M.-E. 1989. "Coherence in interpretation". In W. Heydrich, F.
Neubauer, J. Petöfi and E. Sözer (eds), Connexity and Coherence:
Analysis of Text and Discourse. Berlin: de Gruyter, 275-282.
Conte, M.-E. 1990. "Pronominale Anaphern im Text". Rivista di Linguistica 2:
141-154.
Conte, M.-E. 1991. "Empathie als Kohärenzfaktor". Folia Linguistica 25: 219-
228.
Conte, M.-E. 1994. "Discontinuity in texts". In S. Cmejrkovâ and F. Sticha
(eds), The Syntax of Sentence and Text. A Festschrift for Frantisek
Danes. Amsterdam: Benjamins, 195-204.
Conte, M.-E. (ed). 1989. Kontinuität und Diskontinuität in Texten und
Sachverhaltskonfigurationen. Hamburg: Buske.
Conte, M.-E., Petöfi, J. and Sözer, E. (eds). 1989. Text and Discourse
Connectedness. Amsterdam: Benjamins.
Cooper, A. 1988. "Given-new: enhancing coherence through cohesiveness".

Written Communication 5: 352-367.
Cornish, F. 1996. "Coherence: the lifeblood of anaphora". Belgian Journal of
Couper-Kuhlen, E. 1983. "Intonatorische Kohäsion. Eine makroprosodische
Untersuchung". Zeitschrift für Literaturwissenschaft und Linguistik 49:
74-100.
Couper-Kuhlen, E. 1993. English Speech Rhythm: Form and Function in
Everyday Verbal Interaction. Amsterdam: Benjamins.
Couper-Kuhlen, E. This volume. "Coherent voicing: On prosody in
conversational reported speech".
Couper-Kuhlen, E. and Setting, M. 1996. "Towards an interactional perspective
on prosody and a prosodic perspective on interaction". In E. Couper-
Kuhlen and M. Selting (eds), Prosody in Conversation. Cambridge:
Cambridge University Press, 11-56.
Couper-Kuhlen, E. and Selting, M. (eds). 1996. Prosody in Conversation.
Interactional Studies. Cambridge: Cambridge University Press.
Craig, R. and Tracy, K. (eds). 1983. Conversational Coherence: Form,
Structure, and Strategy. Beverly Hills: Sage.
Crewe, W.J. 1990. "The illogic of logical connectives". ELT Journal 44: 316-
325.
Crow, B.K. 1983. "Topic shifts in couples' conversations". In R. Craig and K.
Dahl, Ö. 1995. "Causality in discourse". In G. Rickheit and Chr. Habel (eds),
Focus and Coherence in Discourse Processing. Berlin: de Gruyter, 251-
260.
Dahlgren, K. 1989. "Coherence relation assignment". Proceedings of the 11th
Annual Conference of the Cognitive Science Society: 588-596.
Dahlgren, K. 1992. "Convergent evidence for a set of coherence relations". In
D. Stein (ed), Cooperating with Written Texts. The Pragmatics and
Comprehension of Written Texts. Berlin: de Gruyter, 631-663.
Danes, F. 1989. "Report of Roger G. van de Velde's paper 'Man, verbal text,
inferencing, and coherence'". In W. Heydrich, F. Neubauer, J. Petöfi and
E. Sözer (eds), Connexity and Coherence: Analysis of Text and
Discourse. Berlin: de Gruyter, 228-239.
Danes, F. 1990. "On stylistic relevance of the choice of anaphoric expressions".
Rivista di Linguistica 2: 121-139 (special issue on Anaphoric Relations in
Sentence and Text).
Bibliography 275
Dascal, M. and Katriel, T. 1979. "Digressions: a study in conversational

coherence". PTL: A Journal for Descriptive Poetics and Theory of
Literature 4: 203-232.
Davies, M. 1989. "Prosodic and non-prosodic cohesion in speech and writing".
Word 40: 255-262.
Dijk, T. van. 1979. "Pragmatic connectives". Journal of Pragmatics 3: 447-
456.
Dijk, T. van. 1980. "The semantics and pragmatics of functional coherence in
discourse". Versus 26/27: 49-65.
Dorfmüller-Karpusa, K. and Dorfmüller, Th. 1985. "Some interdisciplinary
remarks about coherence". In E. Sözer (ed), Text Connexity, Text
Coherence. Aspects, Methods, Results. Hamburg: Buske, 555-565.
Dressler, W. 1978. Current Trends in Textlinguistics. Berlin: de Gruyter.
Dry, H.A. 1985. "Approaches to coherence in natural and literary narrative". In
E. Sözer (ed), Text Connexity, Text Coherence. Aspects, Methods,
Results. Hamburg: Buske, 484-499.
Dusková, L. 1984. "A contrastive view of grammatical means of textual
cohesion". In J. Korensky and J. Hoffmannova (eds), Text and the
Pragmatic Aspects of Language. Praha: Ustav pro jazyk cesky CSAV,
141-162.
Edmondson, W.J. 1981. Spoken Discourse: A Model for Analysis. London:
Longman.
Edmondson, W.J. 1989. "Principles of cooperation in interaction". In R. Dirven
(ed), A User's Grammar of English: Word, Sentence, Text, Interaction.
Frankfurt: Lang, 809-840.
Edmondson, W.J. This volume "If coherence is achieved, then where doth
meaning lie?"
Ehrlich, S. 1988. "Cohesive devices and discourse competence". World
Englishes 7: 111-118.
Eikmeyer, H.-J., Kindt, W., Laubenstein, U., Lisken, S., Polzin, Th., Rieser, H.
and Schade, U. 1991. "Kohärenzkonstitution im gesprochenen Deutsch".
In G. Rickheit (ed), Kohärenzprozesse. Modellierung von
Sprachverarbeitung in Texten und Diskursen. Opladen: Westdcutscher
Verlag, 59-136.
Eikmeyer, H.-J., Kindt, W., Laubenstein, U., Lisken, S., Rieser, H. and Schade,
U. 1995. "Coherence regained". In G. Rickheit and C. Habel (eds), Focus
and Coherence in Discourse Processing. Berlin: de Gruyter, 115-142.
Eimas, P.D. and Nygaard, L.C. 1992. "Contextual coherence and attention in
phoneme monitoring". Journal of Memory and Language 31: 375-395.
Ellis, D. 1983. "Language, coherence, and textuality". In R. Craig and K. Tracy

(eds), Conversational Coherence: Form, Structure, and Strategy. Beverly
Hills: Sage, 222-242.
Ellis, D., Duran, R. and Kelly, L. 1994. "Discourse strategies of competent
communicators: selected cohesive and linguistic devices". Research on
Language and Social Interaction 27: 145-170.
Ellis, D., Hamilton, M. and Aho, L. 1983. "Some issues in conversation
coherence". Human Communication Research 9: 267-282.
Enkvist, N. 1978. "Coherence, pseudo-coherence, and non-coherence". In J.-O.
Östman (ed), Cohesion and Semantics. Åbo: Akademi, 109-128.
Enkvist, N. 1985. "Coherence and inference". In U. Pieper and G. Stickel (eds),
Studia Linguistica Diachronica et Synchronica. Berlin: de Gruyter, 233-
248.
Enkvist, N. (ed). 1982. Impromptu Speech: a Symposium. Åbo: Akademi.
Enkvist, N. (ed). 1985. Coherence and Composition: a Symposium. Åbo:
Akademi.
Erickson, F. 1982. "Money tree, lasagna bush, salt and pepper: social
construction of topical cohesion in a conversation among Italian-
Americans". In D. Tannen (ed), Analyzing Discourse: Text and Talk.
Washington: Georgetown University Press, 43-70.
Erickson, F. 1984. "Rhetoric, anecdote, and rhapsody: coherence strategies in a
conversation among black American adolescents". In D. Tannen (ed),
Coherence in Spoken and Written Discourse. Norwood: Ablex, 81-154.
Erickson, F. 1990. "The social construction of discourse coherence in a family
dinner table conversation". In B. Dorval (ed), Conversational
Organization and its Development. Norwood: Ablex, 207-238.
Fairley, I.R. 1980. "Syntactic deviation and cohesion". In M.K.L. Ching, M.C.
Haley and R. Lunsford (eds), Linguistic Perspectives on Literature.
London: Routledge, 242-256.
Farrell, T.B. 1983. "Aspects of coherence in conversation and rhetoric". In R.
Craig and K. Tracy (eds), Conversational Coherence: Form, Structure,
and Strategy. Beverly Hills: Sage, 259-284.
Field, Y. and Yip, L. 1992. "A comparison of internal cohesive conjunction in
the English essay writing of Cantonese speakers and native speakers of
English". RELC Journal 23: 15-28.
Fine, J. 1994. How Language Works: Cohesion in Normal and Nonstandard
Communication. Norwood: Ablex.
Fox, B.A. 1987. Discourse Structure and Anaphora. Written and
Conversational English. Cambridge: Cambridge University Press.
Francis, G. 1986. Anaphoric Nouns. Birmingham: University of Birmingham.
Bibliography 277
Francis, G 1994. "Labelling discourse: an aspect of nominal group lexical

cohesion", In M. Coulthard (ed), Advances in Written Text Analysis.
London: Routledge, 83-101.
Fries, P. 1982. "On repetition and interpretation". Forum Linguisticum 7: 50-
64.
Fries, P. 1985. "How does a story mean what it does? A partial answer". In J.D.
Benson and W.S. Greaves (eds), Systemic Perspectives on Discourse:
Selected Theoretical Papers from the 9th International Systemic
Workshop. Norwood: Ablex, 295-321.
Fries, P. 1986. "Language features, textual coherence, and reading". Word 37:
13-29.
Fries, P. 1992. "Lexico-grammatical patterns and the interpretation of texts".
Fries, U. 1972. "Textlinguistik". Linguistik und Didaktik 7: 219-234.
Fritz, G. 1975. "Sprachliche Interaktion: gemeinsam planen". Zeitschrift für
Germanistische Linguistik 3: 257-279.
Fritz, G. 1982. Kohärenz. Grundfragen der Linguistischen Kommunikations
analyse. Tübingen: Narr.
Fritz, G. 1991. "Comprehensibility and the basic structures of dialogue". In S.
Stati, E. Weigand and F. Hundsnurscher (eds), Dialoganalyse III, Vol. 1.
Tübingen: Niemeyer, 3-24.
Fritz, G. 1994. "Grundlagen der Dialogorganisation". In G. Fritz and F.
Hundsnurscher (eds), Handbuch der Dialoganalyse. Tübingen: Niemeyer,
177-201.
Fritz, G. 1997. "Coreference in dialogue". In E. Weigand (ed), Dialogue
Analysis'. Units, Relations and Strategies beyond the Sentence. Tübingen:
Niemeyer, 75-88.
Fritz, G. This volume. "Coherence in hypertext".
Garcia, E.C. 1996. "Measured coherence". Belgian Journal of Linguistics 10:
139-160.
Garnham, A., Oakhill, J. and Johnson-Laird, P.N. 1982. "Referential continuity
and the coherence of discourse". Cognition 11: 29-46.
Garrod, S.C. and Doherty, G. 1995. "Special determinants of coherence in
spoken dialogue". In G. Rickheit and C. Habel (eds), Focus and
Coherence in Discourse Processing. Berlin: de Gruyter, 97-114.
Geluykens, R. 1988. "The interactional nature of referent-introduction". Papers
from the 24th Regional Meeting of the Chicago Linguistic Society: 141-
154.
Geluykens, R. 1991a. "Discourse functions of IT-clefts in English
conversation". Communication and Cognition 24: 343-358.
Geluykens, R. 1991b. "Information flow and conversational discourse: a new

approach to the given-new distinction". In E. Ventola (ed), Functional
and Systemic Linguistics. Berlin: de Gruyter, 141-168.
Geluykens, R. 1991c. "Topic management in conversational discourse: the
collaborative dimension". Papers from the 27th Regional Meeting of the
Chicago Linguistic Society: 182-195.
Geluykens, R. 1992. From Discourse Process to Grammatical Construction:
on Left-Dislocation in English. Amsterdam: Benjamins.
Geluykens, R. 1993. "Topic-introduction in English conversation".
Transactions of the Philological Society 91 : 181-214.
Geluykens, R. 1994a. "Interactional and topical coherence in conversation". In
S. Cmejrková, F. Danes and E. Havlová (eds), Writing versus Speaking.
Tübingen: Narr, 251-257.
Geluykens, R. 1994b. The Pragmatics of Discourse Anaphora in English:
Evidence from Conversational Repair. Berlin: de Gruyter.
Geluykens, R. 1995. "On establishing reference in conversation". In R.A.
Geiger (ed), Reference in Multidisciplinary Perspective. Hildesheim:
Olms, 230-240.
Geluykens, R. (in prep). "Information flow through interaction: Topic
organization in English conversation". Ms., University of Münster.
Geluykens, R. This volume. "It takes two to cohere: the collaborative dimension
of topical coherence in conversation".
Geluykens, R. and Swerts, M. 1992. "Prosodic turn- and topic-finality cues".
Proceedings of the IRCS Workshop on Prosody in Natural Speech,
Philadelphia: 63-70.
Geluykens, R. and Swerts, M. 1993. "Local and global prosodic cues to
discourse organization in dialogues". Proceedings of the ESCA Workshop
on Prosody, Lund: 108- 111.
Geluykens, R. and Swerts, M. 1994. "Prosodie cues to discourse boundaries in
experimental dialogues". Speech Communication 15: 69-77.
Gernsbacher, M. 1990. Language Comprehension as Structure Building.
Hillsdale: Erlbaum.
Gernsbacher, M. and Givón, T. (eds). 1995. Coherence in Spontaneous Text.
Giora, R. 1983. "Segment cohesion". Text 3: 155-182.
Giora, R. 1985a. "Notes towards a theory of text coherence". Poetics Today 6:
699-715.
Giora, R. 1985b. "What's a coherent text?" In E. Sözer (ed), Text Connexity,
Text Coherence. Aspects, Methods, Results. Hamburg: Buske, 16-35.
Giora, R. 1996. "Review article to Morton Gernsbacher, Language
Bibliography 279
Comprehension as Structure Building. Hillsdale: Erlbaum. 1990".

Journal of Pragmatics 26: 417-436.
Giora, R. 1997. "Discourse coherence and theory of relevance: stumbling
blocks in search of a unified theory". Journal of Pragmatics 27: 17-34.
Giora, R., Meiran, N. and Oref, P. 1996. "Identification of written discourse
topics by structure coherence and analogy strategies: general aspects and
individual differences". Journal of Pragmatics 26: 455-475.
Givón, T. 1983. "Topic continuity in discourse: an introduction". In T. Givón
(ed), Topic Continuity in Discourse: a Quantitative Cross-Language
Study. Amsterdam: Benjamins, 5-41.
Givón, T. 1992. "The grammar of referential coherence as mental processing
instructions". Linguistics 30: 5-55.
Givón, T. 1995. "Coherence in text vs. coherence in mind". In M. Gernsbacher
and T. Givón (eds), Coherence in Spontaneous Text. Amsterdam:
Benjamins, 59-116.
Givón, T. (ed). 1983. Topic Continuity in Discourse: a Quantitative Cross-
Language Study. Amsterdam: Benjamins.
Givón, T. (ed). 1989. Conversation: Cognitive, Communicative, and Social
Perspectives. Amsterdam: Benjamins.
Goldberg, J. 1983. "A move toward describing conversational coherence". In R.
Goodwin, C. 1981. Conversational Organization. Interaction between
Speakers and Hearers. New York: Academic Press.
Goodwin, C. 1995a. "The negotiation of coherence within conversation". In M.
Gernsbacher and T. Givón (eds), Coherence in Spontaneous Text.
Goodwin, C. 1995b. "Sentence construction within interaction". In U.
Quasthoff (ed), Aspects of Oral Communication. London: Sage, 198-219.
Gordon, P.C. and Scearce, K.A. 1995. "Pronominalization and discourse
coherence, discourse structure and pronoun interpretation". Memory and
Cognition 23: 313-323.
Gough, V. and Talbot, M. 1996. "'Guilt over games boys play': coherence as a
focus for examining the constitution of heterosexual subjectivity on a
problem page". In C.R. Caldas-Coulthard and M. Coulthard (eds), Texts
and Practices: Readings in Critical Discourse Analysis. London:
Routledge, 214-230.
Graesser, A. (ed). 1990. Conceptual Coherence in Text and Discourse.
Symposium conducted at the 12th annual conference of the Cognitive
Science Society, Cambridge, MA.
Granger, S. and Tyson, S. 1996. "Connector usage in the English essay writing
of native and non-native EFL speakers of English". World Englishes 15:
17-27.
Gray, S. 1995. "Linear coherence and relevance: logic in computer-human
'conversations'". Journal of Pragmatics 23: 627-648.
Grimes, J.E. 1975. The Thread of Discourse. The Hague: Mouton de Gruyter.
Grinder, J. and Postal, P.M. 1971. "Missing antecedents". Linguistic Inquiry 2:
269-312.
Grosu, A. 1987. "On acceptable violations of parallelism constraints". In R.
Dirven and V. Fried (eds), Functionalism in Linguistics. Amsterdam:
Benjamins, 425-457.
Grosz, B.J. and Sidner, C.L. 1986. "Attention, intentions and the structure of
discourse". Computational Linguistics 12: 175-204.
Grosz, B.J. and Sidner, C.L. 1990. "Plans for discourse". In P.R. Cohen, J.
Morgan and M.E. Pollack (eds), Intentions in Communication.
Cambridge, MA: MIT Press, 417-444.
Grosz, B.J., Joshi, A.K. and Weinstein, S. 1995. "Centering: a framework for
modeling the local coherence of discourse". Computational Linguistics
21:203-225.
Gülich, E. and Kotschi, Th. 1983. "Les marqueurs de la reformulation
paraphrastique". Cahiers de Linguistique Française 5: 305-351.
Gumperz, J.J. 1982. Discourse Strategies. Cambridge: Cambridge University
Press.
Gumperz, J.J., Kaltman, H. and O'Connor, M.C. 1984. "Cohesion in spoken
and written discourse: ethnic style and the transition to literacy". In D.
Tannen (ed), Coherence in Spoken and Written Discourse. Norwood:
Ablex, 3-20.
Gutwinski, W. 1974. Cohesion in Literary Texts: a Study of some Grammatical
and Lexical Features of English Discourse. The Hague: Mouton de
Gruyter.
Haberlandt, K. and Bingham, C. 1978. "Verbs contribute to the coherence of
brief narratives: reading related and unrelated sentence triples". Journal of
Verbal Learning and Verbal Behavior 17: 419-426.
Hahn, U. 1985. "On lexically distributed text parsing. A computational model
for the analysis of textuality on the level of text cohesion and text
coherence". In F. Kiefer (ed), Linking in Text. Konstanz: Universität
Konstanz.
Halliday, M. A.K. and Hasan, R. 1976. Cohesion in English. London: Longman.
Bibliography 281
Halliday, M.A.K. and Hasan, R. 1985. Language, Context, and Text: Aspects of
Language in a Social-Semiotic Perspective. Geelong: Deakin and 1989,
Oxford: Oxford University Press.
Hardy, D.E. 1996. "Topic versus cohesion in the prediction of causal ordering
in English conversation". Discourse Processes 21: 237-254.
Hartveldt, R. 1987. Pragmatic aspects of coherence in discourse. Groningen.
Harweg, R. 1968. Pronomina und Textkonstitution. München: Fink.
Harweg, R. 1989. "Is coherence a pragmatic phenomenon? Comments on
Dieter Viehweger's paper 'Coherence - interaction of modules'". In W.
Hasan, R. 1968. Grammatical Cohesion in Spoken and Written English.
London: Longman.
Hasan, R. 1984. "Coherence and cohesive harmony". In J. Flood (ed),
Understanding Reading Comprehension: Cognition, Language, and the
Structure of Prose. Newark: International Reading Association, 181-219.
Hatakeyama, K., Petöfi, J. and Sözer, E. 1985. "Text, connexity, cohesion,
coherence". In E. Sözer (ed), Text Connexity, Text Coherence. Aspects,
Methods, Results. Hamburg: Buske, 36-105.
Hawes, L.C. 1983. "Conversational coherence". In R. Craig and K. Tracy (eds),
Conversational Coherence: Form, Structure, and Strategy. Beverly Hills:
Sage, 285-298.
Heinemann, W. and Viehweger, D. 1991. Textlinguistik. Tübingen: Niemeyer.
Hellman, C. 1995. "The notion of coherence in discourse". In G. Rickheit and
C. Habel (eds), Focus and Coherence in Discourse Processing. Berlin: de
Gruyter, 190-202.
Heritage, J. 1984. "A change-of-state token and aspects of its sequential
placement". In M.J. Atkinson and J. Heritage (eds), Structures of Social
Action. Studies in Conversation Analysis. Cambridge: Cambridge
Heydrich, W. and Petöfi, J. (eds). 1986. Aspekte der Konnexität und Kohärenz
von Texten. Hamburg: Buske.
Heydrich, W., Neubauer, F., Petöfi, J. and Sözer, E. (eds). 1989. Connexity and
Coherence: Analysis of Text and Discourse. Berlin: de Gruyter.
Hickmann, M. 1980. "Creating referents in discourse: a development analysis of
linguistic cohesion". Papers from the Sixteenth Regional Meeting of the
Chicago Linguistic Society.
Hickmann, M. 1991. "The development of discourse cohesion: some functional
and cross-linguistic issues", In G. Piéraut-Le Bonniec and M. Dolitsky
(eds), Language bases ... discourse bases. Amsterdam: Benjamins. 157-

185.
Hinds, J. 1979. "Organization patterns in discourse". In T. Givón (ed), Syntax
and Semantics, Vol 12: Discourse and Syntax. New York: Academic
Press, 135-157.
Hintikka, J. 1986. "Logic of conversation as logic of dialogue". In R.E. Grandy
and R. Warner (eds), Philosophical Grounds of Rationality. Intentions,
Categories, Ends. Oxford: Clarendon Press, 259-276.
Hobbs, J. 1979. "Coherence and coreference". Cognitive Science 3: 67-90.
Hobbs, J. 1982. "Towards an understanding of coherence in discourse". In W.
Lehnert and M. Ringle (eds), Strategies for Natural Language
Processing. Hillsdale: Erlbaum, 223-243.
Hobbs, J. 1983. "Why is discourse coherent?" In F. Neubauer (ed), Coherence
in Natural-Language Texts. Hamburg: Buske, 29-70.
Hobbs, J. 1985. "On the coherence and structure of discourse". In L. Polanyi
(ed), The Structure of Discourse. Norwood: Ablex.
Hobbs, J. 1990a. "Topic drift". In B. Dorval (ed), Conversational Organization
and its Development. Norwood: Ablex, 3-22.
Hobbs, J. 1990b. Literature and Cognition. Menlo Park: Center for the Study
of Language and Information.
Hobbs, J. and Agar, M. 1986. "The coherence of incoherent discourse". Journal
of Language and Social Psychology 4: 213-232.
Hoey, M. 1991a. "Another perspective on coherence and cohesive harmony". In
E. Ventola (ed), Functional and Systemic Linguistics. Berlin: de Gruyter,
385-414.
Hoey, M. 1991b. Patterns of Lexis in Text. Oxford: Oxford University Press.
Hoey, M. 1994. "Signalling in discourse: a functional analysis of a common
discourse pattern in written and spoken English". In M. Coulthard (ed),
Advances in Written Text Analysis. London: Routledge, 26-45.
Hoover, M.L. 1997. "Effects of textual and cohesive structure of discourse
processing". Discourse Processes 23: 193-208.
Hopper, R. 1983. "Interpretation as coherence production". In R. Craig and K.
Tracy (eds), Conversational Coherence: Form, Structure, and Strategy;.
Hovy, E. 1990. "Parsimonious and profligate approaches to the question of
discourse structure relations". Proceedings of the 5th International
Workshop on Natural Language Generation : 128-136.
Isenberg, H. 1971. "Überlegungen zur Texttheorie". In J. Ihwe (ed),
Literaturwissenschaft und Linguistik. Frankfurt am Main: Athenäum,
155-172.
Bibliography 283
Jackson, S., Jacobs, S. and Rossi, A.M. 1986. Conversational Relevance:

Three Experiments on Pragmatic Connectedness in Conversation. Paper
presented at the International Communication Association Conference,
Chicago.
Jacobs, S. and Jackson, S. 1983. "Speech act structure in conversation: rational
aspects of pragmatic coherence". In R. Craig and K. Tracy (eds),
Conversational Coherence: Form, Structure, and Strategy. Beverly Hills:
Sage, 47-66.
Janos, M.E., Petöfi, S. and Sözer, E. (eds). 1989. Text and Discourse
Connectedness. Amsterdam: Benjamins.
Jarrett, D. 1984. "Pragmatic coherence in an oral formulaic tradition: I can read
your letters / sure can't read your mind". In D. Tannen (ed), Coherence in
Spoken and Written Discourse. Norwood: Ablex, 155-171.
Jefferson, G. 1978. "Sequential aspects of storytelling in conversation". In J.
Schenkein (ed), Studies in the Organization of Conversational
Interaction. New York: Academic Press, 219-248.
Jefferson, G. 1984. "On stepwise transition from talk about a trouble to
inappropriately next-positioned matters". In J.M. Atkinson and J. Heritage
(eds), Structures of Social Action: Studies in Conversation Analysis.
Johns, A.M. 1986. "Coherence and academic writing: some definitions and
suggestions for teaching". TESOL Quarterly 20: 227-265.
Johnson-Laird, P.N. 1983. Mental Models. Towards a Cognitive Science of
Language, Inference, and Consciousness. Cambridge: Cambridge
University Press.
Jucker, A.H. 1993. "The structure and coherence of discourse". In H. Löfller,
C. Grolimund and M. Gyger (eds), Dialoganalyse IV. Referate der 4.
Arbeitstagung Basel 1992. Tübingen: Niemeyer, 71-78.
Keenan, J.M., Baillet, S.D. and Brown, P. 1984. "The effects of causal cohesion
on comprehension and memory". Journal of Verbal Learning and Verbal
Behavior 23: 115-126.
Kiefer, F. (ed). 1985. Linking in Text. Konstanz: Universität Konstanz.
Kindt, W. and Wirrer, J. 1976. "Zu Teun van Dijks Aufsätzen 'Generative
Semantik und Texttheorie' und 'Textgenerierung und Textproduktion'".
In W. Kindt and S.J. Schmidt (eds), Interpretationsanalysen.
Argumentationsstrukturen in Literaturwissenschaftlichen Interpreta
tionen. München: Fink, 116-132.
Kintsch, W. 1995. "How readers construct situation models for stories: the role
of syntactic cues and causal inferences". In M. Gernsbacher and T. Givón
Kleifgen, J.A. and Saville-Troike, M. 1992. "Achieving coherence in

multilingual interaction". Discourse Processes 15: 183-200.
Knott, A. 1995. A Data-Driven Methodology for Motivating a Set of
Coherence Relations. Ph.D. Thesis, University of Edinburgh.
Knott, A. and Dale, R. 1994. "Using linguistic phenomena to motivate a set of
Knott, A. and Mellish, Chr. 1996. "A feature-based account of the relations
signalled by sentence and clause connectives". Language and Speech 39:
143-183.
Knott, A. and Sanders, T. 1998. "The classification of coherence relations and
their linguistic markers: an exploration of two languages". Journal of
Pragmatics 30: 135-175.
Korensky, J. and Hoffmannova, J. (eds). 1984. Text and the Pragmatic Aspects
of Language. Praha: Ustav pro jazyk cesky CSAV.
Korolija, N. and Linell, P. 1996. "Episodes: coding and analyzing coherence in
multi-party conversation". Linguistics 34: 799-831.
Korpimies, L. 1978. "Some aspects of discourse and cohesion". In J.-O.
Kotthoff, H. This volume. "Coherent keying in conversational humour.
Contextualising joint fictionalisation".
Kreuz, R.J. and Roberts, R.M. 1990. "The elements of conversational
coherence". In A. Graesser (ed), Conceptual Coherence in Text and
Discourse. Symposium conducted at the 12th annual conference of the
Cognitive Science Society, Cambridge, MA.
Kreyß, J. 1995. "Comprehension processes as a means for text generation". In
G. Rickheit and C. Habel (eds), Focus and Coherence in Discourse
Processing. Berlin: de Gruyter, 143-169.
Kroll, N.E. 1996. "Cohesion failure as a source of memory illusions". Journal
of Memory and Language 35: 176-197.
Kroon, C.H.M. 1995. Discourse Particles in Latin: A Study of NAM, ENIM,
AUTEM, VERO and AT. Amsterdam: J.C. Gieben.
Kuo, C. 1995. "Cohesion and coherence in academic writing: from lexical
choice to organization". A Journal of Language Teaching and Research
in Southeast Asia 26: 47-62.
Labov, W. and Fanshel, D. 1977. Therapeutic Discourse. New York: Academic
Press.
Langer, G. 1996. "Textkohärenz und Textspezifität. Textgrammatische
Untersuchungen zu den Gebrauchstextsorten Klappentext, Patienten
information, Garantieerklärung und Kochrezept". Fachsprache 18: 179-
182.
Bibliography 285
Langleben, M. 1981. "Latent coherence, contextual meanings, and the

interpretation of the text". Text 1: 279-313.
Lascarides, A., Copestake, A. and Briscoe, T. 1996. "Ambiguity and
coherence". Journal of Semantics 13: 41-65.
Lautamatti, L. 1978. "Some observations on cohesion and coherence in
simplified texts". In J.-O. Östman (ed), Cohesion and Semantics. Åbo:
Akademi, 165-181.
Lautamatti, L. 1990. "Coherence in spoken and written discourse". In U.
Connor and A.M. Johns (eds), Coherence in Writing. Research and
Pedagogical Perspectives. Alexandria: TESOL, 29-40.
Lenk, U. 1995. "Discourse markers and conversational coherence". In B.
Wårvik, S.-K. Tanskanen and R. Hiltungen (eds), Organization in
Turku, 341-352.
Lenk, U. 1998a. Marking Discourse Coherence: Functions of Discourse
Markers in Spoken English. Tübingen: Narr.
Lenk, U. 1998b. "Discourse markers and global coherence in conversation".
Lewis, D. 1979. "Scorekeeping in a language game". Journal of Philosophical
Logic 8: 339-359.
Linde, C. 1984. The Creation of Coherence in Life Stories. Norwood: Ablex.
Linde, C. 1993. Life Stories. The Creation of Coherence. Oxford: Oxford
University Press.
Linell, P. and Korolija, N. 1997. "Coherence in multi-party conversation:
episodes and contexts in interaction". In T. Givón (ed), Conversation:
Cognitive, Communicative, and Social Perspectives. Amsterdam:
Benjamins, 167-205.
Lintermann-Rygh, I. 1985. "Connector density - an indicator of essay quality?"
Text 5: 347-357.
Lohmann, P. 1988. "Connectedness of texts: a bibliographical survey, part I". In
J. Petöfi (ed), Text and Discourse Constitution. Berlin: de Gruyter, 478-
501.
Lohmann, P. 1989. "Connectedness of texts: a bibliographical survey, part II".
In W. Heydrich, F. Neubauer, J. Petöfi and E. Sözer (eds), Connexity and
Longacre, R.E. 1983. The Grammar of Discourse. New York: Plenum Press.
Lorenz, G. 1998. "Overstatement in advanced learners' writing: stylistic aspects
of adjective intensification". In S. Granger (ed), Learner English on
Computer. London: Addison Wesley Longman, 53-66.
Lorenz, G. 1999. Adjective Intensification - Learners versus Native Speakers:

A Corpus Study of Argumentative Writing. Amsterdam: Rodopi.
Lorenz, G. This volume. "Learning to cohere: causal links in native vs. non-
native argumentative writing".
Lundquist, L. 1985. "Coherence: from structure to processes". In E. Sözer (ed),
Text Connexity, Text Coherence. Aspects, Methods, Results. Hamburg:
Buske, 151-175.
MacNamara, D.S. 1996. "Learning from texts: effects of prior knowledge and
text coherence". Discourse Processes 22: 247-289.
Mann, W.C, Matthiessen, C.M.I.M. and Thompson, S.A. 1992. "Rhetorical
structure theory and text analysis". In W.C. Mann and S.A. Thompson
(eds), Discourse Description. Diverse Linguistic Analyses of a Fund-
Raising Text. Amsterdam: Benjamins, 39-79.
Mann, W.C. and Thompson, S.A. 1986. "Relational propositions in discourse".
Mann, W.C. and Thompson, S.A. 1987. "Rhetorical structure theory:
description and construction of text structures". In G. Kempen (ed),
Natural Language Generation. New Results in Artificial Intelligence,
Psychology and Linguistics. Dordrecht: Martinus Nijhoff, 85-95.
Mann, W.C. and Thompson, S.A. 1988. "Rhetorical structure theory: toward a
functional theory of text organization". Text 8: 243-281.
Marello, C. 1989. "Ellipsis between connexity and coherence". In M.-E. Conte,
J. Petöfi and E. Sözer (eds), Text and Discourse Connectedness.
Martin, J.R. 1992. English Text: Systems and Structure. Amsterdam:
Benjamins.
Maschler, Y. 1994. "Metalanguaging and discourse markers in bilingual
conversation". Language in Society 23: 325-366.
McCarthy, J.C, Wright, P.C. and Monk, A.F. 1992. "Coherence in text-based
electronic conferencing: coupling text and context". Journal of Language
and Social Psychology 11: 267-277.
McCown, R.R. and Miller, R.B. 1986. "Referential coherence and structural
hcight in text processing". American Educational Research Journal 23:
77-86.
McNeill, D. and Levy, E. 1993. "Cohesion and gesture". Discourse Processes
16: 363-386.
Meyer, B.J.F. 1975. The Organization of Prose and its Effects on Memory.
Amsterdam: North-Holland.
Bibliography 287
Milton, J. and Tsang, E. 1993. "A corpus-based study of logical connectors in

EFL students' writing". In R. Pemberton and E. Tsang (eds), Studies in
Lexis. Hong Kong: HKUST, 215-246.
Moeschler, J. 1989. "Pragmatic connectives, argumentative coherence and
relevance". Argumentation 3: 321-339.
Moeschler, J. 1993. "Relevance and conversation". Lingua 90: 149-171.
Moore, J.D. and Pollack, M.E. 1992. "A problem for RST: The need for multi
level discourse analysis". Computational Linguistics 18: 537-544.
Morris, J. and Hirst, G. 1991. "Lexical cohesion computed by thesaural
relations as an indicator of the structure of text". Computational
Mosegaard Hansen, M.-B. and Skytte, G. (eds). 1996. Le Discours: Cohérence
et Connexion. Copenhagen.
Mulder, W. de, Tasmowsky de Ryck, L. and Vetters, C. (eds). 1996. Anaphores
temporelles et (in-)coherence. Amsterdam: Rodopi.
Mulder, W. de, Tasmowsky, L. (eds). 1996. Coherence and Anaphora.
(Special issue of Belgian Journal of Linguistics 10). Amsterdam:
Benjamins.
Neubauer, F. (ed). 1983. Coherence in Natural-Language Texts. Hamburg:
Buske.
Newman, J.E. 1985. "Processing spoken discourse: effects of position and
emphasis on judgements of textual coherence". Discourse Processes 8:
205-227.
Nofsinger, R.E. 1983. "Tactical coherence in courtroom conversation". In R.
Norgard-Sørensen, J. 1992. Coherence theory: the case of russian. Berlin: de
Gruyter.
O'Brien, T. 1995. "Rhetorical structure analysis and the case of the inaccurate,
incoherent source-hopper". Applied Linguistics 16: 442-482.
Ochs Keenan, E. and Klein, E. 1975. "Coherency in children's discourse".
Journal of Psycholinguistic Research 4: 365-386.
Ochs Keenan, E. and Schieffelin, B. 1976. "Topic as a discourse notion: a study
of topic in the conversations of children and adults". In C. Li (ed), Subject
and Topic. New York: Academic Press, 335-383.
Östman, J.-O. 1978. "Introduction: text, cohesion, and coherence". In J.-O.
Östman, J.-O. 1979. "On the pragmatic meaning of textual particles". In B.
Hammarberg (ed), Kontrastiv Lingvistik och Sekundärspräksforskning.
Stockholm: Stockholms Universitet, 171-191.
Östman, J.-O. 1982. "Pragmatic particles in an applied perspective".

Neuphilologische Mitteilungen 83: 135-153.
Östman, J.-O. 1995. "Pragmatic particles twenty years after". In B. Wårvik, S.-
K. Tanskanen and R. Hiltungen (eds), Organization in Discourse.
Proceedings from the Turku Conference. Turku: University of Turku, 95-
108.
Östman, J.-O. This volume. "Coherence through understanding through
discourse patterns: focus on news reports".
Östman, J.-O. (ed). 1978. Cohesion and Semantics. Åbo: Akademi.
Pander Maat, H. 1994. "Coherentie-relaties hebben zowel semantische als
pragmatische aspecten". In A. Maes, P. van Hauwermeiren and L. van
Waes (eds), Perspectieven in Taalbeheersingsonderzoek. Dordrecht:
ICG, 118-129.
Pander Maat, H. 1998. "Classifying negative coherence relations on the basis of
linguistic evidence". Journal of Pragmatics 30: 177-204.
Pander Maat, H. and Sanders, T. 1995. "Nederlandse causale connectieven en
het onderscheid tussen inhoudelijke en epistemische coherentie-relaties".
Pappas, C.C. 1985. "The cohesive harmony and cohesive density of children's
oral and written stories". In J.D. Benson and W.S, Greaves (eds),
Systemic Perspectives on Discourse: Selected Applied Papers from the 9th
International Systemic Workshop. Norwood: Ablex, 169-186.
Parret, H. 1989. "Paraphrase as a coherence principle in conversation". In M -
E. Conte, J. Petöfi and E. Sözer (eds), Text and Discourse
Connectedness. Amsterdam: Benjamins, 281-289.
Parsons, G. 1990. Cohesion and Coherence: Scientific Texts. Nottingham:
University of Nottingham.
Passonneau, R.-J. and Litman, D.-J. 1996. "Empirical analysis of three
dimensions of spoken discourse: segmentation, coherence, and linguistic
devices". In E.H. Hove and D.R. Scott (eds), Computational and
Conversational Discourse: Burning Issues - an Interdisciplinary
Account. Berlin: Springer, 161-194.
Penman, R. 1987. "Discourse in courts: cooperation, coercion, and coherence".
Petöfi, J. 1979. Text vs. Sentence. Basic Questions of Text Linguistics, Vol. 2.
Hamburg: Buske.
Petöfi, J. (ed). 1986. Text Connectedness from Psychological Point of View.
Hamburg: Buske.
Petöfi, J. (ed). 1988. Text and Discourse Constitution. Berlin: de Gruyter.
Bibliography 289
Petöfi, J. and Sözer, E. (eds). 1983. Micro and Macro Connexity of Texts.
Hamburg: Buske.
Pilus, Z. 1996. "Coherence and students' errors. Weaving the threads of
discourse". English Teaching Forum 34: 44-49.
Pit, M. 1997. "Subjective and objective construal in coherence relations". In J.
Don and T. Sanders (eds), Utrecht Institute for Linguistics. Yearbook
1997. Utrecht: Utrecht Institute for Linguistics, 45-70.
Polanyi, L. 1976. "The role of redundancy in cohesion and evaluative
functioning in narrative". Rackham Literary Studies 7: 19-38.
Polanyi, L. 1985. The Structure of Discourse. Norwood: Ablex.
Ragan, S. 1983. "Alignment and conversational coherence". In R. Craig and K.
Rambow, O. (ed). 1993. Intentionality and Structure in Discourse Relations.
Proceedings of the ACL-Workshop, Ohio State University, Columbus,
Ohio.
Randquist, M. 1985. "The barely visible glue - some aspects of textual
connectedness". In E. Sözer (ed), Text Connexity, Text Coherence.
Aspects, Methods, Results. Hamburg: Buske, 189-218.
Redeker, G. 1990. "Ideational and pragmatic markers of discourse structure".
Redeker, G. 1991. "Linguistic markers of discourse structure". Linguistics 29:
1139-1172.
Redeker, G. (forthcoming). "Coherence and structure in text and discourse". In
W. Black and H. Bunt (eds), Computational Pragmatics: Abduction,
Belief and Context. London: University College Press.
Reichman, R. 1978. "Conversational coherency". Cognitive Science 3: 283-327.
Reinhart, T. 1980. "Conditions for text coherence". Poetics Today 1: 161-180.
Rentel, V.M., King, M.L., Pappas, C.C. and Pettegrew, B.S. 1983. A
Longitudinal Study of Coherence in Children's Written Narratives. Final
Report to National Institute of Education. Columbus: The Ohio State
University Research Foundation.
Ribeiro, B.T. 1994. Coherence in Psychotic Discourse. Oxford: Oxford
University Press.
Rickheit, G. and Strohner, H. 1991. "Towards a cognitive theory of linguistic
coherence". Theoretical Linguistics 18: 209-237.
Rickheit, G , Günther, U. and Sichelschmidt, L. 1992. "Coherence and
coordination in written text: reading time studies". In D. Stein (ed),
Cooperating with Written Texts. The Pragmatics and Comprehension of
Written Texts. Berlin: de Gruyter, 103-127.
Rickheit, G., Sichelschmidt, L. and Strohner, H. 1995. "Economical principles

in coherence management: a cognitive systems approach". In G. Rickheit
and C. Habel (eds), Focus and Coherence in Discourse Processing.
Berlin: de Gruyter, 170-189.
Rickheit, G. (ed). 1991. Kohärenzprozesse. Modellierung von
Sprachverarheitung in Texten und Diskursen. Opladen: Westdeutscher
Verlag.
Rickheit, G. and Habel, C. (eds). 1995. Focus and Coherence in Discourse
Processing. Berlin: de Gruyter.
Riley, Ph. 1980. "When communication breaks down: levels of coherence in
discourse". Applied Linguistics 1: 201-216.
Roberts, R.M. and Kreuz, R.J. 1993. "Nonstandard discourse and its
coherence". Discourse Processes 16: 451-464.
Rochester, S. and Martin, JR. 1979. Crazy Talk: a Study of the Discourse of
Schizophrenic Speakers. New York: Plenum Press.
Roncador, M. von and Bublitz, W. 1979. "Abschweifungen". In H. Weydt (ed),
Die Partikeln der Deutschen Sprache. Berlin: de Gruyter, 285-298.
Rosch, E. 1988. "Coherences and categorization: a historical view". In F.S.
Kessel (ed), The Development of Language and Language Researchers:
Essays in Honor of Roger Brown. Hillsdale: Erlbaum, 373-392.
Rudolph, E. 1989. "The role of conjunctions and particles for text
connectivity". In M.E. Janos, S. Petöfi and E. Sözer (eds), Text and
Discourse Connectedness. Amsterdam: Benjamins, 175-190.
Rugaleva, A. 1983. "On typology of lexical cohesion in discourse". LACUS 9,
1982: 357-366.
Samet, J. and Schank, R. 1984. "Coherence and connectivity". Linguistics and
Philosophy 7: 57-82.
Sanders, R. 1983. "Tools for cohering discourse and their strategic utilization:
markers of structural connections and meaning relations". In R. Craig and
K. Tracy (eds), Conversational Coherence: Form, Structure, and
Strategy. Beverly Hills: Sage, 67-80.
Sanders, T. 1997. "Semantic and pragmatic sources of coherence: On the
categorization of coherence relations in context". Discourse Processes
24: 117-145.
Sanders, T. and Spooren, W. This volume. "Communicative intentions and
coherence relations".
Sanders, T., Spooren, W. and Noordman, L. 1992. "Toward a taxonomy of
Bibliography 291
Sanders, T., Spooren, W. and Noordman, L. 1993. "Coherence relations in a

cognitive theory of discourse representation". Cognitive Linguistics 4: 93-
133.
Sanford, A., Barton, S., Moxey, L. and Paterson, K. 1995. "Cohesion
processes, coherence, and anomaly detection". In G. Rickheit and Chr.
Habel (eds), Focus and Coherence in Discourse Processing. Berlin: de
Gruyter, 203-213.
Sanford, A. and Moxey, L. 1995. "Aspects of coherence in written language: a
psychological perspective". In M. Gernsbacher and T. Givón (eds),
Coherence in Spontaneous Text. Amsterdam: Benjamins, 161-188.
Sarangi, S. and Slembrouck, S. 1992. "Non-cooperation in conversation: A
reassessment of Gricean pragmatics". Journal of Pragmatics 17: 117-154.
Schade, U., Langer, H., Rutz, H. and Sichelschmidt, L. 1991. "Kohärenz als
Prozeß". In G. Rickheit (ed), Kohärenzprozesse. Modellierung von
Sprachverarbeitung in Texten und Diskursen. Opladen: Westdeutscher
Verlag, 7-58.
Schegloff, E. 1990. "On the organization of sequences as a source of
'coherence' in talk-in-interaction". In B. Dorval (ed), Conversational
Schiffrin, D. 1982. Cohesion in Discourse: the Role of Non-Adjacent
Paraphrase. Austin: Southwest Educational Development Laboratory.
Schiffrin, D. 1985. "Conversational coherence: the role of 'well'". Language
61:640-667.
Schiffrin, D. 1986. Coherence in Spoken and Written Discourse. Norwood:
Ablex.
Schiffrin, D. 1987. Discourse Markers. Cambridge: Cambridge University
Press.
Schleppegrell, M. 1996. "Strategies for discourse cohesion. Because in ESL
writing". Functions of Language 3: 235-254.
Schmitt-Ackermann, S. 1996. Kohärenz in Redewiedergaben. Heidelberg:
Groos.
Seidlhofer, B. 1986. "Cohesion in Austrian learners' English". In O.
Rauchbauer (ed), A Yearbook of Studies in English Language and
Literature 1985/86. Festschrift für Siegfried Korninger. Wien:
Braumüller, 221-236.
Seidlhofer, B. 1995. Approaches to Summarization. Discourse Analysis and
Language Education. Tübingen: Narr.
Seidlhofer, B. and Widdowson, H. This volume. "Coherence in summary: the
contexts of appropriate discourse".
Sgall, P. 1983. "On one aspect of textual coherence". In F. Neubauer (ed),

Coherence in Natural-Language Texts. Hamburg: Buske, 135-143.
Shiro, M. 1994. "Inferences in discourse comprehension". In M. Coulthard (ed),
Advances in Written Text Analysis. London: Routledge, 167-178.
Smith, J. and Elkins, J. 1992. "Coherence and the sharing of meaning:
supporting underachieving readers". International Journal of Disability,
Development and Education 39: 239-249.
Souza, F. de. 1985. "Dialogue breakdown". In M. Dascai (ed), Dialogue: an
Interdisciplinary Approach. Amsterdam: Benjamins, 415-426.
Sözer, E. (ed). 1985. Text Connexity, Text Coherence. Aspects, Methods,
Results. Hamburg: Buske.
Spooren, W. 1997. "The processing of underspecified coherence relations".
Discourse Processes 24: 149ff.
Spooren, W., Tates, T. and Sanders, T. 1995. "Coherentierelaties en
connectieven bij basisschoolkinderen". Nederlandse Taalkunde 1: 26-51.
Sternberg, M. 1981. "Ordering the unordered: time, space, and descriptive
coherence". Yale French Studies 61: 60-88.
Stoddard, S. 1991. Text and Texture: Patterns of Cohesion. Norwood: Ablex.
Stubbs, M. (forthcoming). "Computer-assisted text and corpus analysis:
collocation, cohesion and communicative competence". In D. Tannen et
al. (eds), The Handbook of Discourse Analysis. Oxford: Blackwell.
Talbot, M.M. 1996. "'Randy fish boss branded a stinker': coherence and the
construction of masculinities in a British tabloid newspaper". In S.
Johnson and U. Meinhof (eds), Language and Masculinity. Oxford:
Blackwell, 173-187.
Tannen, D. 1990a. "Gender differences in topical coherence: creating
involvement in best friends' talk". Discourse Processes 13: 73-90.
Tannen, D. 1990b. "Gender differences in conversational coherence: physical
alignment and topical cohesion". In B. Dorval (ed), Conversational
Tannen, D. (ed). 1984. Coherence in Spoken and Written Discourse. Norwood:
Ablex.
Tannen, D. (ed). 1988. Linguistics in Context: Connection Observation and
Understanding. Norwood: Ablex.
Tanskanen, S.-K. 1994a. "Variation in lexical cohesion in spoken and written
English". In S. Cmejrková, F. Danes and E. Havlová (eds), Writing versus
Speaking. Tübingen: Narr, 237-242.
Tanskanen, S.-K. 1994b. "Continuity and interaction: aspects of lexical
cohesion in conversations". In S.-K. Tanskanen and B. Wårvik (eds),
Bibliography 293
Topics and Comments. Papers from the Discourse Project. Turku:

Department of English, University of Turku, 99-114.
Tanskanen, S.-K. 1995. "Lexical cohesion meets communicative conditions". In
B, Wårvik, S.-K. Tanskanen and R. Hiltungen (eds), Organization in
Turku, 531-538.
Tasmowski, L. and Mulder, W. de (eds). 1996. Belgian Journal of Linguistics
10 (special issue on Coherence and Anaphora).
Taylor, A. 1994. "The role of repetition in perceptions of discourse coherence".
Thompson, S. 1994. "Aspects of cohesion in monologue". Applied Linguistics
15: 58-75.
Tierney, R. and Mosenthai, J. 1983. "Cohesion and textual coherence".
Research in the Teaching of English 17: 215-229.
Tomlin, R. (ed). 1987. Coherence and Grounding in Discourse: Outcome of a
Symposium, Eugene, Oregon, June 1984. Amsterdam: Benjamins.
Trabasso, T. and Suh, S. 1993. "Understanding text: achieving explanatory
coherence through on-line inferences and mental operations in working
memory". Discourse Processes 16: 3-34.
Trabasso, T., Secco, T. and van den Broek, P. 1984. "Causal cohesion and
story coherence". In H. Mandl, N.L. Stein and T. Trabasso (eds),
Learning and Comprehension of Text. Hillsdale: Erlbaum, 83-111.
Trabasso, T., Suh, S. and Payton, P. 1995. "Explanatory coherence in
understanding and talking about events". In M. Gernsbacher and T. Givón
Tracy, K. 1981. The Issue-Event Distinction as an Aspect of Conversational
Coherence: a Rule and its Scope Conditions. Madison: University of
Madison.
Traxler, M.J. 1995. "Improving coherence in written communication". In M.
Gernsbacher and T. Givón (eds), Coherence in Spontaneous Text.
Tyler, A. 1994. "The role of repetition in perceptions of discourse coherence".
Tyler, A. 1995. "Patterns of lexis: how much can repetition tell us about
discourse coherence?" In J.E. Alatis et al. (eds), Georgetown University
Round Table on Language and Linguistics 1995. Washington:
Georgetown University Press, 268-280.
Vater, H. 1992. Einführung in die Textlinguistik. München: Fink.
Vauras, M , Hyona, J. and Niemi, P. 1992. "Comprehending coherent and

incoherent texts: evidence from eye movement patterns and recall
performance". Journal of Research in Reading 15: 39-54.
Velde, R. van de. 1981. Interpretation, Kohärenz und Inferenz. Hamburg:
Buske.
Velde, R. van de. 1985. "Inferences and coherence in text interpretation". In E.
Sözer (ed), Text Connexity, Text Coherence. Aspects, Methods, Results.
Hamburg: Buske, 261-298.
Velde, R. van de. 1989. "Man, verbal text, inferencing, and coherence". In W.
Ventola, E. 1987. The Structure of Social Interaction. London: Pinter.
Ventola, E. 1996. "Packing and unpacking of information in academic texts". In
E. Ventola and A. Mauranen (eds), Academic Writing. Intercultural and
Textual Issues. Amsterdam: Benjamins, 153-194.
Ventola, E. This volume. "Semiotic spanning at conferences: cohesion and
coherence in and across conference papers and their discussions".
Ventola, E. and Mauranen, A. 1991. "Non-native writing and native revising of
scientific articles". In E. Ventola (ed), Functional and Systemic
Linguistics. Approaches and Uses. Berlin: de Gruyter, 457-492.
Viehweger, D. 1989. "Coherence - interaction of modules". In W. Heydrich, F.
Neubauer, J. Petöfi and E. Sözer (eds), Connexity and Coherence:
Analysis of Text and Discourse. Berlin: de Gruyter, 149-167.
Vuchinich, S. 1977. "Elements of cohesion between turns in ordinary
conversation". Semiotica 20: 229-257.
Wennerstrom, A. 1998. "Intonation as Cohesion in Academic Discourse: A
Study of Chinese Speakers of English". Studies in Second Language
Acquisition 20.
Werth, P. 1984. Focus, Coherence and Emphasis. London: Croom Helm.
Widdowson, H. (ed). 1978. Teaching Language as Communication. Oxford:
Oxford University Press.
Wilkes-Gibbs, D. 1995. "Coherence in collaboration: some examples from
conversation". In M. Gernsbacher and T. Givón (eds), Coherence in
Spontaneous Text. Amsterdam: Benjamins, 239-267.
Winterowd, W.R. 1970. "The grammar of coherence". College English 31:
828-835.
Witte, S. and Faigley, L. 1981. "Coherence, cohesion, and writing quality".
College Composition and Communication 32: 189-204.
Wolfson, N. 1982. The Conversational Historical Present in American English
Narrative. Dordrecht: Foris.
Bibliography 295
Zabrucky, K. 1986. "The role of factual coherence in discourse

comprehension". Discourse Processes 9: 197-220.
Zammuner, V. 1981. Speech Production. Strategies in Discourse Planning: a
Theoretical and Empirical Enquiry. Hamburg: Buske.
Zammuner, V. 1986. "Children's coherence in discourse: a review". In J. Petöfi
(ed), Text Connectedness from Psychological Point of View. Hamburg:
Buske, 24-63.
Ziff, P. 1984. "Coherence". Linguistics and Philosophy 7: 31-42.
Index 169f; 196; 201; means of securing
~ 157; 159f; 162; negotiation of ~
-A- 3; 6; 121; 181; 183f; 186; 189;
212; over- ~ 160; partial ~ 3; 5; 6;
academic styles 88f 153; 155; - relations 235-247;
account 153; 162; 212-218; 239; 241 source of ~ 241; topical ~ 3; 35;
action 167; 170; 288 180; 187; 189; 195; 197; 231
action game 181 cohesion: ~ devices 7; 56; 198; 200f;
activity type 190; 192; 197; 200; 204 207; means of- 104; 109; 159;
adjacency pair, question-answer 4; 6; 165; dialogic - 109; - cues 235; -
7; 37; 45; 51; 163; 187; 192f; 198f; ties 55f; 222; 23 1f
226 collaboration 36ff; 50ff; 172
ambiguity 175; 178; 181f; 183; 186; comical discourse 128
208; 262 communication breakdown 184f
animation (in reported speech) 12-30 competence, communicative 71f; 106;
appropriation 211 ; 218 217; 261ff
argumentative: ~ style 4; 7; 18; 62; 71; comprehension 160; 175; 181; 219
~ text 96; ~ writing 55; 57; 63; 70 conceptualisation 72f; 81f; 86; 98
avoidance technique 118 context: linguistic - 2; 183ff; ~ of
awareness 218 ; 281ff situation 2; 5; 22; 101; 105; 109f;
114; 116; 119; 127-145; 176; 177;
179 180;190; 200; 209; 211; 245f;
-C-
254
contextualisation cues 11; 17; 125;
casual speech 71 127ff; 145
causal connectors/ ~ labelling/ ~ contrastive rhetoric 56
markers 4; 56-73; 241 conventions 187; 202; 254
clarification, quest for 156; 157; 164; conversation analysis 4; 6; 11f; 35f;
165; 181; 198; 226
38; 125f; 130f; 148ff; 173ff
Clinton, B. 77f
conversational goal 169
code-switching 11; 125
conversational humour 5; 125-145;
cognitive account of coherence 235ff
164
cognitive domain 88
conversational maxim 191
cognitive schemata 83; 97
cooperation (*cooperate, *cooperative,
coherence: - approximate 155; ~ as a
*cooperativeness, cooperation) 3;
process 2; ~ as a relational concept
49f; 154; 157; 162f; 166; 168; 178;
7; ~ as a scalar notion 3; 5; 6;
183; 191; 200; 209; 260
153ff; 172;~breakdown 30, 153; Cooperative Principle 50; 157; 166;
164; 181; Coherence Principle, see 208ff
also default principle of coherence; Corpora: Learner ~ 57-73; Native
~ contextual 106; 109; 125f; 129; speaker ~ 57-74; ~ of spoken
145; 156; 190; 206; 209f; 262; discourse 5
254f; ~ disturbed 5; 153-172; 189; critical discourse analysis 251ff; 256
~global 98; 156; 167f; 170; 196; cross-cultural comparisons 261
226; 230; ~ impairment 161ff; - in
interaction 12; local ~ 155; 166f;
298 Index
-D- goals: communicative ~ 200; social ~

197; 200; 201
default principle of coherence 155- Grice, H.P. 157; 166; 200
158; 162; 172
deixis: shifted ~ 15; vocal ~ 14-17 -H-
dialogue 178; 203; 221
digression 158; 174 hearer-knows-best principle 155; 161;
discourse: ~ analysis 1; 4; 5; 7; 35; 77; 253f
92; 206; 251-263; ~ intention 239;
246; ~ interpretation 2; 5; 6; 14; -I-
96; 127; 154; 169; 175f; 178ff;
182; 208ff; 216; 228; 231; 252; Idealized Cognitive Model (ICM) 83
255; 258;~marker 16; 63; 158; ideational / experiential 118; 121
166; 181; 185;~purpose 236; ideational meaning 3; 154
theory of- 7; 127; 236f; 241f; idiomaticity 70ff;217
255-263; ~ topic 4; ~ universals illocutionary: ~ force 3; 154; ~
260; 261; universe of- 197; ~ relations 242-247
worlds 259; 26Iff incoherence: global ~ 170; local - 3; 5;
discourse pattern 4; 77-101; Contact 155; 161; 168; 170; 172; 202
Ad pattern 86f; Death Notice intentional relation 2; 7; 55
pattern 84f; news pattern 4; 77- interaction 12f; 17; 37; 125; 127; 170;
101; Recipe pattern 83f; 90f 179; 182; 192; 197; 201; 206f;
253; 260; dynamics of ~ 111;~
-F- modality 126
interlanguage 7If
footing: changes of ~ 11; shifts of- 4; international English 72
11; 12 intertextuality 5; 97; lOlf, 109f; 112;
formal writing 61; 63f; 71; 113; 122 120f; 192
frame 97; 99; 128; 130; 145
frame break 5; 153; 163; 170ff -J-
frame of interpretation 129
framing: vocal - 34; ~ procedures 136; jokes 125f; 128; 131; 136
prosodic ~ 18f; 23; 27;
paralinguistic - 18f -K-
Kaplan, R. 56; 88
keying 5; 126-145
Galtung, J. 88f
gambits 165 -L-
generic structures 111; 125
generic unity 111
Lakoff, G. 83
genre 37; 77-81; 85ff; 91, 96; 98; 100;
laughter 125-145
lOlf; 111f;119-122; 126; 129; laughter particles 17; 27; 32; 126; 133;
135; 145; 155; 165
137; 139; 141; 143
learner language 61; 72
Index 299
learners' strategies 4; 57; 64ff; 146 prosodic: ~ animation 12; 23; ~ cues 4;
lexical field 66 11-30; 125; 127; 175;~ effects 14;
lexico-grammatical deficits 56 ~ signals 18; ~ variation 3
linguistic maturation/ maturity 57; 59f punchline 125; 127f; 139; 133; 136;
logical connectors 56f; 200 139; 145
-M- -Q-
metapragmatic 27; ~ function 126; 147 questions 4; 37-54; 123; 163f; 194-200
metatext, means and functions 199;
201 -R-
mind 216; 252f; 263
misunderstanding 97; 175ff reference 5; 116; 121;125; 154; 165ff;
modulation 120 172; 189
multi-level thesis 236-245 register 6 57; 62; 67; 69; 71f; 106;
llff; 119; 153; 173
-N- register break 5; 154; 172
relations: classification of ~ 240f;
Natural Language Generation 236 logical ~ 55; 57; propositional ~
negotiation of topic, see negotiation of 242; 247; rhetorical ~ 236f
coherence relevance 11; 47; 118f; 127; 129; 236;
nominal style 64 242
non-coherence, see incoherence repair 6; 11-30; 176; 181; 183ff
non-native discourse 4; 55-73; 101; reported speech 4; 11-30
103 Rhetorical Structure Theory 236
non-nativeness, see non-native rhythm 19ff ; 167
discourse routines 254
-O- -S-
open-choice use 64; 69 schizophrenic discourse 160f; 172

order of Segments 4; 13; 19; 72; 91; self-presentations 119; 134f
123; 141 semantic prosody 67; 72
over-extension 60 semiosis 5; 102; 113; 122
semiotic (spheres of) knowledge 109
-P- semiotic modalities 102; 112f
shared knowledge 55; 115; 145
paraphrasing strategies 58; 60; 62; 238 side-track 119
participant 36; 39; 47; 101-123; 183f; simplification 61
200 speaker meaning 260; 263
participant roles 11; 13; 17; 18; 19; spontaneous speech 37
215-218 stylistic maturity 4 57; 59f; 69ff; 195
phrasal building-block 69 summarising, summary 43; 91; 97;
polarity 240 211-218
pragmatic purpose 17; 229
300 Index
-T-
text type 77; 85ff; 96; ~ narration 2; 5;

78ff; 82; 85
theory of discourse coherence, see also
discourse, theory of: intentional
approach 235-247; relational
approach 235-247
topic: ~ break 95; 170;~change 5; 36;
43; 153; 168-172; 192; 194; 201;
closing of ~ 39; 42; ~ discontinuity
201;~drift 5; 153; 166f; -
elicitation 4; 38ff; 44; 50f; - flow
37; 48; global ~ 107; 168-171;~
introduction 37 f; 46; 48f; -
management 4; 53; - negotiation
37ff; 44f; 47, 52; ~ organisation
35f; ~ progression 2; 230; -
proposition 38; 41f; 44, 51f; - shift
38f;41;43; 168; 185;
superordinate ~ 158; 168-171; 196
transitivity 66
transitivity structure 114; 120
-U-
understanding 6; 84; 90f; 97; 121f;

153ff; 176; 181ff; 221; 258
-V-
visual mode 119-122
-W-
wild-card use 60ff; 67; 69
-Y-
Yeltsin, B. 77

Coherence

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Coherence

Hochgeladen von

Copyright:

Verfügbare Formate

COHERENCE IN SPOKEN

AND WRITTEN DISCOURSE

JOHN BENJAMINS PUBLISHING COMPANY

Library of Congress Cataloging-in-Publication Data

Introduction: Views of Coherence 1

Part I: How to (Re-)Create Coherence: Means of Coherence

Coherent Voicing: On Prosody in Conversational Reported Speech 11

It Takes Two to Cohere: The Collaborative Dimension 35

Learning to Cohere: Causal Links in Native vs. 55

Coherence through Understanding through Discourse Patterns: 77

Semiotic Spanning at Conferences: Cohesion and Coherence 101

Coherent Keying in Conversational Humour: 125

Part II: How to Negotiate Coherence: Degrees of Coherence

Coherence and Misunderstanding in Everyday Conversations 175

The Effect of Context in the Definition and Negotiation of Coherence 189

Coherence in Summary: The Contexts of Appropriate Discourse 205

Coherence in Hypertext 221

Part III: How to Describe Coherence: Views of Coherence

If Coherence Is Achieved, Then Where Doth Meaning Lie? 251

A Bibliography of Coherence and Cohesion 267

Wolfram Bublitz is Professor of English Linguistics at the University of

Anna Ciliberti is Professor of Applied Linguistics at the University for

Elizabeth Couper-Kuhlen is Professor of English Linguistics at the University

Rossana Damiano is currently a Ph.D. student in Computer Science at the

Willis Edmondson is Professor of Foreign Language Learning Research at the

Gerd Fritz is Professor of German Linguistics at the University of Gießen,

Ronald Geluykens is Professor of English Linguistics at the University of

Helga Kotthoff is full-time researcher in Linguistics at the University of

Uta Lenk teaches Linguistics in the English Department of the University of

Gunter Lorenz teaches Linguistics and Foreign Language Methodology in the

Jan-Ola Östman is Professor of Linguistics at the English Department and at

Ted Sanders is Assistant Professor at the Utrecht Institute of Linguistics and

Barbara Seidlhofer is Associate Professor at the English Department of the

1994) and Approaches to Summarization. Discourse Analysis and Language

Wilbert Spooren is Assistant Professor at the Discourse Studies Group,

Eija Ventola is Professor of English Linguistics at the Martin-Luther-

Henry Widdowson holds professorial appointments in Language Education

Our aim in putting together this collection of papers is to help relieve

management, meaning and other kinds of semiotic relations and discourse

traditional concepts of cohesion and coherence are bound to individual texts

at one frequent cause of disturbed coherence: conversational mis­

in the hierarchical structure of the text. Construing coherence across the

Sinclair, J. 1991. Corpus, Concordance, Collocation. Oxford: Oxford University

Augsburg August 1998

Coherence should be findable for everything that is a demonstrably relevant aspect

exactly which tasks the 'voicing' of reported speech confronts

1 Coherence as a conversationalist's practice and an analyst's

2 Three types of 'trouble' in reported speech sequences

interaction do not always explicitly introduce different 'voices' with reporting

Given the deictic nature of prosodic and paralinguistic phenomena,

?f2 ('n) they're gonna die.

In recounting how he 'nailed' his interlocutor on the political

3 Prosodie and paralinguistic framing in 'successful' reported

3.1 Vocalframing of an 'other voice '

If 'trouble' in reported speech sequences involves problematic answers to the

1 J: we had a professor from Carleton;

25 f2 and as long as those things happened

so as to come at approximately equal intervals in time, i.e. they are perceptually

1 J: They just decided what.

J: fl that [<<ƒ> /tGALI/LEO:: <scanded,

Janet's anecdote is occasioned by Steve's prior remark even the Vatican

3.2 Vocal framing of several 'other voices '

Wanda's anecdote involves three figures, two of which are animated:

at one frequent cause of disturbed coherence: conversational mis