Emotion Talk Across

Corpora

Monika Bednarek
 Emotion Talk Across Corpora

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
 Also by Monika Bednarek
EVALUATION IN MEDIA DISCOURSE

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
 Corpora
Monika Bednarek
Emotion Talk Across

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
 © Monika Bednarek 2008
Foreword © J.R. Martin 2008
All rights reserved. No reproduction, copy or transmission of this
publication may be made without written permission.
No paragraph of this publication may be reproduced, copied or transmitted

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

save with written permission or in accordance with the provisions of the
Copyright, Designs and Patents Act 1988, or under the terms of any licence
permitting limited copying issued by the Copyright Licensing Agency, 90
Tottenham Court Road, London W1T 4LP.
Any person who does any unauthorized act in relation to this publication
may be liable to criminal prosecution and civil claims for damages.
The author has asserted her right to be identified as the author of this work
in accordance with the Copyright, Designs and Patents Act 1988.
First published 2008 by
PALGRAVE MACMILLAN
Houndmills, Basingstoke, Hampshire RG21 6XS and
175 Fifth Avenue, New York, N.Y. 10010
Companies and representatives throughout the world
PALGRAVE MACMILLAN is the global academic imprint of the Palgrave
Macmillan division of St. Martin’s Press, LLC and of Palgrave Macmillan Ltd.
Macmillan® is a registered trademark in the United States,
United Kingdom and other countries. Palgrave is a registered trademark in
the European Union and other countries.
ISBN-13: 978 0 230 55146 6 hardback
ISBN-10: 0 230 55146 7 hardback
This book is printed on paper suitable for recycling and made from fully
managed and sustained forest sources. Logging, pulping and manufacturing
processes are expected to conform to the environmental regulations of the
country of origin.
A catalogue record for this book is available from the British Library.
Library of Congress Cataloging-in-Publication Data
Bednarek, Monika, 1977
Emotion talk across corpora / Monika Bednarek.
p. cm.
Includes bibliographical references and index.
ISBN-13: 978 0 230 55146 6 (hardback : alk. paper)
ISBN-10: 0 230 55146 7 (hardback : alk. paper) 1. Language and
emotions. I. Title.
BF582.B43 2008
152.4 dc22 2008009902
10 9 8 7 6 5 4 3 2 1
17 16 15 14 13 12 11 10 09 08

Printed and bound in Great Britain by

CPI Antony Rowe, Chippenham and Eastbourne

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

Contents

Foreword by J.R. Martin vi

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Acknowledgements vii
List of Figures, Tables and Image ix
System Network Notation xii

1 Analyzing Language and Emotion 1

2 Emotion Profiling 27
3 A Local Grammar of Affect 65
4 Patterns of Affect Across Corpora 100
5 Mapping and Analyzing Affect 142
6 Enacting Affect: Pragmatic Analysis 183
References 224
Index 239

Additional data and discussion, complementary to the material in the

book and referred to occasionally in the text, appears on the author’s
website at www.MonikaBednarek.com

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

Foreword

It is of course one of the great pleasures of academic life to watch a new

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

generation of scholars emerging to build on and develop the work that
has gone before. In the case of Monika Bednarek we have the pleasure
of enjoying a scholar working at the interface of two complementary
perspectives on language, corpus linguistics and systemic functional
linguistic theory, and developing new insights into the language of eval-
uation at the edge of knowledge in both these interacting domains. The
challenge here is on the one hand to enrich the counting which corpus
linguists undertake, and on the other to quantify the qualitative dis-
course analyses in which systemic functional linguists engage. Bednarek
rises effectively to this challenge, paving the way for a generation of
transdisciplinary rapprochement across these two fields.
Initially trained in Germany, Bednarek has pursued her doctoral and
post-doctoral studies as a visiting scholar at both the University of Birm-
ingham, where she worked with Susan Hunston and Peter White, and
later here at the University of Sydney, where she has been an active
member of our functional linguistic research community. Her generous
mentoring, collaborative style and zest for intellectual challenge have
been a boon to all concerned. And we are delighted to see her research
come to fruition in this, her second monograph, which for the second
time establishes her as an inspiring intellectual leader as far as text based
work on language and evaluation is concerned.
I am honoured to have been asked to write the foreword for this book,
and commend it to all readers, who will be treated to a voice they will
be enjoying many times in the future as her career unfolds.

Prof J.R. Martin

Sydney
August 2007

vi

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

Acknowledgements

A number of people and institutions supported me in writing this book,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

and I would like to thank them here:
Most importantly, I wish to say thanks to my friends and family for
their friendship and support throughout, and for spending so much time
and money on overseas telephone calls while I was in Australia! I also
wish to thank my friends in Sydney for making the hours not spent on
research so very enjoyable.
Books cannot happen without institutional support, and I am very
grateful to the University of Augsburg for funding a six month research
project on emotions across corpora. I also appreciate the university grant-
ing me study leave to write this book, which would not have been
possible without the financial support of the Deutsche Forschungsgemein-
schaft (DFG). The majority of the research for this book was undertaken
at the University of Sydney, and I am very grateful to the Department
of Linguistics for their support throughout my stay there, in particular
Jim Martin, who also kindly agreed to write a foreword to this book.
It was also very helpful to have the opportunity to present parts of my
research at the research seminars in Sydney which provided important
feedback.
I would also like to thank Jim Martin and Wolfram Bublitz for valu-
able academic advice, discussion and support on draft material, Nicola
Bednarek for the cover design, Naomi Knight, Helen Caple, Ariane
Welch, and Steven Davey for native speaker help, Elisabeth Reber for
bibliographical advice, and Martin Wynne, Ylva Berglund (both from
the Oxford Text Archive), Sebastian Hoffmann, Peter White and Sally
Humphrey for help with corpus design and computer software, as well as
Michael Oakes for help with statistics. Susan Hunston and Nick Groom
kindly agreed to read draft chapters, and their advice has been much
appreciated. Needless to say, all remaining mistakes are mine.
This book was first conceived as a Habilitationsschrift for the University
of Augsburg, and I appreciate that Wolfram Bublitz, Stephan Elspaß and
Jim Martin agreed to be part of my Fachmentorat. I am grateful too to Jill
Lake for commissioning the book and the staff at Palgrave Macmillan
for their editorial support as well as an anonymous reviewer for his/her
review of a previous version of this book.

vii

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

viii Acknowledgements

Permission has been granted by Pearson Education Ltd for the repro-
duction of a table adapted from the Longman Grammar of Spoken and
Written English (1999, London: Longman, first edition, fourth impression
2004 by Biber, D., Johansson, S., Leech, G., Conrad, S. & E. Finegan).
Unless otherwise noted, examples of usage are taken from the British

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

National Corpus (BNC) and were obtained under the terms of the BNC
End User Licence. Copyright in the individual texts cited resides with the
original IPR holders. For information and licensing conditions relating
to the BNC, please see the website at http://www.natcorp.ox.ac.uk.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

List of Figures, Tables and Image

Figures

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

1.1 Language and emotion 10
1.2 Emotional talk vs. emotion talk 12
2.1 Nature vs. nurture 34
3.1 Words and patterns 66
3.2 Directed and undirected affect 90
5.1 Affect and evaluation 157
5.2 Authorial affect and evaluation 158
5.3 Non-authorial affect and evaluation 158
5.4 Four sets of affect type 161
5.5 Affect type and (ir)realis trigger 167
5.6 Modified affect system 168
5.7 Core and marginal emotions 168
5.8 Interaction between affect type, direction and
trigger 172
6.1 Distribution of affect sub-types (but not emoters)
in the conversation sub-corpus of the BRC baby 186
6.2 Affective key in academic discourse 200

Tables
1.1 Attitudes towards emotion 3
1.2 The BRC (British Register Corpus) 20
1.3 Major situational differences among the BRC registers 21
2.1 Key words in the BRC sub-corpora 29
2.2 Emotion terms in the BRC 32
2.3 Emotion terms excluding sorry, want, like 33
2.4 Ten most frequent emotion terms in the BRC sub-corpora
(raw frequency) 36
2.5 Part of speech distribution of 50 most frequent emotion
terms in each sub-corpus 37
2.6 Shared emotion terms 38
2.7 Part of speech variation in the BRC 40
2.8 Register preferences for part of speech 40
2.9 Ten most frequent emotion nouns in the BRC
(raw frequency) 41

ix

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

x List of Figures, Tables and Image

2.10 Shared emotion nouns 42

2.11 Ten most frequent emotion adjectives in the BRC
(raw frequency) 44
2.12 Shared emotion adjectives 45
2.13 Ten most frequent emotion adverbs in the BRC

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(raw frequency at least 2) 46
2.14 Ten most frequent emotion verbs in the BRC
(raw frequency) 47
2.15 Shared emotion verbs 48
2.16 Frequent emotion terms 49
2.17 Positive, negative, neutral emotion terms 49
2.18 Positive, negative, neutral emotion terms 51
2.19 Sharing of emotion terms 52
2.20 Syntactic variation 54
2.21 Variation in verb form 57
2.22 Distribution of emoters 59
3.1 Analyzed emotion terms 65
3.2 Local grammar analysis 67
3.3 Patterns and meaning groups 68
3.4 Coding of affect 70
3.5 to 3.34 Series of tables for different affect patterns 72–82,
84–89, 91–92
N1 99
5.1 Cline of implicitness 147
5.2 Ways of portraying emotions 150
5.3 Un/happiness 155
5.4 In/security 155
5.5 Dis/satisfaction 155
5.6 Dis/inclination 156
5.7 Sub-categories of un/happiness and
dis/satisfaction 160
5.8 Sub-categories of in/security 160
5.9 Modified in/security system 160
5.10 Behavioural surges of surprise in fiction 162
5.11 Fear vs. disquiet 165
5.12 Modified dis/inclination system 165
5.13 Affect types and irrealis trigger 167
5.14 Affect types and realis trigger 167
5.15 Comparison of affect types 169
5.16 Affect types and basic emotions 170
5.17 Modifications of affect system 171

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 List of Figures, Tables and Image xi

5.18 Emotion terms analyzed in the BRC baby 173

5.19 Analysis of covert affect in the BRC baby 176
5.20 Analysis of valence in the BRC baby 177
6.1 Affect in KE2 (BRC) 185
6.2 Affect types of characters 191

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

6.3 Affect sub-types in news reportage categories in the
BRC baby 194
6.4 Affect types of 50 most frequent emotion terms in the
BRC news reportage sub-corpus 195
6.5 Affect and mediation in the BRC baby news reportage
sub-corpus 197
6.6 Affect types in AS6 (BRC) 207
6.7 Number of emotion terms in news reportage categories
of the BRC baby 209
6.8 Affective keys and disciplines 210
6.9 A three-pronged approach 220

Image

5.1 XML Spy Analysis 153

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

System Network Notation

‘realized by’

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

|–
– ‘or’
--
---

-- ‘or’ (fuzzy system)

{ ‘and’

Appraisal theory’s labels for systems are usually in (small) capitals (e.g.
AFFECT, APPRECIATION), but to show that I am making use of appraisal in a
more theory-neutral fashion (see Section 1.4) I have not used this con-
vention in this book. Appendices to chapters from this book are available
online at http://www.MonikaBednarek.com. The book can also be read
without consulting the appendices, but researchers interested in more
detailed comments on methodological and other issues might find them
useful.

xii

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

1
Analyzing Language and Emotion

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

1.1 Introduction

While I was preparing the manuscript for this book in November 2006,
a youth in Germany shot and wounded several of his fellow students,
a teacher and the caretaker at his former school, before killing himself.
Coincidentally, just days earlier I had finished a novel by the American
writer Lionel Shriver about a similar scenario at an American high-school,
which includes the following speech by the shooter (Kevin), addressed
to his father:

‘I don’t care how your camera works.’ he continued levelly. ‘I don’t

want to be a location scout for a bunch of crappy products. I’m not
interested. I’m not interested in baseball or the founding fathers or deci-
sive battles of the Civil War. I hate museums and national monuments
and picnics. I don’t want to memorize the Declaration of Indepen-
dence in my spare time or read de Tocqueville. I can’t stand reruns
of Tora,Tora,Tora! or documentaries about Dwight Eisenhower. I don’t
want to play Frisbee in the backyard or one more game of Monopoly
with a snivelling, candy-ass, one-eyed midget. I don’t give a fuck
about stamp collecting or rare coins or pressing colorful autumn leaves
in encyclopedias. And I’ve had it up to my eyeballs with heart-to-heart
father–son talks about aspects of my life that are none of your business.’
(Shriver 2006: 425–6, italics in original).

This is Kevin’s response to his father’s patronizing and rather simplis-

tic behaviour towards him, and comes on the very day that Kevin kills
members of his family, a teacher and several pupils at his high school.
What is interesting about his response in the context of this book is

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

2 Emotion Talk Across Corpora

that it contains a lot of ‘emotion talk’: I don’t care, I don’t want (three
occurrences), I’m not interested (two occurrences), I hate, I can’t stand,
I don’t give a fuck about, I’ve had it up to my eyeballs with. These function
to position Kevin contrary to the activities endorsed by his father, his
father’s desire for Kevin to be just like him, and, simultaneously, against

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

his father himself. They provide evaluations of entities and behaviour,
asserting Kevin’s indifference, dislike and distaste towards what his father
so whole-heartedly believes in (i.e. mainstream American ideologies and
values). At the same time, they provide a characterization of Kevin’s
personality, rather than just pointing to momentary or transitional emo-
tional states. The passage is about interests, likes and values, though
tellingly these are only referred to in terms of what they are not, rather
than in terms of what they are. The extract seems to represent Kevin’s
final emotional reaction towards his father’s continual attempt to mould
him, their relationship, and their family life into some idolized ‘All-
American’ perfection, rather than accepting Kevin on his own terms.
Crucially, Kevin’s parents do not challenge or take him up on his words,
though his emotional outburst is very unusual and extremely significant
in light of his later deeds, indicating something of his ‘true’ state-of-mind
and foreshadowing what later happens. His emotion talk is indeed very
noteworthy.
Although this book is in no way about high-school killings or related
phenomena, it is about emotion talk (using emotion terms) and its func-
tions in discourse. It seeks to investigate how we use emotion talk in
different types of text (the four registers of casual conversation, fiction,
news reportage, and academic discourse) to position ourselves, to express
evaluations and to provide information, and is aimed at all researchers
interested in the use of emotion talk in naturally occurring discourse.

1.2 Emotion talk

Arguably, our emotions and how we talk about them are an essential part
of what makes us all human. Even if animals may also have emotional
experiences (Ekman 1992), humans can reasonably be regarded as the
most emotional of all sentient beings (Mees 2006: 3). The study of human
discourse about emotion therefore probes one of the most fundamental
human characteristics.1
However, our attitudes towards emotions themselves have in fact
always been rather mixed, oscillating between the negative and the pos-
itive, with diverging dichotomies emerging in Western culture (Table
1.1 on p. 3).2 The view of emotions as irrational probably goes back

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Analyzing Language and Emotion 3

Table 1.1 Attitudes towards emotion

Emotion Non-emotion
Emotion regarded as emotion ratio/intelligence/cognition
negative (distrust) natural cultural

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

irrational rational
heart head
passion reason
chaotic ordered
subjective universal, objective
physical mental/intellectual
unintended intended
uncontrollable controllable
dangerous not dangerous
Emotion regarded as emotion alienation
positive (appreciation) life death
interpersonal connection estrangement
free nature shackling civilization
authentic artificial

(After Lutz 1990: 69, Parrott 1995: 80–2, Daneš 2004, Oatley et al. 2006: 58).

as far as Plato’s and Darwin’s observations on emotions and emotional

expression (Oatley et al. 2006: 58). Their conception as dangerous and
uncontrollable is reflected in frequent talk about emotional control in
interviews (Lutz 1990, Parrott 1995), and psychological discourse (for
example, Fiehler 1990: 60, Ekman 1992: 189). The aspect of control is
also very important in the metaphorical construction of emotions, with a
focus on attempt at control, loss of control and lack of control (Kövecses
2000: 43). It is embodied in the ‘master’ metaphor for emotion, EMOTION
IS FORCE (Kövecses 2000: 17), which is exemplified by fixed expressions
such as:

• He was seized by emotion.

• He was struggling with his emotions.
(Kövecses 2000: chapter 5)

Further, Kidron & Kuzar (2002) point out that different cultural
opinions about correct emotional behaviour are associated with the
conceptualization of control with respect to emotional experience. The
Anglo-American culture tends to emphasize ‘self-restraint and control of
emotions’ (Kidron & Kuzar 2002: 134), which, as they argue, is reflected

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

4 Emotion Talk Across Corpora

by the way emotions are syntactically encoded in English. More specif-

ically, for Americans, being emotional has negative connotations, and is
linked to ‘losing control, confusing or mixed emotions, becoming irra-
tional’ (Parrott 1995: 78). Males in Western culture appear reluctant to
accredit (de Beaugrande 1992: 247) emotions, that is, accept and express

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

emotional experience (but see Galasiński 2004).
Our attitudes towards, and thoughts and feelings about, emotions have
been described as meta-emotion philosophy, and vary among cultures as
well as individuals (Gottman et al. 1996: 243–5). From a different, more
general perspective, the sociologist Arlie Hochschild talks of cultural
‘emotion ideologies about appropriate attitudes, feelings, and emotional
responses in basic spheres of activity.’ (Turner & Stets 2005: 36). Meta-
emotion philosophies and presumably also emotion ideologies, are
reflected in discourse, with statements such as Getting angry can be a
relief opposed to remarks like Her shouting scares me (Gottman et al. 1996:
267). Emotion talk can also reveal emotional culture (Gordon 1990) – what
Stearns (1994, 1995 in Bamberg 1997b) calls the emotional style of a cul-
ture. Our ‘emotion talk’, then, may reveal our personal and cultural
attitudes towards emotional experience but what else is its function?
How do we talk about emotion in different situations, when do we do
so and what is the purpose of this ‘emotion talk’? These are all questions
that will be discussed throughout this book. The following sections give
an overview of emotion research and outline the framework of analysis.

1.3 A brief history of emotion research

Emotions have been subject to a large number of empirical and theo-

retical studies, and it is impossible to do justice to all. Instead, only the
most important ones (mainly in psychology, sociology and linguistics)
are briefly reviewed. Concise overviews of many aspects of emotions can
be found in Davidson et al. (2003) and Oatley et al. (2006), and there are
also a myriad of handbooks and encyclopaedias of emotion research.
Essentially, emotion research was founded in the 19th century by
Darwin, James and Freud (Oatley et al. 2006: 4–10), but emotions
themselves have been a topic for much longer, starting with ancient
philosophers such as Aristotle. They are a topic that is of interest to many
disciplines: philosophy (for example, Marty 1908, Richards 1948, Black
1948, Bedford 1956/57, Wittgenstein 1967, Goldberg 1971, Belfrage
1986, Wollheim 2001, Robinson 2005); psychology (for example, Ortony
et al. 1988, Ekman 1992, 1997, 1999a, Johnson-Laird & Oatley 1989,
Braun 1992, Gallois 1994, Gottman et al. 1996, Mees 2006, Oatley et al.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Analyzing Language and Emotion 5

2006); neuroscience (for example, Davidson et al. 2003: part 1); anthro-
pology (for example, Abu-Lughod & Lutz 1990, Goodwin & Goodwin
2000, Milton & Svašek 2005); sociology (for example, Gordon 1990,
Turner & Stets 2005); communication studies (Planalp 1999); linguistics
(see below) and so on.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Some recurring issues in emotion research are:

• What is an emotion? How can emotions be defined?

• What is the structure of emotions?
• How can emotions be distinguished from each other?
• How can emotions be studied, measured, and described?
• Are emotions innate (biological) and universal, or acquired and
culturally construed?
• Are there ‘basic’ (innate, universal, primary, cognitively salient)
emotions and what are they?
• What is the relation between linguistic resources (providing labels for
emotions) and emotional experience?

To review discussions of these aspects in detail would take us too far:

for instance, there are more than a hundred definitions of emotion (Jahr
2000: 7; see for example, Mees 2006, Oatley et al. 2006 for overviews) and
heated discussions on most of the other questions can be found as well.
However, it is interesting that several definitions note that emotions
include ‘an eliciting condition, a cognitive evaluation, physiological
activation, a change of action readiness, and finally an action’ ( Johnson-
Laird & Oatley 1989: 82). The current consensus in emotion research
seems to be that emotions are neither wholly universal nor wholly
culturally determined, and that these two views are not completely
incompatible (Parrot & Harré 1996: 2, Planalp 1999: 195). Some aspects
of emotional behaviour are universal and rooted in biology (for exam-
ple autonomic and central nervous system activity, facial expression)
with the limbic system (the amygdala), the neocortex, subcortical regions
of the brain, as well as hormones, neuromodulators, and transmit-
ter substances related to emotional experience and emotion regulation
(Turner & Stets 2005: 4–9, Oatley et al. 2006: chapter 6). At the same time,
there is no doubt that other aspects of emotional experience are deter-
mined by socialization and cultural construal (even with respect to uni-
versal aspects such as facial expression), and there is both cultural, sub-
cultural and individual variation (Ekman 1997, 1999b: 14, Ellsworth &
Scherer 2003: 584, Schrauf & Sanchez 2004: 282, Oatley et al. 2006: 68–9,
97, 180). While there is a wealth of research on cultural constructivism

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

6 Emotion Talk Across Corpora

and emotions (for example Harré 1986, Fiehler 1990, Abu-Lughod

& Lutz 1990, Harré & Parrott 1996, Oatley et al. 2006: 70–2), the
most well-known (non-linguistic) research probably relates to Goffman’s
dramaturgical/cultural theory, Hochschild’s establishment of feeling rules
and Ekman’s display rules (Ekman 1997, Turner & Stets 2005: 36–7).

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Dramaturgical theories see human behaviour as a scripted on-stage per-
formance which is influenced by cultural norms and beliefs about emo-
tional experience and expression, including feeling rules as generated
by culture (Turner & Stets 2005: 23–4). These rules regulate emotional
experience and management. Emotion rules, manifestation rules, correspon-
dence rules, and coding rules regulate which emotions are to be expected
in which intensity in social situations, how they are conventionally
manifested, and how to react with the appropriate emotion – varying
depending on social roles, gender, situation, culture (Fiehler 2002: 82–
3). In Downes’s words, ‘culture specifies “what you are supposed to feel” ’
(Downes 2000: 108).
Aspects that seem to have an influence on the socialization of emo-
tions are the development of Über-ich, conscience and ratio (Wilk 2005:
132), art (van Meel 1994: 163), as well as parent–child interaction and
other microsocial interpersonal relationships through which macroso-
cial structures work (Gordon 1990: 147). Emotional intelligence, it
seems, is learned in childhood, and emotion talk plays a crucial role in
the socialization of emotions (Planalp 1999: 142–3). The talk of emotions
and events that evoke them:

teach children about what events appropriately elicit emotions in

their community, inducting the child into the cultural rules of emo-
tional expression. Emotion talk also structures the child’s own internal
experience, and lets the child know about the internal experience of
others
(Oatley et al. 2006: 302).

Though the connection between language and emotion had, for a

long time, been neglected in linguistics (cf. Lyons 1982: 103, Finegan
1995: 2, Scheibman 2002: 7), by now a great variety of linguistic stud-
ies on language and emotion exist. However, while we may indeed talk
of a new interest in emotive language, or the ‘ecology of subjectivity’
(Bublitz 2003: 389), there is, as yet, no unified theory of affect or emo-
tion. Instead, we find a range of at times widely-differing approaches to
the expression of emotion in general. One reason for this may be that
the relation between language and emotion is itself quite complex: we

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Analyzing Language and Emotion 7

can express feelings that we have, we can have feelings that we do not
express, and we can express feelings that we do not have (Daneš 1987:
174f, Caffi & Janney 1994). The feelings may be expressed consciously or
subconsciously, intentionally or unintentionally, spontaneously (auto-
matically) or strategically (Marty 1908, Daneš 1987, Planalp 1999: 71ff),

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

and may relate to various aspects of the communicative context (Arndt &
Janney 1987: 78–9). We can look at them in terms of the speaker’s (self-)
expression, the potential of language to express emotion, or the pre-
sumable influence on the hearer. All these different perspectives can be
found in the various approaches to affect/emotion in linguistics, which
furthermore adopt different viewpoints according to the sub-discipline
of linguistics in which they can be situated:3

• The cognitive approach: cognitive-linguistic research on words that

refer to emotions, that is, the ‘emotion-lexicon’ (Palmer & Occhi
1999: 15). This approach examines how emotions are conceptual-
ized (for example in terms of emotion schemata) and is concerned with
the question of the universality of emotions, the origin of linguistic
expressions of emotions, and the relation between emotions and their
linguistic labels. This strand of study is sometimes called emotionol-
ogy (Athanasiadou & Tabakowska 1998b: xii) and is exemplified by
Athanasiadou & Tabakowska (1998a), Harkins & Wierzbicka (2001),
and Kövecses (for example Kövecses 2000).
• The cross-linguistic approach: the study of emotion terms across
languages. This approach seeks to demonstrate that the expression
of emotions relies on culturally determined notions of emotions.
In this context, Wierzbicka’s (for example 1992b, 1999) concept
of ‘semantic primitives’ is most well-known (for a critique see
Bamberg 1997b, Weigand 2004b). Other studies are Athanasiadou &
Tabakowska (1998a), Ochs & Schieffelin (1989), Ungerer (1997),
Harkins & Wierzbicka (2001), Kidron & Kuzar (2002), Dem’jankov
et al. (2004) and Teubert (2004a, b).
• The linguistic-anthropological approach: studies in this area (for
example Lutz & Abu-Lughod 1990, Irvine 1990, Palmer & Occhi
1999, Goodwin & Goodwin 2000) include work on language acquisi-
tion and ethnographic research on poetics and performance (Besnier
1990: 420ff). Studies of emotion in linguistic anthropology are
also interested in ‘[t]he problem of how emotions are conceptual-
ized, described, expressed, and realized in purposive actions in each
language and culture’ (Palmer & Occhi 1999: 2). Another focus is on
emotion as social practice (Goodwin & Goodwin 2000).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

8 Emotion Talk Across Corpora

• The diachronic approach: the attempt to trace the diachronic devel-

opment of language that is associated with emotion. For example,
Hübler (1998) explores the evolution of some grammatical means
of expressivity in English; Györi (1998) is interested in semantic
change concerning the conceptualization of emotions in different lan-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

guages; and Teubert (2004a) tries to trace the origin of the feeling of
guilt.
• The functional approach: many modern studies on emotion can be
traced back to the historical tradition of research on the functions
of language (for example, Marty 1908, Bühler 1934, Jakobson 1960,
Stankiewicz 1964; see also Sarangi 2003 for research by Richards in the
1920s), and some more recent approaches to emotion also deal with
the notion from a functional point of view (for example, Šabršula
1982, Péter 1984, Daneš 1987, Stankiewicz 1989, Martinet 1991,
Foolen 1997). Related to these are approaches to affective/expressive
language (for example, Charleston 1960, Leech 1974, Schneider 1991)
or studies which are concerned with the specific status of signs with
emotive meaning (usually in terms of the difference between indexi-
cal and symbolic meaning as elaborated by Peirce 1978: for example,
Volek 1977, 1987, Konstantinidou 1997).
• The syntactic approach: studies concerning the syntax of emotion
terms, for example the use of different prepositions (Dirven 1997,
Osmond 1997, Radden 1998) or emotion verb complementation
(Werth 1998).
• The conversation analytic approach: studies taking up the legacy of
researchers such as Schegloff and Sacks to focus on the display of emo-
tion in discourse, specifically talk in interaction, with an interest in
turn-taking and other types of structural organization ( Jefferson 1988,
Sandlund 2004), and the ‘embodied performance of affect’ (Goodwin &
Goodwin 2000: 254, original emphasis).
• The stylistic/literary approach: the study of the emotional impact of
stylistic devices, as well as analyses of narrative perspective, modality,
evidentiality, and expressions of emotion in literature (Busse 1992:
177–236, Simpson 1993, van Meel 1994, Watson 1999, Downes 2000,
Čmejrková 2004, Robinson 2005); the analysis of emotion as reader
response (Oatley 1994, 2003). On Bally’s influential stylistic approach,
see Caffi & Janney (1994: 333–5) and Hübler (1998).
• The psycholinguistic approach: research concerning the devel-
opment of emotions and related language in childhood – the
ontogenetic perspective – for example, studies on the variation of
emotional talk in different social strata (Burger & Miller 1999).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Analyzing Language and Emotion 9

Compare Shaver et al. (1987: 1083–4), Bamberg (1997b), Niemeyer &

Dirven (1997: section III), Planalp (1999: 140ff), Reilly & Seibert
(2003), Painter (2003), Russell et al. (1995: section 3), and Oatley
et al. (2006: 213–14, chapter 8).
• The pragmatic/textlinguistic approach: Studies in this area are

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

interested in many aspects of language and emotion. Some examine
‘the conventional displaying of affect through linguistic means [for
example pronouns, mood, tense/aspect/voice, intonation, lexis, dis-
course structure, affective speech acts]’ (Ochs & Schieffelin 1989: 7).
Others are interested in emotive communication, where the notion
is related to strategic uses of language, and is regarded as interac-
tional, interpersonal, and other-directed (Caffi & Janney 1994: 328f).
Arndt & Janney (1987) deal with the general influence of attitude
on communicative decisions and regard emotive communication as
a complex verbal, vocal, and kinesic phenomenon; Janney (1996)
establishes emotive ‘strategies’ such as approach and avoidance ( Janney
1996: 343ff). Ungerer (1997), on the other hand, looks at how emo-
tions can be evoked in hearers, and Bublitz (2002, 2003) is interested
in the ‘emotive prosody’ of texts, which express the speaker’s atti-
tudes and emotions. Daneš focuses on how the expression of emotion
is organized in text, in what he calls ‘the global emotional course
(profile)’ of discourse (Daneš 1987: 177). Other studies analyze the
connection between emotions and speech acts (Weigand 2004b: 16–
18), and are interested in social constructivism (Bamberg 1991, 1997a,
b) or ideological uses of emotion terms (Stubbs 1996: 85ff). Galasiński
(2004) studies male emotion talk in connection with masculinity.
A special strand within the pragmatic approach is the inten-
sity/involvement approach: involvement is the speaker’s ‘emotional
engagement in the interaction, or ego-identification with the topic
or partner of conversation’ ( Janney 1996: 136f). Besnier (1994),
Daneš (1994), Caffi & Janney (1994) and Watson (1999) give an
overview of approaches to involvement (the classic reference is Chafe
1982). Intensity is defined as ‘the emotional expression of social ori-
entation toward the linguistic proposition: the commitment of the
self to the proposition’ (Labov 1984: 43f). Intensity markers are
concerned with the degree of personal involvement of the speaker
towards the described states of affairs (Dorfmüller-Karpusa 1990).
Studies on intensity markers, or intensifiers, have focused on all
aspects of speech that are capable of being modulated in terms of
a higher on lesser degree of force. For examples see Janney (1996:
154, 160).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

10 Emotion Talk Across Corpora

• The systemic-functional approach: systemic-functional linguis-

tics is interested in affect in connection with appraisal theory, an
approach to the interpersonal function of language (Martin &
White 2005).

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

These approaches differ according to whether they concern the
language about emotion (linguistic expressions denoting emotions) or
language as emotion (linguistic expressions as conventionalized reflexes
or indices of speakers’ emotions) (Grondelaers & Geeraerts 1998: 357).
Different terms have been used by researchers for these two aspects
(Figure 1.1).

Language and emotion

gefühlsbezeichnend emotional gefärbt (Péter 1984: 246–7)

language about emotion language as emotion (Grondelaers &

Geeraerts 1998: 357)

conceptually descriptive, signal-like, hot emotions (Daneš 1987: 170–3)

cold emotions

discourse on emotions emotional discourse (Abu-Lughod &

Lutz 1990: 10)

speaking of emotions expressing emotions (Athanasiadou &

Tabakowska 1998b: xi)

communicating emotion communicating emotionally (Planalp

1999: 43)

emotion talk/talk about emotions expression of emotions (Bamberg 1997b)

descriptive emotion words expressive emotion words (Kövecses

2000: 2)

thematization of emotion emotional expression (Fiehler 2002: 86–7)

talk about emotion emotional talk (Koven 2004: 47)

emotions referred to directly emotions referred to indirectly (Dem’jankov

et al. 2004: 164)

Figure 1.1 Language and emotion

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Analyzing Language and Emotion 11

Simplifying the matter slightly, the language about emotion or emo-

tion talk is constituted by all those expressions in the dictionary that
denote affect/emotion, for example love, hate, joy, envy, sad, mad, enjoy,
dislike and so on (as well as fixed expressions such as He had a bro-
ken heart). Language as emotion or emotional talk relates to all those

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

constituents (verbal, non-verbal, linguistic, non-linguistic) that conven-
tionally express or signal affect/emotion (whether genuinely experienced
or not, whether intentional or not). Examples that are mentioned in
the relevant literature (see above) include intonation, mental process
verbs, grading (intensifiers, comparison, quantifiers, mood, modality,
negation), repetition, interpersonal metaphor, figurativeness, punctu-
ation, interjections, affective derivation (diminutives/augmentatives),
inversion, exclamation, syntactic markedness, pronoun use, emphatic
particles, intensifiers, expletives, vagueness, affective connotations,
evaluative adjectives, and many more. Paralinguistic devices are facial
expressions, vocal cues, gestures, body posture, body movement, and
physiological cues (see for example, Planalp 1999: 44f f).
However, approaches to language and emotion cannot neatly be classi-
fied according to these two broad categories, since ‘the two [talk/writing
about emotions and the interweaving of emotions and discourse] are
related in a complex manner’ (Besnier 1990: 437). Hence, emotional
talk also often involves the first person usage of emotion terms (I love) –
which on a functional level can be linked to expressing emotion (see
also Kövecses 2000: 2) – whereas emotion talk comprises all usages of
emotion terms, but excludes other emotional talk devices (Figure 1.2 on
page 12). Consequently, some studies of affect/emotion only deal with
emotional talk (sometimes including, sometimes excluding the usage of
emotion terms); others deal only with emotion talk (for example the cog-
nitive approach), and in others both approaches are combined (Harkins
& Wierzbicka 2001, Lutz & Abu-Lughod 1990).
To make matters less complicated I shall from now on work with a strict
definition of emotional talk as including all sorts of human behaviour
that signal emotion without the recourse to linguistic expressions that
explicitly denote emotion (emotion talk). I will thus use a strict dichotomy
between signalling and denoting affect. This means that all expressions
that denote affect whether they refer to the self or the other will be
considered as part of emotion talk (rather than as part of emotional talk):

affect: signal (Oh, fuck)
affect: self (1st person)
affect: denote (I’m really angry)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

12 Emotion Talk Across Corpora

⎧
⎪
⎨affect: signal (And then he goes
affect: other (non-1st person) ‘Oh, fuck’)
⎪
⎩affect: denote (And he was very angry)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Whereas much research on affect (outside cognitive linguistics) con-
centrates on affect: self: express (that is, indications/signals of the
emotions of the speaker), this study focuses on affect: self/other: denote.
The main starting point of the analysis is thus simply lexis that denotes
affect. The focus is on the usage in discourse of emotion terms (emotion
talk) in British English, and no claims are made concerning a general
(universal) theory of emotion or human psychology or languages other
than English. The theoretical approach that is adopted here for the
investigation of emotion terms is appraisal theory.

Intonation, punctuation,
interjections, affective derivation,
inversion, exclamation, syntactic
Emotion terms
markedness, pronoun use, emphatic
(First, second, third person)
particles, intensifiers, swear words,
vagueness, first person usage of
emotion terms, and so on.

Emotional talk: Emotion talk :

linguistic expressions linguistic expressions that denote

that conventionally (the speaker’s and others’) emotions
signal the speaker’s emotions

Figure 1.2 Emotional talk vs. emotion talk

1.4 Appraisal theory

Even though appraisal theory works within systemic functional lin-

guistics (SFL), it can also be adopted in a more theory-neutral way to
the analysis of language. This is the aim of my analyses in this book,

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Analyzing Language and Emotion 13

which – while being sympathetic to systemic functional linguistics

(SFL) – is nevertheless not rooted in SFL as such, and is aimed both at the
systemic functional (appraisal) analyst as well as any researcher interested
in emotion talk (specifically, corpus linguists). It adopts appraisal theory
as one of its methodological tools, with corpus and cognitive linguistics

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

and pragmatic analysis as complementary tools. For this reason, the fol-
lowing discussion of appraisal theory ignores considerations of SFL as a
theory of language in general (for outlines, see Halliday & Matthiessen
2004, Eggins 2004, Martin & Rose 2007).
But why adopt appraisal theory, if I do not consider myself a systemic
linguist as such? It does seem to me that its classification of resources
of interpersonal meaning lends itself in particular to discourse analytical
purposes. Its focus is on language in its social function, rather than on
language and the mind (as in cognitive linguistics), and it is based on the
analysis of naturally occurring discourse, in line with corpus linguistic
and textlinguistic principles. In contrast to the pragmatic approaches
mentioned above, which focus more on general emotive strategies or
on the broad concepts of intensity/involvement, it is also specifically
suited to the analysis of emotion talk, since one of its sub-systems (affect)
specifically describes this aspect of social meaning.
Appraisal deals with the expression of interpersonal meanings, includ-
ing ‘resources for modalising, amplifying, reacting emotionally (affect),
judging morally (judgment) and evaluating aesthetically (appreciation).’
(Martin 1995: 28). Although this is a comparatively new theory, there is
already a large body of research making use of it (for example Christie &
Martin 1997, Coffin 2006, Eggins & Slade 1997, Iedema et al. 1994,
Macken-Horarik & Martin 2003, Martin 1995, 1997, 2000a, 2002, 2004a,
Rothery & Stenglin 2000, White 2000, 2002, 2003a, b, 2004a, b,
Martin & White 2005, Adendorff & de Klerk 2005) which continues to
grow. Unless otherwise noted, the outline of appraisal in this chapter
follows its most recent description by Martin & White (2005).
Appraisal is divided into three sub-systems: attitude, engagement, and
graduation, with further sub-divisions:

Engagement

Affect

Appraisal Attitude Judgement

Appreciation

Graduation

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

14 Emotion Talk Across Corpora

Graduation and engagement concern the modification of the inten-

sity or force of an utterance (graduation) as well as the degree of speaker
commitment towards the utterance (engagement) and are not partic-
ularly relevant in the context of this book. Much more important for
the analysis of affect is the attitude system. Attitude is concerned with

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

evaluations relating to emotion, morality/ethics and aesthetics, con-
sisting of the three sub-systems of affect, judgement and appreciation.
(Incidentally, the relation between affect and appraisal had already been
noted by the philosopher Bedford in the 1950s: see Bedford 1956/57.)
Affect has to do with describing positive and negative emotions, both
of the speaker (authorial affect: I love) and third parties (non-authorial
affect: s/he loves, you love, they love). Affect can be realized by adjectives
(a sad person), verbs (he cried, he loved him), nouns (his grief ) and adverbs
(desperately). Affect can furthermore be classified according to six factors:

The feelings are culturally construed as positive or negative: positive

affect (the boy was happy) vs. negative affect (the boy was sad ).
The feelings are realized as a surge of emotion involving para- or
extralinguistic manifestation or are more mentally experienced as an
ongoing emotional state: behavioural surge (the boy laughed, the cap-
tain wept) vs. mental disposition (the boy liked the present/felt happy, the
captain disliked the present/felt sad).
The feelings are construed as directed at/reacting to some external
agency or as a general mood: reaction to other (the boy liked the
teacher/the teacher pleased the boy) vs. undirected mood (the boy was
happy).
The feelings are graded in terms of a cline of intensity: low (like) –
median (love) – high (adore).
The feelings relate to future states or existing ones: realis (the boy liked
the present) vs. irrealis (the boy wanted the present).
Emotions are grouped into three major sets: in/security (the boy was
anxious/confident) – dis/satisfaction (the boy was fed up/absorbed) –
un/happiness (the boy was sad/happy).

These factors will be described in more detail in Chapter 5, which

deals with a modification of appraisal theory. It remains to be noted
at this stage that whoever experiences the emotion is classified as the
emoter, and what evokes the emotion as the trigger, for example: The

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Analyzing Language and Emotion 15

boy [emoter] liked the present [trigger]. This will become important in
Chapter 3.
Judgement systems consist of resources for morally evaluating human
actions, behaviour or character, by reference to a set of ethic norms.
Judgement is subdivided into two broad categories: judgements of social

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

esteem (normality: standard–odd, capacity: clever–stupid, tenacity: brave–
cowardly) and judgements of social sanction (veracity: honest–deceitful,
propriety: moral–immoral). These can be positive or negative (admiration
vs. criticism).
Appreciation systems include resources used to evaluate the
(aesthetic) quality of processes, things and products (and human beings
when they are seen as entities), for example It’s a fantastic book. Like affect
and judgement, appreciation also has a positive and negative dimension,
and is organized around three variables: reaction, composition, valua-
tion. Reaction concerns the impact of the text/process on our attention
(impact: captivating–dull) and its attitudinal impact (quality: beautiful–
ugly). Composition concerns perceptions of proportionality/balance
(balance: harmonious–discordant) and detail (complexity: intricate–
simplistic) in a text/process. Valuation has to do with our assessment
of the social significance of the appreciated entity (profound–shallow).
As becomes evident, not all sub-systems of attitude are of relevance
here; rather, it is only the system of affect that concerns us. However,
since there are some connections between judgement, appreciation and
affect (White 2001: 3–4), and the borders between them are far from clear
(Martin & White 2005: 57–61), both appreciation and judgement will at
times come up in subsequent discussions. It must also be pointed out
that both judgement and appreciation are considered as instituational-
izations or recontextualizations of affect in appraisal theory:

AFFECT can perhaps be taken as the basic system, which is then insti-
tutionalized in two major realms of uncommon sense discourse.
As JUDGEMENT, AFFECT is recontextualized as an evaluation matrix for
behaviour, with a view to controlling what people do. As APPRECIATION,
AFFECT is recontextualized as an evaluation matrix for the products
of behaviour (and wonders of nature), with a view to valuing what
people achieve.
(Martin 2000a: 147)

Appraisal has so far predominantly been applied to individual

texts or relatively small corpora (for example Miller 2006), with
the help of detailed manual analyses. (More recently, slightly larger

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

16 Emotion Talk Across Corpora

corpora – 500,000–2.6 million words – are used for the analysis of

appraisal by Miller 1999, 2002, Adendorff & de Klerk 2005, and
Kaltenbacher 2006, 2007; Coffin & O’Halloran 2005, 2006 also explore
corpus linguistic methods for analyzing appraisal.)
In contrast, I shall use both large- and small-scale corpus data in inves-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

tigating affect in this book. The study will consider affect in four different
registers, adopting this vantage point from the register analyses by Biber
and his colleagues (for example Biber et al. 1999) with respect to stance
analysis, described in the following sections.

1.5 Stance analysis and register variation

Since his seminal study of variation in spoken and written language

(Biber 1988), Biber has been identified with the corpus-based study of
linguistic variation. Of particular concern here are studies that ana-
lyze the expression of stance in different registers.4 Stance is similar
to appraisal and can be defined as ‘the expression of personal feel-
ings and assessments’ (Conrad & Biber 2000: 57). The notion of stance
includes three broad categories: epistemic stance (certainty/doubt),
style stance (discourse comments), and attitudinal stance (positive/
negative attitudes/feelings). Epistemic and style stance are not relevant
here, but attitudinal stance is. This conveys speakers’ attitudes, feelings
or value judgements, including both emotion vocabulary (happy, love),
and evaluative expressions (wonderful, lovely, good) (Biber et al. 1999:
968). Consequently, attitudinal stance is a broader notion than affect,
making no distinction between the systems of affect, judgement and
appreciation, and is more or less equivalent to attitude rather than affect.
Furthermore, in some analyses of attitudinal stance no difference is made
between emotion talk and emotional talk, so that, for example, expletives
(God, damn) are included (Precht 2000: 67).
At the same time, analyses of attitudinal stance pick up only a small
percentage of emotion talk, since the number of expressions consid-
ered as emotion terms is very low. This is because few emotion verbs,
adjectives, adverbs and nouns are frequent enough to be included in
a factor analysis, the corpus linguistic methodology most often used
(Precht 2000: 68–71). Dry even argues that 60 per cent of emotion nouns
in academic discourse are ignored by Biber & Finegan’s (1989) method-
ology (Dry 1992, cited in Precht 2000: 13). Additionally, many studies
of attitudinal stance are limited to the analysis of first person usage of
emotion terms, since only these are said to be ‘direct and explicit expres-
sions of speaker attitude’ (Biber & Finegan 1989: 97; for criticism see

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Analyzing Language and Emotion 17

Watson 1999). In any case, ‘affect is the least-thoroughly analysed aspect

of stance’ (Precht 2000: 12).
Unfortunately, these three aspects (attitudinal stance equals the sys-
tem of attitude rather than that of affect; only some emotion terms
are included in the analysis; third-person references to emotion are

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

disregarded) mean that the results of stance analyses are not directly com-
parable to the results undertaken here. What this study owes to stance
analysis, and the analysis of register variation in general, is the insight
that ‘[l]inguistic variation is central to the study of language use’, that
it is systematic, depends on many contextual factors, and that it should
be analyzed with the help of corpora (Reppen et al. 2002a: vii). In this
respect, I assume that emotion terms vary across registers, that only a
corpus-based analysis can tell us about the details of this variation, and
that the usage of emotion terms is related to the characteristics of the
given registers. The analysis must include more than individual words,
however, since much corpus linguistic research (for example Sinclair
2004a,b, Hoey 2006) has shown the importance of lexico-grammatical
patterns such as collocation and colligation. With respect to this,
Chapters 3 and 4 will take up the notions of local grammar (Hunston
& Sinclair 2000) and FrameNet (http://framenet.icsi.berkeley.edu/), and
will move beyond the investigation of individual emotion terms which is
the focus of Chapter 2. The aims and methodology of this investigation
will be outlined in the following sections.

1.6 Bringing it all together

1.6.1 Aims
As mentioned, the aim of this book is to study the usage of emotion
terms (emotion talk) in four varieties (or registers) of British English in a
large corpus including casual conversation, fiction, news reportage and
academic discourse. An emotion term is defined as a lexical item that
denotes emotion in a broad sense, namely affect, feelings, emotional
states, moods, and so on (see Note 1). Prototypical examples are adjec-
tives such as happy, sad, nouns such as joy, anger, adverbs such as happily,
and verbs such as love, hate. Together, such lexical items make up the
resources for emotion talk in British English.
More specifically, this book aims firstly at the quantitative establish-
ment of emotion profiles for each register. This includes:

• Lexical variation: frequency and distribution of emotion terms

(emotion profile I )

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

18 Emotion Talk Across Corpora

• Part of speech variation: frequency and distribution of word classes

(emotion profile II )
• Syntactic variation: frequency and distribution of choices in gram-
matical paradigms (emotion profile III )
• Lexico-grammatical variation: frequency and distribution of syntag-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

matic lexico-grammatical patterns (emotion profile IV )

This large-scale corpus study of emotion terms (making use of a 19.5 mil-
lion word corpus) is followed by an exploratory pragmatic/discourse-
analytic analysis of emotion talk in a small subset of the corpus (∼ 85,000
words). As a background to these analyses, I use a modified version of
appraisal theory which is developed with the help of corpus linguistic
data and insights from cognitive linguistics/psychology (Chapter 5).
Summing up, this book combines both large-scale and small-scale
empirical studies, and takes up three important current approaches to the
study of language, offering complementary perspectives: the systemic,
the corpus linguistic, and the discourse-analytic. As Partington has
pointed out as late as in 2004, discourse analysis and corpus linguistics
are still rarely combined (Partington 2004: 11). The same is true to some
extent for the combination of SFL and corpus linguistic data (Butler 2004:
147; for existing research see Thompson & Hunston 2006, Kaltenbacher
2007). The main focus will be on human discourse about emotion in
actual, naturally-occurring language, and its polyfunctionality.

1.6.2 The corpus

The corpus used for the analysis of affect is a register-sensitive corpus of
British English. Register is here used in Biber et al.’s (1999) sense, defining
a variety of language that is based on external, non-linguistic, situational
criteria:5

Register distinctions are defined in non-linguistic terms, with respect

to situational characteristics such as mode, interactiveness, domain,
communicative purpose, and topic. For example, newspaper editori-
als are distinguished as being (a) written, (b) published in a newspaper,
and (c) primarily intended to express an informed opinion on matters
already in the news.
(Biber et al. 1999: 15)

The chosen registers also broadly correspond to those analyzed by Biber

and his colleagues (but see below): conversation, fiction, news, and aca-
demic discourse. As they note, ‘[t]hese registers have the virtue of being

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Analyzing Language and Emotion 19

(a) important, highly productive varieties of the language, and (b) dif-
ferent enough from one another to represent a wide range of variation’
(Biber et al. 1999: 15–16). For the analysis of the four chosen registers,
I used a custom-made corpus of British English compiled from various
parts of the British National Corpus (BNC) with the help of Lee’s (2002)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

classification of this corpus (see also Lee 2001). The aim was to com-
pile a corpus that was as big as possible (to increase the reliability of the
language data) while being as representative of the chosen registers as
possible (within the limitations imposed by the contents of the BNC and
based on language-external criteria). The main advantage of an approach
that uses a publicly available classification system (Lee 2002) and corpus
(the BNC) is its replicability (Oakey 2002: 115) – ‘the scientists’ favourite
criterion’ (Kilgarriff 1997a: 147). This corpus will from now on be referred
to as British Register Corpus (BRC). (I will only give a brief outline of
the corpus here, but its design is described in detail in Appendix A 1.1
online). The BRC consists of a conversation sub-corpus, a news reportage
sub-corpus, a fiction sub-corpus and an academic discourse sub-corpus:

• Conversation: 4,206,058 words of casual conversational British

English;
• News reportage: 2,613,399 words of British tabloid and broadsheet
news reports, including arts/cultural material, commerce/finance,
home/foreign news, science, lifestyle/leisure/belief and thought, and
sports;
• Fiction: 6,688,459 words of adult fiction by male and female authors
from 1985 to 1994 in book form;
• Academic discourse: 5,960,933 words of different types of writ-
ten academic discourse from the humanities, medicine, natural
sciences, politics/law/education, social/behavioural sciences, and
technology/computing/engineering.

Although the chosen registers are roughly equivalent to those investi-

gated by Biber, the equivalences are not total. For instance, the BRC
sub-corpus of news reportage does not include (persuasive) editorials,
and is thus more narrow than Biber’s register of News. The corpus is also
restricted to British English. (For further differences between the two
corpora compare the design of the BRC as described in Appendix A 1.1
online, and Biber et al.’s 1999 description of the LSWE corpus). In total,
the BRC consists of about 19.5 million words (Table 1.2 on p. 20).
The BRC is a parallel or contrastive corpus, since it aims to investi-
gate differences between externally-identified varieties of British English

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

20 Emotion Talk Across Corpora

Table 1.2 The BRC (British Register Corpus)

Conversation 4,206,058 words

News reportage 2,613,399 words
Fiction 6,688,459 words
Academic discourse 5,960,933 words

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Total 19,468,849 words

(Sinclair 2004c: 3). The four registers are differentiated by a variety of

contextual/situational factors, as visualized in Table 1.3 (on p. 21).
As Biber et al. (1999) have outlined, conversation is characterized by
the individual, local and personal, by immediacy and interactiveness;
news reportage, in contrast, is more public, appeals to the nation as a
whole, and has low degrees of interactiveness and immediacy. Fiction
is mainly distinguished from news reportage by its communicative pur-
pose, and its even more global appeal. Finally, academic discourse is
much more specialist than all other registers, but shares some features
with news discourse (low degrees of interactiveness and immediacy, pro-
viding information). These and other situational differences among the
registers will be taken up again later when relating the findings of the
analyses to a functional interpretation.

1.6.3 Emotion terms

In examining emotion talk in a large corpus the first task is to compile a
comprehensive list of lexical items to be included in the analysis. After
surveying much of the existing research on and lists of emotion terms
(for example Wallace & Carson 1973, Nissenbaum 1985, Ortony et al.
1987, Storm & Storm 1987, Biber & Finegan 1989, Johnson-Laird &
Oatley 1989, Nöth 1992, Janney 1996, Dirven 1997, Osmond 1997,
Moore et al. 1999, Precht 2000),6 and consulting different dictionaries
and thesauri, I decided to base my list of emotion terms on the classifi-
cation provided by the 2001 Encarta Thesaurus (ET ), since this is, on the
one hand, corpus-based, and, on the other hand, seemed the most com-
prehensive and accessible. Since the relevant thematic section (labelled
Emotions and States of Mind) in the ET does not differentiate between
emotions and states of minds, only a subset of this category was used
for establishing the list of emotion terms (focusing solely on adjectives,
adverbs, nouns and verbs), including the following semantic categories:

• Feelings about the past (n)

• Feelings about the future (n)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
Table 1.3 Major situational differences among the BRC registers

Conversation News reportage Fiction Academic discourse

Mode Spoken Written (+ reported dialogue) Written (+ fictional dialogue) Written

Interactiveness and online Yes Only in reported dialogue Only in fictional dialogue No
production
Shared immediate situation Yes No No No

Main communicative Personal Information Pleasure reading Information/

purpose communication argumentation/
explanation

Audience Individual Wide-public Wide-public Specialist

Dialect domain Local National Global Global

(adapted from Biber et al. 1999: 16).

21
10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek
22 Emotion Talk Across Corpora

• Pleasure, excitement, and elation (adj, adv, n) & Please and amuse (v)
• Appreciation and gratitude (adj, adv, n)
• Positive impatience, enthusiasm, and alertness (adj, adv, n)
• Sadness, distress, and despair (adj, adv, n) & Upset, distress, and
humiliate (v)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

• Confusion, anxiety, and worry (adj, adv, n) & Confuse and
bewilder (v)
• Irritation and anger (adj, adv, n) & Anger and annoy (v) & Dislike and
hate (v)
• Embarrassment and humiliation (adj, adv, n)
• Fear and panic (adj, n) & Frighten and shock (v)
• Insecurity and loss of composure (adj, adv, n)
• Surprise, shock, and amazement (adj, adv, n) & Surprise and
impress (v)
• Envy and jealousy (adj, n)
• Love, respect, and goodwill (n) – Like, love, value, and enjoy (n, v)
• Compassion and forgiveness (n)
• Antagonism (n)
• Be concerned and care (v)
• Desire and want (adj, adv, n, v)

The result was a list of over 1500 lexical items, of which some items
were excluded after careful scrutiny for a variety of reasons.7 Even though
the final list of 1060 potential emotion terms (see Appendix A 1.3 online)
is more comprehensive than many other lists, it is not argued that it is
exhaustive, but rather that it contains a large selection of British English
emotion terms.
The second step in the analysis of affect lies in determining which
of these potential emotion terms actually occur and in which meaning,
since it is only the ‘emotion’ meaning of a given form that is relevant.
This issue – the problem of polysemy and homonymy (see for example
Kilgarriff 1997b on polysemy) – lies at the heart of specifically lexical
sense-sensitive corpus analysis: ‘if a corpus is not annotated for sense
it is not possible to quantify sense distributions and if the corpus is
a large one, annotating each polysemous item for sense is not practi-
cal.’ (Neale 2006: 147). Since computer software (in my case the Zurich
BNCweb interface http://escorp.unizh.ch/, which allows different types
of searches of the BNC, and my subset of it, the BRC) does not recog-
nize meaning when searching for words, it will list all occurrences of a
given word form, regardless of its meaning. For example, a computer

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Analyzing Language and Emotion 23

program looking up afterglow will come up with all occurrences of the

form afterglow, referring to:

1 the light that is left in the sky after the sun has set
2 a pleasant feeling after a good experience

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(Oxford Advanced Learner’s Dictionary, henceforth OALD).

Only meaning 2 would be included as denoting affect/emotion, whereas

all occurrences of meaning 1 would need to be excluded. Since auto-
mated tools simply cannot reliably recognize semantic differences at
present,8 the classification of all words according to meaning had to
be done manually. In other words, all occurrences of the 1060 emotion
terms were classified by me as denoting emotion (or not). This is a rather
cumbersome method, but comes much closer to finding out about what
we are interested in than would a purely automatic computer analysis.
As Kilgarriff notes:

A computer word count program defines a word as any string of char-

acters separated by blanks or punctuation. … [T]hat makes counting
easy and has the advantage that everyone knows where they stand
and will arrive at the same numbers. The disadvantage, of course, is
that it doesn’t tell the truth. … Any step towards the truth (as lin-
guists strive to define it) tends to be a step away from anything that
is computationally straightforward.
(Kilgarriff 1997a: 144).

At the same time, this means that the analysis is less easily replicable than
a computationally straightforward, automated corpus study, and retains
some elements of subjectivity in the interpretation of the meaning of
emotion terms. The result of this manual analysis is a sense frequency
list, not a word frequency list.
Concerning cases where the same word form can realize different parts
of speech (for example love, hate as noun or verb), this was only a small
problem since all of the BNC (and therefore also the BRC) is POS-tagged
using the CLAWS system (Garside 1987). This tagging has an accuracy
rate of 96.5 per cent (Leech et al. 2001: 14), which means that about
3.5 per cent of mistakes remain (see Sinclair 2004d: 81 on this problem).
In any case, my interest lies more in reporting tendencies than exact fig-
ures, taking into account the subjective nature of the meaning-sensitive
analysis reported above.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

24 Emotion Talk Across Corpora

It remains to be pointed out that emotion terms and emotion talk

do not necessarily stand for or represent the speaker’s or others’ ‘real’
internal affective state; rather emotion talk represents what Galasiński
(2004: 6) calls a discursive practice (compare also Edwards 1999), and
‘reflects what one displays to others either in a conscious and deliber-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

ate manner, or as a result of habit of expression that accumulates with
experience’ (Anderson & Leaper 1998: 410). To give an example from
contemporary culture, film critic Philip Lopate has pointed out (in an
interview included on the DVD) that in Noah Baumbach’s 2005 film The
Squid and the Whale, the father (Bernard Berkman) keeps using emotion
talk – statements such as That hurt me or I feel bad now – to manipu-
late others, while remaining strangely detached from his own emotional
experience.

1.7 Outline of this book

To sum up the most important aims of this book again, these are:

• to examine emotion terms and their patterns in terms of register

variation (emotion profiling);
• to develop appraisal theory;
• to analyze the functions of emotion talk in the different
registers.

The hope is that the combination of a functional approach (Martin’s

systemic-functional appraisal theory) with two corpus-linguistic
approaches (Biber’s theories concerning register variation, and Hunston’s
local grammar approach, which will be described in Chapter 3) will result
in a development of present studies on emotion talk.
Chapter 2 outlines the results of the large-scale corpus investigation of
emotion profiles in the four registers in terms of lexical variation, part of
speech variation and syntactic variation. Chapters 3 and 4 deal with the
local grammar approach to affect, and outline the most important affect
patterns and their functions in conversation, news reportage, fiction and
academic discourse (lexico-grammatical emotion profiling). Chapter 5
develops the modified version of appraisal theory that was applied in
the manual analysis of the 85,000-word subset of the BRC. The results
of this exploratory analysis are described in Chapter 6. Hence, corpus
linguists may be particularly interested in Chapters 2 to 4, and systemic
functional linguists in Chapters 5 and 6.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Analyzing Language and Emotion 25

Notes
1. In this book I use the terms emotion and affect more or less interchangeably
as referring to emotional experience. I will, however, consistently employ
the terms emotion talk, emotional talk and emotion terms (rather than, say,
affect talk, affective talk and affect terms). Nevertheless, it is also necessary

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

to use affect, because this book makes reference throughout to the system
of affect as recognized in appraisal theory (Section 1.4). In more narrow
definitions, distinctions are made between affect and emotion, emotions and
feelings, moods, emotional attitudes, emotional traits, emotional disorders,
emotional plots, emotion-related states and so on (Ortony et al. 1987, Ekman
1992, Caffi & Janney 1994: 327–8, Anderson & Leaper 1998: 426, Jahr 2000:
9–10, Downes 2000: 102–3, Mees 2006: 15–16, Oatley et al. 2006: 29–30).
2. Dichotomies always simplify and emotions involve many aspects at the same
time: they are both biological and cultural, personal and social and so on (see
for example Planalp 1999: 134). It may be worth noting that neurological
research has shown that the juxtaposition of emotion vs. ratio is misguided,
with human rationality depending crucially on emotion (Turner & Stets
2005: 21–2). For a discussion of the debate concerning emotion vs. ratio
see Milton & Svašek (2005: 2–4).
3. Since there is such a wealth of studies on language and emotion (for exten-
sive bibliographies, see Arndt & Janney 1987, Besnier 1990, Janney 1996,
van Dijk 2001), it was necessary to exclude from this overview: (a) studies
that focus on aspects such as intonation, prosody, pitch, facial expression,
gazing patterns, gesture (for example Arndt & Janney 1987, Selting 1994);
(b) most studies that focus on languages other than English (for example
Fiehler 1990, Haviland 1991, Fries 1995, Günther 1997, Jahr 2000, Koven
2004); (c) much non-linguistic (for example psychological, philosophical,
sociological, anthropological) emotion research. It must also be noted that
there are many overlaps, and that only the most important approaches are
covered; more recently, corpus-based methodology has been used in some
of these studies (Teubert 2004a, b, Dem’jankov et al. 2004).
4. The term stance is also closely associated with analyses of academic dis-
course (for example Hyland 1999 and references in Bednarek 2006a), and
there is also a research project on analyzing stance in spoken American
English (http://www.ekl.oulu.fi/stance/index.html, accessed 27 September
2005), with a focus on intersubjectivity and conversation analysis. Such
studies usually include more than the analysis of affect (for example modal-
ity, evidentiality, evaluation), and are therefore only partly relevant to the
research undertaken here.
5. The notion of register overlaps with related concepts such as genre, domain, or
text type (compare Lee 2001). For systemic-functional notions of genre and
register cf. Martin (1993), Halliday & Matthiessen (1999), Martin & Rose
(2007).
6. Nöth (1992) gives an overview of lists of emotion terms from psychologi-
cal research. Most such research is intuition- and/or informant-based (for
example using elicited or free-listed emotion terms), with some also using
information from dictionaries and thesauri or previous research. The num-
ber of emotion terms identified by this research varies depending on how an

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

26 Emotion Talk Across Corpora

emotion term is defined: Moore et al. (1999) mention 415 emotion terms,
Johnson-Laird & Oatley (1989) list 590 words, and Storm & Storm (1987) use
787 different terms, while Wallace & Carson (1973) and Oatley et al. (2006:
183) both mention a list of over 2000 emotion terms, and other figures also
appear in the relevant research. In a corpus linguistic study, Precht (2000)
included 366 different word-forms indicating attitudinal stance, but these

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

include both expressions with evaluative meanings (for example awful, bitch)
and emotion terms (adore, angry). In a sense, any list of emotion terms will
to a certain extent remain subjective, as emotion terms – like other lexical
items – make up a fuzzy set with prototypical, core, and marginal members
(Section 5.3.2.2).
7. Words in the ET that were on the border of affect were excluded (for example
alert, alive, be sure of ) – see Appendix A 1.2 online. Similarly, references to
behaviour associated with emotion were not included, since the focus was
on mental disposition terms rather than behavioural surge terms (see Section
1.4 above). Others had to be excluded for purely methodological purposes;
for instance, they were too polysemous, too difficult to analyze objectively,
or occurred too often to analyze manually (for example the modal verb will
as indicating volition vs. other modal meanings).
8. One possible solution would be to use corpus analysis in order to identify
the most common words that occur in the context of afterglow when used
in meaning 2, a methodology similar to the one used by Teubert (2004a),
or to use a collocation dictionary for the same purpose when the number of
words analyzed is too large to investigate separately. It would then become
possible to search only for occurrences of afterglow in the context of these
common collocates. However, not all emotion terms are listed in all their
meanings in collocation dictionaries such as the Oxford Collocations Dictio-
nary, and the analysis would still not be completely reliable. Watters (2002)
shows that cluster analysis (grouping word senses on the basis of their collo-
cates) works in 69 per cent of classifications for primary word senses, but only
in 25 per cent of classifications for secondary word senses. See also Kilgarriff
(1997b) for an overview of word sense disambiguation in natural language
processing. A related approach is to identify the grammatical frame in which
a lexical expression is used to exclude non-affective meanings (Precht 2000,
2003). For example, afraid + prepositional phrase (I’m afraid of spiders) indi-
cates fear, whereas afraid + that-clause (I’m afraid that’s impossible) refers to
the speaker’s thoughts (Precht 2000: 43) and is a more formulaic usage. But
see Werth (1998) who argues that the usage of emotion terms as indicating
either ‘genuine’ emotion or as a conventional expression does not depend
on grammatical frames alone.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

2
Emotion Profiling

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

In this chapter I discuss the results of the analysis of the emotion
profiles of the registers of academic discourse, conversation, fiction
and news reportage with respect to lexical, part-of-speech (POS) and
syntactic variation concerning emotion terms. I use the term emo-
tion profiling to refer to the process of investigating these frequency
profiles. As mentioned in Section 1.6.3, this is based on a combi-
nation of automated computer analysis (using the Zurich BNCweb
interface at http://escorp.unizh.ch/) and careful manual scrutiny of the
data. It must also be emphasized that the research reported in this
chapter is more corpus-based than corpus-driven, in Tognini-Bonelli’s
(2001) terms.
Recent research has shown that linguistic choices (including lexico-
grammatical patterns or chains of word-forms) are mainly motivated
by function, resulting in many linguistic differences between registers
(Biber et al. 1999, Reppen et al. 2002b, Stubbs & Barth 2003). However,
little systematic linguistic research has considered the question of regis-
ter variation in the usage of emotion terms (with the exception of stance
analysis, for example Precht 2000, 2003). At the same time, it is noted
that linguistic affect may differ across social groups (Besnier 1990: 435),
discourse types (Daneš 1987: 177) and ‘sub-cultures’ (Daneš 2004: 30). As
Daneš suggests, ‘it is just this “sub-cultural” research that could represent
a very interesting and rewarding field of study’ (Daneš 2004: 30). Simi-
larly, Hyland notes that ‘[o]ur knowledge of how evidentiality and affect
are typically expressed in different registers is limited’ (Hyland 1999:
105). This chapter aims to shed some light on these issues by means of
an emotion profiling of casual conversation, news reportage, fiction and
academic discourse.

27

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

28 Emotion Talk Across Corpora

2.1 Key words in the registers

Before describing the results of the emotion profiling, let’s look more
closely at the contents of the BRC with the help of a key word analy-
sis. A key word analysis (Scott 1997, Scott & Tribble 2006) allows the

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

researcher to compare corpora in terms of their key words, that is, words
that are statistically significant in terms of frequency. Usually, a smaller
corpus (the node corpus) is compared with a larger corpus (the refer-
ence corpus). A key word analysis highlights register-specific features:
it provides information about content and other characteristics of the
corpus (for example style features, cultural background) that may be rel-
evant when interpreting corpus findings. Here we are mainly concerned
with content, in order to ascertain whether any of the sub-corpora are
unusual (‘skewed’) in terms of content and thus not suitable for analysis.
A preliminary key word analysis was thus undertaken before the emotion
profiling. In each case, the respective node corpus (academic discourse,
news reportage, fiction and conversation) was compared against the BRC
(used as reference corpus). In Table 2.1 (on pages 29 to 30) the most fre-
quent key words (see also Appendix A 1.1.6 online) are sorted according
to part of speech, though in some cases this is not clear (for example study
can be both noun or verb, like can be preposition, conjunction, verb, or
pragmatic marker). Such key words are enclosed in brackets in Table 2.1.
As can be seen from the table, the respective corpora do not appear
to be greatly skewed in terms of content, and conform very much to
what we would expect to find in the respective register (compare for
example the descriptions in Biber et al. 1999): academic discourse
includes both words from the hard and the soft sciences, though there is
perhaps a slight preponderance of key words belonging to medical dis-
course ( patients, cells, disease, acid, gastric). It does, however, include a
large number of words relating to academic discourse in general (shown,
described, associated, figure, analysis, study, data, information, different,
important, particular, specific, possible, significant, similar, general). (Note
also that there is a large overlap with the top word-forms listed for
academic discourse by Stubbs & Barth (2003: 65), based on different
corpora of academic language.) News discourse is dominated by nouns
and adjectives referring to cities, countries and people (Britain, Germany,
American, Soviet, Europe, west, east, David, John, London, Hong Kong) as
well as by nouns that are part of commercial and sports discourse ( pound,
cent, million, market, shares, sales, season, team, league, players, club,
game, champion), and words from the world of entertainment (TV, star).
These represent different sections of the news such as home/foreign

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
Table 2.1 Key words in the BRC sub-corpora
Corpus Grammatical Nouns Verbs Adjectives Adverbs
words
Academic of, the, in, which, patients, formula, example, data, (study), is, are, may, social, different,
discourse by, or, these, such, system, (form), fig, analysis, language, be, has, shown, important, particular,
between, also, cells, disease, (figure), information, using, described, specific, possible,
however, this, as, (studies), (results), section, (effect), associated, significant, gastric,
thus, therefore, (number), treatment, development, required individual, similar,
an, within, both, (act), cases, areas, level, (state), groups, general
each, per (function), evidence, process, systems,
education, DNA, structure, (order),
(control), (use), activity, case, acid,
(values), species, population,
women, chapter

News by, per, pound, Mr, year, government, cent, has, will, win last, new, former, west, yesterday
reportage for, the, after, million, Britain, world, market, (shares), international, east,
who, its, against, company, London, England, European, united, first, German,
their season, city, chairman, team, union, British, American,
David, Minister, President, league, dollar, Soviet, financial,
players, John, spokesman, (profits), national
(party), companies, Hong, month,
Kong, labour, director, secretary, club,
game, (match), Europe, tax, (bid), (star),
sport, TV, (bank), years, conference,
page, week, price, Germany, manager,
correspondent, edition, sales, cup,
corpus, (cash), leader, industry,
university, Smith, police, (share),
investors, group, (final), billion, Tory,

29
Champion

(Continued)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
Table 2.1 (Continued)

Corpus Grammatical Nouns Verbs Adjectives Adverbs

30
words
Fiction her, she, he, his, eyes, face, man, (head), room, door, had, was, looked, dark suddenly
him, my, me, (back), voice, father, hand, hands, said, knew, turned,
herself, into, at, mother, moment, woman, could, felt, seemed,
away, himself, down, (smile), hair, feet, sir, mouth, bed, asked, would, stood,
out, again, never, Corbett, house, lips, arms, smiled, came, sat,
nothing, too, behind, window, silence, floor, girl, thought, took,
across, up, around, Mrs, Athelstan nodded, saw, shook,
beside, someone, began, stared, walked,
towards heard, told, looking,
laughed, am, replied,
rose, tell, opened,
pulled, wondered,
wanted, glanced, hadn

Conversation I, you, yeah, it, bit, mum, (sort), thing, lot, dad t, s, got, know, right, alright, just, really,
oh, well, what, no, don, ve, do, nice, okay, actually
that, they, yes, re, get, ll, bloody, good,
cos, we, there, go, think, gonna, fucking
so, your, one, (like), mean,
then, me, them, can, want, going,
up, here, twenty, put, see, didn,
if, five, all, haven, come, isn,
hundred, anyway, look, say, have,
why, now, aye, did, gotta, doing,
my, he, something, goes, d, doesn,
(like), (er, won, wanna,
mm, erm, m, ah, innit, done,
ooh, mhm, eh) says, said,
ain, wouldn

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 31

news, commerce/finance, and sports discourse, arts/cultural material,

lifestyle/leisure/belief. Fiction is characterized by first and third person
singular pronouns (her, she, he, his, him, my, me) – reflecting the fact
that most fiction contains either first- or third-person narrative. It also
includes words referring to human appearance (eyes, face, head, voice,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

smile, hair, feet, mouth, lips, arms) and human behaviour (specifically
interaction), such as: looked, said, knew, turned, felt, asked, stood, smiled,
came, sat, thought, took, nodded, saw, shook, began, stared, walked, heard,
told, looking, laughed, replied, rose, tell, opened, pulled, wondered, wanted,
glanced. This is probably attributable to the fact that the physical appear-
ance of characters needs to be described in order for readers to get a
mental ‘picture’ of them. Furthermore, human interaction seems to be
a key feature of most narrative fiction, and both verbal and non-verbal
behaviour are an important part of its description. (Note again the con-
siderable overlap with the most frequent word-forms listed by Stubbs &
Barth 2003 for their FICTION corpus.) Finally, in conversation, we find
personal pronouns referring to speaker and listener (I, you), a range of
words belonging to evaluation in a broad sense, some of which may func-
tion as response words (yeah, yes, right, alright, nice, okay, good, bloody,
fucking), many hesitation/response phenomena (er, mm, erm,m, ah, ooh,
mhm, eh), discourse markers and downtoners/upgraders (well, anyway,
just, really, actually), vague language (sort, thing, lot, bit, something), con-
tracted forms (t, s, ve, don, didn, haven, isn, re, ll, did, doesn), colloquial/
informal words (mum, dad, gonna, gotta, wanna, innit, ain), words refer-
ring to mental and verbal processes (know, think, mean, say, says, said, pos-
sibly goes) and modality (can, want, possibly going (to)), and what we may
call basic words referring to human behaviour ( put, see, come, look, have,
doing, done). Again, none of the words are particularly unexpected. This
indicates that the corpus is fairly well balanced in terms of content, and
that there is, presumably, no great bias towards a particular topic/content
which might influence the findings to any significant extent.

2.2 Lexical variation

Let’s now move on to considering lexical variation of emotion terms,

both in terms of their frequency and in terms of their distribution.1

2.2.1 Frequency of emotion terms

The first analysis concerns the frequency of emotion terms in each sub-
corpus. As said, this is an examination of frequency of meaning rather
than form since ‘non-affect’ meanings of lexical items were excluded

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

32 Emotion Talk Across Corpora

from the analysis (Section 1.6.3). No cut-off level was used below which
occurrences were to be disregarded. Rather, all occurrences of emotion
terms (if denoting emotion) contributed to the overall sum of emo-
tion term occurrences in the given sub-corpus. This figure is thus a
‘composite variable’ (Precht 2000: 53–4; compare also Kaltenbacher’s

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

2007 virtual frequency ranks). This decision was taken because I am inter-
ested in how frequent affect is in general, and each individual emotion
term contributes to this overall picture. Even though fewer occurrences
render results less reliable from a statistical point of view, this is out-
weighed by the fact that as many as 1060 words were included in the
quantitative analysis. Likewise, dispersion could not be investigated sys-
tematically, meaning that some frequency findings might be partially
distorted (see Appendix A 1.2 online). Bearing these caveats in mind,
Table 2.2 shows the distribution of all 1060 emotion terms in each
sub-corpus (ppm = occurrences per million).2

Table 2.2 Emotion terms in the BRC

Conversation News reportage Fiction Academic discourse

Raw freq ppm Raw freq ppm Raw freq ppm Raw freq ppm
40,570 9,645.6 15,917 6,090.5 72,170 10,790.2 12,991 2,179.4

From Table 2.2, it initially appears that emotion terms are most fre-
quent in fiction, followed by conversation, news reportage and academic
discourse. However, it must be pointed out that three emotion terms are
uncharacteristically frequent in conversation and skew the analysis to a
very large extent: the verb want (13,060 occurrences), the adjective sorry
(10,787 occurrences) and the verb like (6,353 occurrences) (see Table A.8
in Appendix A 2.1 online). Furthermore, these terms are so frequent
that even a manual analysis based on extrapolation (see Appendix A
1.2 online) was not possible (10 per cent of all occurrences for each of
these terms still make up more than 500 occurrences), and no manual
analysis could therefore be undertaken. This means that no meaning
distinctions were taken into account. Moreover, it might be argued that
these terms are, as it were, ‘bleached’ (Martin & White 2005: 85) to a
certain extent, in that they are not really – or at least not primarily –
used to talk about our emotions, but rather to be conventionally polite,
for example to apologize, to disagree or to introduce bad news (see OALD
entry for sorry), to make offers, invitations or demands (see OALD entry
for want) or to make polite demands and express evaluations (see OALD
entry for like). For these reasons, and since the interest is not so much in

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 33

these conventionalized usages of emotion terms, but rather in their more

‘literal’ emotive uses, sorry, want and like were excluded. The results for
this re-calculation are shown in Table 2.3.

Table 2.3 Emotion terms excluding sorry, want, like

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Conversation News reportage Fiction Academic discourse
Raw freq ppm Raw freq ppm Raw freq ppm Raw freq ppm
10,370 2,465.5 12,972 4,963.6 54,402 8,133.7 11,444 1,919.8

As Table 2.3 suggests, news reportage includes more emotion terms

than conversation, in contrast to the calculation above, where the oppo-
site was true. The rank order for fiction and academic discourse remains
stable: fiction contains the most emotion terms (ca 8000 per million),
followed by news reportage (ca 5000 per million), conversation (ca 2500
per million) and academic discourse (ca 1900 per million). This is in stark
contrast to the findings of Biber & Finegan (1989), where 90 per cent of
General fiction, 92 per cent of Adventure fiction, 85 per cent of Mystery
fiction, and 100 per cent of Science fiction are included in the clus-
ter named ‘faceless stance’ (Biber & Finegan 1989: 103), because they
exclude non-authorial affect from their analysis (for criticism of this see
Watson 1999: 226–7). It is therefore in the realm of fiction that the lan-
guage of emotion seems to be most alive. This is not surprising since
fiction deals with the description of human characters, the relationships
and interactions among them, their emotions and emotional reactions.
This also suggests that prose fiction constitutes an important arena for
the mediation of emotions in English (together with newspaper stories,
films, TV shows, and other popular media outlets which cannot be ana-
lyzed here). Arguably, it is through these cultural constructs that we at
least partly come to make assumptions (in our adult life) about how we
ought to behave emotionally, what kinds of emotions we are ‘supposed’
to feel in certain situations, what an emotional experience entails and
so on. In cognitive linguistic terms, fiction contributes to our construal
of emotion schemata (see Sections 1.3; 5.3.1). In sociological terms, works
of fiction are cultural products that play a part in construing feeling rules
(Section 1.3). Our emotional responses are thus to a certain extent not
naturally given but rather culturally mediated, and fiction certainly plays
a role in this cultural mediation. However, much more research is neces-
sary to determine the extent to which emotional responses are socialized
(or culturally transmitted) through the personal (face-to-face interaction)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

34 Emotion Talk Across Corpora

personally (face-to-face)

culturally transmitted
(socialization)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

impersonally (via mass media, …)
Emotional responses

innate/universal

Figure 2.1 Nature vs. nurture

rather than through the impersonal (fiction, mass media3 ), and to what
extent they are innate (Figure 2.1).
As mentioned in Section 1.3, there is a large body of research on the
question of innateness vs. cultural construal of emotional experience,
which seems to allow the conclusion that emotions involve both aspects.
A significant body of research has also investigated emotion and art (van
Meel 1994, Oatley 2003, Robinson 2005), but without paying specific
attention or systematically investigating the role of cultural artefacts
(such as novels, paintings, films) in the socialization of affect, though
Painter argues that the reading of children’s literature helps children to
learn ‘the conventions of visual representations of affect as well as gain-
ing experience in the role of affect construal in story telling’ (Painter
2003: 197), and that literacy is influential in children’s development of
an emotion vocabulary (Painter 2003: 206). More informally, Planalp
states that:

. . . the love each of us feels seems to fit, more or less, the expecta-
tions of the dominant culture. Somehow most people ‘fall in love’
with someone of the appropriate sex, age, social class, and marital
status; exceptions are unusual, frowned upon, or viewed as decidedly
immoral. We are engulfed in messages about love – love songs on the
radio, talk about love with friends, lectures about love from parents,
TV sitcoms, magazines, romance novels, love stories at the movies . . . .
All these messages shape our own unique love.
(Planalp 1999: 135)

However, in general terms, we need more precise investigations of how

emotion cultures are learned, and ‘a better explanation of how [the

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 35

elements of emotion cultures] are used strategically in the micropolitics

and microeconomics of all encounters’ (Turner & Stets 2005: 297).
Moving on to news reportage now, the relatively high frequency of
emotion terms in this register was also to be expected: after all, refer-
ences to emotions contribute to the news values (Galtung & Ruge 1965,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Bednarek 2006a: 16–18) of personalization and colour. In fact, an even
higher frequency of emotion terms might have been expected, since the
corpus contains tabloid as well as broadsheet newspapers, and the for-
mer are characterized in particular by their frequent references to news
actors’ emotions (Bednarek 2006a: 194). Concerning the relatively low
frequency of emotion terms in conversation, this is somewhat surpris-
ing: after all, it could be intuited that speakers choose to reveal more
about their emotional reactions in face-to-face conversations than in
written discourse. On the other hand, Planalp does suggest that people
rarely communicate feelings in Western cultures (Planalp 1999: 211).
However, it must also be mentioned that speakers may resort to a vari-
ety of other means of communicating emotionality, that is, expressing
rather than denoting affect/emotion (emotional rather than emotion
talk; compare Section 1.3). For example, in Burger & Miller’s data,
speakers employ ‘emotion state words in only one-third of their anger
narratives’ (Burger & Miller 1999: 137), even though this may vary across
social class (Burger & Miller 1999: 154–5). The fact that speakers have
access to many different means of signalling emotionality may explain
the relatively low frequency of emotion terms in conversation. With
respect to academic discourse, the finding that this register contains the
least frequent use of emotion terms of all registers is not unexpected,
although the fact that it contains as many as 1900 words per million is
something of a surprise, as an even lower frequency might reasonably
have been expected. For instance, the analyses by Hood (2005: 32) show
that research articles strongly prefer appreciation over affect.

2.2.2 Most frequent emotion terms

Again excluding want, sorry and like, the ten most frequent emotion
words in each sub-corpus are listed in Table 2.4. With regard to these lexi-
cal items, conversation has a clear verbal–adjectival affect style (six of the
ten most frequent emotion terms are verbs, four are adjectives), and aca-
demic discourse has a distinct nominal affect style (nine nouns), whereas
both news reportage (five nouns, two verbs, three adjectives) and fiction
(four verbs, four nouns, two adjectives) are more varied in their affect
style, with news reportage nevertheless tending towards a nominal, and
fiction tending towards a verbal–nominal, affect style. However, if we

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
36
Table 2.4 Ten most frequent emotion terms in the BRC sub-corpora (raw frequency)

Conversation News reportage Fiction Academic discourse

1 love (V) 1,185 fear (N) 402 love (V) 2,106 feeling (N) 425
2 hate (V) 647 happy (A) 386 love (N) 1,358 fear (N) 410
3 mind (V) 635 love (V) 367 happy (A) 1,152 concern (N) 385
4 worry (V) 627 hope (N) 360 enjoy (V) 1,041 anxiety (N) 365
5 enjoy (V) 515 love (N) 272 fear (N) 927 expectation (N) 356
6 happy (A) 504 enjoy (V) 205 feeling (N) 920 desire (N) 353
7 care (V) 332 concern (N) 191 worry (V) 899 stress (N) 334
8 glad (A) 281 surprise (N) 186 hate (V) 854 love (N) 317
9 worried (A) 245 prepared (A) 172 surprised (A) 800 hope (N) 217
10 surprised (A) 232 angry (A) 168 pleasure (N) 776 concerned (A) 215

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 37

extend the analysis to the 50 most frequent emotion terms in each cor-
pus (see Appendix A 2.2 online), it appears that although the findings
for the other registers remain more or less the same, fiction is in fact
characterized by a nominal–adjectival affect style (Table 2.5). These ten-
dencies will be explored in more detail later on in this chapter, based on

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

the analysis of all emotion terms (Section 2.3.1).

Table 2.5 Part of speech distribution of 50 most frequent emotion terms in each
sub-corpus

Sub-corpus No. of Adjectives No. of Nouns No. of Verbs

Conversation 28 6 17
News reportage 19 25 6
Fiction 17 20 13
Academic discourse 8 35 7

Of the ten most frequent emotion terms none occurs in all four cor-
pora, but love (V), enjoy (V) and happy (A) occur in three sub-corpora
(conversation, news reportage, fiction) as do the nouns fear and love
(news reportage, fiction, academic discourse). Conversation and fiction
share the verbs hate and worry as well as the adjective surprised; news
reportage and academic discourse share the nouns hope and concern; and
fiction and academic discourse have in common the noun feeling. Fre-
quent only in conversation are mind (V), care (V), glad (A) and worried
(A); in news discourse we can find surprise (N), prepared (A) and angry (A);
pleasure (N) is common in fiction, and in academic discourse the nouns
anxiety, expectation, desire, stress and the adjective concerned occur fre-
quently (Table 2.6 on page 38).4 (Note that horizontal lines here group
items shared in three corpora, two corpora, or appearing only in one
corpus.)
Even though some emotion terms occur in three of the sub-corpora,
there is a considerable amount of lexical variation concerning the ten
most frequent emotion terms in each corpus. If we again extend the
analysis to the 50 most frequent words in each sub-corpus, the lexical
variation hypothesis is confirmed: only one noun is shared among all
four corpora (shock), and only eleven adjectives, ten nouns and two verbs
are shared among three of the four corpora (of 39 adjectives, 49 nouns,
24 verbs, and 0 adverbs). For now, I want to postpone the discussion of
possible reasons for this variation, but I shall come back to this below
when exploring part-of-speech variation.5

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

38 Emotion Talk Across Corpora

Table 2.6 Shared emotion terms

Emotion term Conversation News Fiction Academic

reportage discourse
love (V)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

enjoy (V)



happy (A)



fear (N)



love (N)



hate (V)


worry (V)


surprised (A)


hope (N)


concern (N)


feeling (N)


mind (V)

care (V)

glad (A)

worried (A)

surprise (N)

prepared (A)

angry (A)

pleasure (N)

anxiety (N)

expectation (N)

desire (N)

stress (N)

concerned (A)

2.3 Part-of-speech variation

2.3.1 Distribution of POS

Since the BNC, and therefore also the BRC are tagged corpora, it
is relatively straightforward to analyze part-of-speech variation, even
though an error margin remains (Appendix A 1.3 online shows the POS
classification of all terms). Comparing the overall occurrences of all emo-
tion terms (excluding sorry, want, like) we find the distribution in Table
2.7. Looking at all analyzed emotion terms in Table 2.7 (on page 40),
it now becomes possible to identify a distinct POS affect style for each
register: conversation is characterized by a verbal–adjectival affect style
(with more verbs than adjectives); both news reportage and fiction have a
nominal–adjectival affect style (with more nouns than adjectives but also
a considerable number of verbs, particularly in fiction); and academic
discourse has a clear nominal affect style.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 39

It may be instructive to compare these findings with some related

research. The findings for conversation stand in contrast to the adjec-
tival stance style described for conversation by Precht (2000) (but note
again that the results are not directly comparable in terms of methodol-
ogy, corpus design and investigated features), but they are in line with

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Biber et al.’s findings that ‘it is more common for the spoken texts to
refer to mental states as verbs and adjectives [rather than using nouns]’
(Biber et al. 1998: 64). The results for fiction also compare with Biber
et al.’s finding that -ness nominalizations (which include but are not lim-
ited to emotion nouns, for example bitterness, happiness) as well as -ment
nominalizations (which often describe emotions, for example amaze-
ment, astonishment, disappointment, embarrassment, excitement) are quite
common in fiction, even ‘though fiction has the lowest frequency of
nominalizations overall’ (Biber et al. 1998: 65). The analyses of the BRC
have shown that fiction in general has a nominal–adjectival affect style
beyond the particular kinds of nominalizations investigated by Biber
et al. However, these data should be related to what we know about
register POS variation in general: this will tell us whether the identified
affect style simply reflects a general POS style or not. Table 2.8 (on page
40) shows some findings for the POS preferences of different registers.
Viewing the findings in Table 2.8 comparatively, it would seem that
the news and academic affect style partly reflects a general POS style,
whereas the affect style in fiction and conversation reflects this to a
lesser extent. However, since these findings are based on different cor-
pora, these conclusions must remain preliminary – an investigation of
all POS distributions in the BRC was not undertaken. (Data reported for
the distribution of mental verbs, stance and so on in Biber et al. 1999
cannot be compared directly because they include more than emotion
terms. Biber 2006 was not available at the time of writing.) The following
sections describe the frequencies of emotion nouns, adjectives, adverbs
and verbs in more detail.

2.3.2 Nouns
We start by looking at the ten most frequent emotion nouns in the BRC,
as visualized in Table 2.9 (on page 41). Here we can find some overlap
between the four registers: hope, love, fear and feeling are amongst the
ten most frequent emotion nouns in all four corpora; shock and surprise
occur very frequently in three corpora (conversation, news reportage and
fiction), and worry, pleasure, concern and expectation in two. Characteristic
for the conversation register are the nouns joy, nerves, panic; for news

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
Table 2.7 Part of speech variation in the BRC

POS Conversation News reportage Fiction Academic discourse

40
Raw freq ppm Raw freq ppm Raw freq ppm Raw freq ppm
Nouns 1,009 239.9 5,730 2,192.5 21,568 3,224.7 7,478 1,254.5
Adjectives 3,904 928.2 4,490 1,718.1 17,511 2,618.1 2,319 389
Adverbs 22 5.2 118 45.1 1,159 173.3 54 9.1
Verbs 5,435 1,292.2 2,634 1,007.9 14,164 2,117.7 1,593 267.2
Total 10,370 2,465.5 12,972 4,963.6 54,402 8,133.7 11,444 1,919.8

Table 2.8 Register preferences for part of speech

Corpus Conversation News reportage Fiction Academic discourse

London-Oslo-Bergen (LOB) – – most frequent: lexical learned and scientific
(Johansson 1981: 2) verbs, but nouns also English dominated by
relatively frequent nouns and adjectives
Brown, Frown, LOB, FLOB – – fiction: verbal style learned English:
(Stubbs & Barth 2003: 79) (past tense) nominal style
LSWE corpus −nouns ++nouns +verbs +nouns
(Biber et al. 1999: 65–66; 359) −adjectives +adjectives +adverbs +adjectives
+verbs nouns > verbs −verbs
+adverbs nouns > verbs
nouns = verbs
BRC (key word analysis) many modal and lexical many nouns in key many nouns in key many nouns in key
verbs in key words list words list words list; lexical words list
verbs also frequent
BRC (affect style) verbal–adjectival nominal–adjectival nominal–adjectival nominal

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
Table 2.9 Ten most frequent emotion nouns in the BRC (raw frequency)

Conversation News reportage Fiction Academic discourse

1 shock 75 fear 402 love 1,358 feeling 425
2 surprise 67 hope 360 fear 927 fear 410
3 hope 55 love 272 feeling 920 concern 385
4 worry 52 concern 191 pleasure 776 anxiety 365
5 pleasure 48 surprise 186 surprise 733 expectation 356
6 love 35 shock 154 anger 582 desire 353
7 joy 31 horror 146 hope 491 stress 334
8 fear 30 feeling 140 shock 470 love 317
feeling 30
9 nerves 28 expectation 115 excitement 413 hope 217
10 panic 23 worry 107 pain 355 wish 175

41
10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek
42 Emotion Talk Across Corpora

Table 2.10 Shared emotion nouns

Noun Conversation News reportage Fiction Academic discourse

hope




love

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

fear




feeling




shock



surprise



worry


pleasure


concern


expectation


joy

nerves

panic

horror

anger

excitement

pain

anxiety

desire

stress

wish

reportage it is horror; for fiction anger, excitement, pain and for the aca-
demic register it is anxiety, desire, stress and wish (Table 2.10, on page 42;
here and in Table 2.15 horizontal rules group items shared between four,
three, two corpora, or appearing only in one corpus).
Since hope, love, fear and feeling are common in all corpora, they
represent common emotion nouns in English in general. It seems impor-
tant for speakers/writers to talk about what emoters hope for or not, what
they love or not, as well as what they fear (or not). Both hope and fear can
also be used to introduce ( project) speech and thought representation:

(1) The family’s hopes that Caroline was alive slumped when her visa
expired three weeks ago. (CH2 6513, BRC news reportage)
(2) Authorities have themselves acknowledged fears that ‘counter-
revolutionaries’, still at large despite a nationwide crackdown on
dissent, might try to sabotage the anniversary. (A1G 165, BRC news
reportage)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 43

Feeling is a general noun that can be employed in all sorts of constructions

(see OALD), and is therefore potentially very useful for speakers. It does
not refer to a particular emotion but can be used to introduce a variety
of emotions (as well as other feelings), for instance:

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

a feeling of/ feelings of
smugness/ expectation/ mutual respect/ wonder/ terror/ disquiet/
anger/ pleasure/ joy/ panic/ let-down/ unease/ guilt /malaise/ elation/
relief/ pride/ dread/ sadness/ satisfaction/ euphoria/ longing/ tri-
umph/ gratitude/ guilt/ anguish/ annoyance/ animosity/ antic-
ipation/ anxiety/ contentment/ depression/ desire/ desolation/
discomfort/ disgust/ dismay/ excitement/ exhilaration/ fear/ gloom/
happiness/ hatred/ homesickness/ loathing/ love/ optimism/ pity/
relief/ shock
(from BRC fiction sub-corpus)

The nouns shock and surprise are also very frequent in conversation, news
reportage and fiction but less frequent in academic discourse (which
presumably expresses unexpectedness meanings differently). In news
reportage this almost certainly relates to the news value of unexpected-
ness (Bednarek 2006a: 171), but the usage of these terms in conversation
and fiction invites further research.

2.3.3 Adjectives
Moving on to the analysis of emotion adjectives, the ten most fre-
quent emotion adjectives (excluding sorry) in the BRC are presented in
Table 2.11 (on page 44). Of these adjectives, some are again shared among
the different sub-corpora: happy, worried and surprised occur frequently
in all corpora; sad, keen, anxious and angry occur very commonly in three
sub-corpora, whereas glad, pleased, frightened, proud, prepared, concerned
and disappointed occur in two. Conversation has frequent occurrences of
fed up and scared; fiction likes afraid and academic discourse prefers will-
ing, unhappy and content (news reportage has no particular preferences:
all its ten most frequent adjectives occur in at least one other corpus as
well). Compare Table 2.12 (on page 45).
It is interesting to note that the three adjectives that are shared between
all four corpora as most frequent can be classified as positive (happy),
negative (worried) and neutral (surprised), so it appears that all options
are covered: using just these three adjectives allows speakers/writers
to express positive and negative emotional responses, and to make
reference to unexpectedness.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
44
Table 2.11 Ten most frequent emotion adjectives in the BRC (raw frequency)

Conversation News reportage Fiction Academic discourse

1 happy 504 happy 286 happy 1,152 concerned 215
2 glad 281 prepared 172 surprised 800 prepared 203
3 worried 245 angry 168 glad 764 willing 157
4 surprised 232 worried 150 pleased 679 happy 131
5 pleased 207 concerned 126 angry 650 anxious 128
6 fed up 149 keen 125 afraid 613 angry 73
7 sad 130 proud 118 worried 439 unhappy 55
8 keen 124 sad 113 frightened 385 worried 53
9 frightened 101 anxious 111 anxious 378 keen 44
disappointed 111 surprised 44
10 scared 83 surprised 106 proud 333 content 42
sad 333 disappointed 42

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 45

Table 2.12 Shared emotion adjectives

Adjective Conversation News reportage Fiction Academic discourse

happy




worried

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

surprised




sad



keen



anxious



angry



glad


pleased


frightened


proud


prepared


concerned


disappointed


fed up

scared

afraid

willing

unhappy

content

2.3.4 Adverbs
Adverbs are in general the least frequent word class concerning emotion
terms. The occurrences of even the ten most frequent emotion adverbs
may therefore be too low to be considered indicative of register variation.
I nevertheless want to report on them here for the sake of complete-
ness. The ten most frequent emotion adverbs are listed in Table 2.13 (on
page 46).
The figures in Table 2.13 (on page 46) point to a general preference
for happily, desperately and passionately across corpora. However, since
the raw occurrences are very infrequent, these tendencies require much
more research, so no more shall be said about emotion adverbs here.

2.3.5 Verbs
In contrast to adverbs, verbs are relatively frequent in all sub-corpora, and
the findings are therefore more indicative of register variation. Table 2.14
lists the ten most frequent emotion verbs (excluding want and like) in
the BRC. Let’s again compare which verbs are shared among the four

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
46
Table 2.13 Ten most frequent emotion adverbs in the BRC (raw frequency at least 2)

Conversation News reportage Fiction Academic discourse

1 happily 11 happily 43 happily 168 happily 20
2 desperately 4 desperately 25 desperately 132 sadly 7
3 cheerfully 2 passionately 9 sadly 131 passionately 6
passionately 2
4 cheerfully 8 cheerfully 110 gratefully 5
furiously 8
5 sadly 7 gratefully 75 desperately 3
unhappily
6 miserably 4 furiously 69 joyfully 2
gloomily 2
7 gratefully 3 excitedly 63
8 blissfully 2 gloomily 60
9 miserably 44
10 expectantly 38
fearfully 38

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 47

Table 2.14 Ten most frequent emotion verbs in the BRC (raw frequency)

Conversation News reportage Fiction Academic discourse

1 love 1,185 love 367 love 2,106 love 109
2 hate 647 enjoy 205 enjoy 1,041 enjoy 98

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

3 mind 635 worry 146 worry 899 desire 91
4 worry 627 hate 95 hate 854 care 81
5 enjoy 515 care 83 care 747 worry 80
6 care 332 admire 66 mind 636 value 77
7 fancy 187 upset 50 admire 314 dislike 58
8 miss 133 anger 46 miss 300 admire 46
9 bother 120 shock 43 hurt 265 resent 41
surprise 43
10 annoy 94 impress 42 long 261 appreciate 38

sub-corpora of the BRC: love, worry, enjoy and care occur in all regis-
ters; hate and admire occur in three sub-corpora; and mind and miss are
amongst the ten most frequent verbs in two sub-corpora (conversation
and fiction). Fancy, bother and annoy are characteristic for conversation,
upset, shock, surprise, impress and anger are particularly frequent in news
reportage; fiction prefers long and hurt and in academic discourse we find
a frequent usage of desire, value, dislike, resent and appreciate (Table 2.15).
On closer inspection, it becomes apparent that, when using emotion
verbs, speakers talk about someone ‘loving’ or ‘enjoying’ some-
thing/someone (positive) or perhaps caring about something/someone
(neutral) as well as ‘worrying’ (negative). These verbs are used to indi-
cate likes and dislikes, preferences and anxieties – key emotions that
may be relevant to a variety of circumstances, and which are also used
in conversational formulae.

2.4 Summary of lexical and POS variation

Summing up the findings so far, it appears that a number of emotion

terms are frequent overall, occurring in all or at least three of the sub-
corpora of the BRC, and also seem frequent in the BNC overall. These are
listed in Table 2.16 (Appendix A 2.3 online details which of their mean-
ings were included). These overlap partly with emotion labels that have

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

48 Emotion Talk Across Corpora

Table 2.15 Shared emotion verbs

Verb Conversation News reportage Fiction Academic discourse

love




worry

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

enjoy




care




hate



admire



mind


miss


fancy

bother

annoy

upset

shock

surprise

impress

anger

long

hurt

desire

value

dislike

resent

appreciate

a high ‘psychocultural salience’ (Schrauf & Sanchez 2004: 274) (mean-

ing that they are free-listed by subjects) – for instance happy, sad, angry,
love, hate – but more research is necessary to explore systematically the
connection between frequency of usage and psychocultural or psycho-
logical salience. Neither is it clear how the findings can be compared to
frequency data based on elicited emotion words (for example Wallace &
Carson 1973), and space does not permit a discussion of this complex
issue here.
The terms represent common emotion terms in the (British) English
language in general: these are the words that speakers/writers use to talk
about their own and other’s emotional experience. They can be grouped
together as representing emotions such as love/admiration (love – N, V,
admire), happiness (happy, happily, cheerfully, enjoy), sadness (sad, sadly),
fear/worry (fear, worried, worry, anxious), hate/anger (angry, hate), desire
(hope, keen) and surprise (shock, surprise, surprised). Of these, happiness,

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 49

Table 2.16 Frequent emotion terms

All sub-corpora Three sub-corpora Frequency (ppm) in BNC

(sense-based) (sense-based) (based on Leech et al 2001:
271ff) (Not sense-based)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Nouns hope, love, fear, shock, surprise love (150), feeling (126), fear
feeling (93), hope (80), surprise (51),
shock (45)
Adjectives happy, worried, sad, keen, anxious, happy (129), surprised (47),
surprised angry angry (43), worried (38), keen
(38), sad (36), anxious (31)
Adverbs happily, cheerfully, sadly (19), happily (18),
desperately gratefully, sadly desperately (20),
Verbs love, worry, hate, admire love (150), enjoy (146), care
enjoy, care (81), worry (62), hate (50),
admire (22)

sadness, fear, anger and surprise are recognized by a large number of emo-
tion researchers as basic (possibly universal) emotions (Turner & Stets
2005: 13–15; compare also Section 5.3.2.2 below). Some of these terms
refer to (culturally) ‘positive’ rather than ‘negative’ emotions (compare
the suggestions made by appraisal theory and stance analysis) whereas
others are not as easy to classify or are more ‘neutral’:
positive: hope (N), love (N), happy, happily, love (V), enjoy,
cheerfully, gratefully, admire
neutral/ambiguous: feeling (N), surprised, care (V), surprise (N), keen
negative: fear (N), worried, desperately, worry (V), shock (N),
sad, anxious, angry, sadly, hate (V)
If only those terms that occur in all four sub-corpora are taken into
account, it seems that speakers talk more about positive than nega-
tive emotions; otherwise, the situation is more balanced. If we extend the
analysis to the 50 most frequent emotion words in general, we find the
distribution of ‘positive’, ‘negative’ and ‘neutral/ambiguous’ emotion
terms (no. of different emotion terms) as shown in Table 2.17.6

Table 2.17 Positive, negative, neutral emotion terms

Sub-corpus Positive Negative Neutral/ambiguous

Conversation 10 30 11
News reportage 15 23 12
Fiction 17 22 11
Academic discourse 16 21 13

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

50 Emotion Talk Across Corpora

All corpora prefer negative emotion terms over positive ones, even
though this tendency is most striking in conversation. Disregarding neu-
tral/ambiguous terms, about 75 per cent of the 50 most frequent emotion
terms in conversation are negative, about 61 per cent in news reportage,
about 56 per cent in fiction, and about 57 per cent in academic discourse.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

This preference is roughly in line with the general preference for nega-
tive emotion terms in language found by Nöth (1992: 82), with about
40 per cent positive vs. 60 per cent negative emotion terms (types). It
is also supported by Schrauf & Sanchez’s (2004) finding that the working
emotion vocabulary – defined as ‘the set of psychologically salient emotion
words to which an individual has immediate access when asked to make
a list of emotion words’ (Schrauf & Sanchez 2004: 270) – consists of more
negative (50 per cent) than positive (30 per cent) emotions. These com-
bined findings show that there are more negative emotion labels than
positive ones in the language as such (system); there are more negative
emotion labels than positive ones in the working emotion vocabulary
(psychological salience); and there are more negative than positive emo-
tion labels (types) in actual discourse (use). It has been suggested that
this ‘does not imply that humans have more negative than positive emo-
tional experience’ (Schrauf & Sanchez 2004: 280), but rather, that it has
to do with the importance of negative cues (for example in registering
danger) (Turner & Stets 2005: 12) or with different types of cognitive
processing (Schrauf & Sanchez 2004: 266).
These findings concern affect types, but what about tokens? In other
words, how often are positive emotion terms actually used in compar-
ison with negative and neutral ones? In Table 2.18 below we can see
the frequencies for the occurrences of the 50 most frequent emotion
terms.
As can be seen from the table, only in fiction are positive emotion
terms used more often than negative emotion terms; in the three other
sub-corpora, negative emotion terms are the most frequent. (However, it
must be pointed out that these figures and the figures above do not tell us
how often these terms are negated.) This supports Shimanoff (1985) and
Anderson & Leaper (1998), who found that negative emotion terms were
used more frequently by college students in conversations than positive
ones, but stands somewhat in contrast to Nöth’s findings that posi-
tive emotion words are more frequently used in the LOB corpus (Nöth
1992: 83). (However, it is unclear if Nöth’s analysis of emotion tokens
takes into account polysemy and non-emotion meanings, and the anal-
ysis does not investigate differences between the text types that make
up LOB.)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 51

Table 2.18 Positive, negative, neutral emotion terms

Sub-corpus Valence Raw frequency Frequency ppm

Conversation neutral/ambiguous 1,861 442.5
negative 3,488 829.3

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

positive 2,951 701.6
News reportage neutral/ambiguous 1,306 499.7
negative 2,818 1,078.3
positive 2,327 890.4
Fiction neutral/ambiguous 5,491 821
negative 9,529 1,424.7
positive 10,934 1,634.8
Academic neutral/ambiguous 2,214 371.4
discourse negative 2,987 501.1
positive 1,750 293.6

It is also interesting to note which registers are similar to each other

with regard to the frequency of emotion terms (excluding words that
occur in all registers). Considering only the ten most frequent words
in each POS category, news reportage (N) and fiction (F) are the most
similar registers (13 terms in common), followed by news reportage and
academic discourse (A) as well as conversation (C) and fiction (11 terms),
conversation and news reportage (8 terms), fiction and academic dis-
course (6 terms). Conversation and academic discourse are furthest apart,
sharing only one term (Table 2.19 on page 52).
These similarities shown in the table might be interpreted as follows:
both news reportage and fiction are concerned with story telling, both
report on emotional reactions (of news actors or characters) and both
have a nominal–adjectival affect style. Both also include features of
spoken conversation when representing (news actors’ or characters’) dia-
logue, explaining also the similarities between conversation–fiction and
conversation–news. In a sense, both fiction and news reportage are ‘com-
posite’ texts, inheriting features of (mediated) spoken discourse. As noted
in Section 1.6.2, academic discourse and news reportage have in common
some situational features (low degrees of interactiveness and immedi-
acy, providing information), accounting perhaps for the shared emotion
terms in these registers. Conversation and academic discourse are fur-
thest apart both situationally (1.6.2) and with reference to their POS
style (verbal–adjectival vs. nominal).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
52
Table 2.19 Sharing of emotion terms

C-N (8) C-F (11) C-A (1) N-F (13) N-A (11) F-A (6)
worry (N), pleasure (N), keen (A) proud (A), concern (N), anxious (A),
passionately (Adv), glad (A), furiously (Adv), expectation (N), angry (A),
shock (N), pleased (A), miserably (Adv), prepared (A), gloomily (Adv),
surprise (N), frightened (A), shock (N), concerned (A), admire (V),
sad (A), mind (V), surprise (N), disappointed (A), gratefully (Adv),
keen (A), miss (V), sad (A), keen (A), sadly (Adv)
cheerfully (Adv), shock (N), anxious (A), anxious (A),
hate (V) surprise (N), angry (A), angry (A),
sad (A), cheerfully (Adv), gratefully (Adv),
cheerfully (Adv), gratefully (Adv), sadly (Adv),
hate (V) sadly (Adv), admire (V)
hate (V),
admire (V)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 53

2.5 Syntactic variation

In this section I am interested in syntactic variation concerning different

parts of speech, in particular:

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

• Nouns: singular vs. plural
• Adjectives: positive vs. comparative (more) vs. superlative (most) (but
not: less, least)
• Verbs: base form (love) vs. -ed form (loved) vs. -s (loves) vs. -ing form
(loving)

Since it was not possible to analyze syntactic variation with respect to

more than 1000 emotion terms, this section concentrates on a subset of
these. Methodologically, the simplest solution was to examine in one
register all emotion terms that: (1) occur at least twice, (2) do not just
occur in (idiomatic) phrases (for example in a huff ) or have alternative
spellings (for example demoralisation – demoralization), and (3) require no
further (sense-sensitive) manual analysis.7 Altogether, 146 lexical items
were included in the analysis: a subset of 46 nouns from each register
comprising 12.5 per cent of all emotion nouns; a subset of 65 adjectives
from each register comprising 16.2 per cent of all emotion adjectives,
and a subset of 35 verbs from each register comprising 14.2 per cent of
all emotion verbs. The verb hate was excluded from the analysis because
of its high frequency, in order not to skew the results. The necessity of
restricting the analysis of syntactic variation to these subsets becomes
apparent when noting that even with only 146 emotion terms, 584 indi-
vidual searches had to be carried out for the analysis of syntactic variation
in the four registers (46 × 4 + 65 × 4 + 35 × 4), corresponding to about
20,000 occurrences.

2.5.1 Results
Let’s start by looking at the results for all types of syntactic variation
investigated here, as listed in Table 2.20.8 Starting with emotion nouns,
we can see that 92–96 per cent of all nouns in the four registers are
used in the singular form and only 3.5–6.9 per cent in the plural form.
The differences between the registers are negligible, and the focus is on
tendencies in any case. As for adjectives, 96–99 per cent are used in the
positive form, 1–3 per cent in the comparative form and only 0.2–0.5
per cent in the superlative form.9 Again, the distribution is very similar
in all four registers. Moving on to verbs, the occurrence of the -ing form

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
54
Table 2.20 Syntactic variation

Conversation News reportage Fiction Academic discourse

Raw freq % Raw freq % Raw freq % Raw freq %
Nouns
Singular 291 92.1 683 95.1 5,341 96.3 1,256 92.8
Plural 14 4.4 34 4.7 195 3.5 93 6.9
Total 316 – 718 – 5,546 – 1,354 –
2
Adjectives some expected values < 5 (i.e. X not applicable)
Positive 1,850 98.6 952 98.2 5,866 98.5 878 96.2
more 19 1.0 12 1.2 73 1.2 27 3.0
most 8 0.4 4 0.4 14 0.2 5 0.5
Total 1,877 – 969 – 5,953 – 913 –
Verbs
Base 185 41.9 178 32.1 841 28.9 207 36.7
-ed 119 26.9 246 44.3 1,549 53.3 237 42.0
-s 87 19.7 55 9.9 117 4.0 33 5.9
-ing 48 10.9 60 10.8 308 10.6 56 9.9
Total 442 – 555 – 2,907 – 564 –
Grand total 2,635 – 2,242 – 14,406 – 2,831 –

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 55

is roughly the same in all four registers (circa 10 per cent), but with the
other forms (base, -ed, -s) there are more differences: conversation is
distinguished from the other three registers in that the base form is most
frequent (∼40 per cent), in contrast to the -ed form in news reportage,
fiction and academic discourse (∼40–50 per cent). In conversation, the

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

s- form also has the highest percentage of all four corpora (∼20 per cent).
News reportage has a similar distribution of the base form to fiction, and
of the -ed form to academic discourse. It has the highest percentage of
the -s form of the three written corpora. Fiction is characterized by the
highest frequency of the -ed form, and the lowest frequency of the -s
form of all four corpora. Finally, academic reportage is distinguished by
having the smallest difference in distribution between the base form and
the -ed form of all four registers.

2.5.2 Interpretation
On the one hand, the fact that plural occurrences for emotion nouns are
less common than singular occurrences corresponds to an overall ten-
dency of the language system in which singular is the norm: ‘The higher
frequency of singular nouns agrees with their status as the unmarked
form (which generally has the widest distribution)’ (Biber et al. 1999:
291). This also becomes apparent from the frequencies of singular and
plural common nouns in the BNC (Leech et al. 2001: 295):

Grammatical word classes Frequency per million tags

common noun, neutral for 2896
number, for example sheep
singular common noun 121078
plural common noun 41320

On the other hand, the fact that no differences in the percentage

of singular and plural occurrences are found among the four registers
(which differs from findings for nouns in general by Leech et al. 2001 for
the similar but not equivalent corpora ‘Imaginative’, ‘Informative’ and
‘Conversation’) relates to the specific character of emotion nouns. Nouns
denoting emotion are usually non-count/uncountable nouns, and the
low frequency of the plural form is hence not surprising. Those nouns
that do allow the plural often have different affect meanings for the
non-count and the count usage. Nouns that occur in the plural include
(f > five in plural in one of the corpora): whim, worry, disappointment,
affection, antagonism, leaning, surprise, regret, resentment, apprehension,
sorrow, anxiety, frustration, humiliation, jealousy, grudge. Many of these

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

56 Emotion Talk Across Corpora

have different affect meanings typically associated with the plural form:
for example, worry is usually ‘the state of worrying about something’
(OALD) whereas worries often refers to ‘something that worries you’
(OALD). Similarly, disappointment typically means ‘sadness because sth
has not happened or been as good, successful, etc. as you expected or

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

hoped’ (OALD), whereas disappointments are ‘[persons or things] that [are]
disappointing’ (OALD).
With verbs – where the occurrence of the -ing form is about the same
in all registers – there is a similar grammatical restriction: many verbs
denoting emotion typically do not allow the progressive (for a list of
exceptions, see Biber et al. 1999: 472). In ‘traditional’ grammar, they
belong to the so-called stative verbs (Biber et al. 1999: 471); in SFL gram-
mar, they belong to the category of mental processes, where the simple
present is the unmarked tense (Halliday & Matthiessen 2004: 206).10
As with plural occurrences for nouns, the progressive form for emotion
verbs ‘is less frequent and . . . carries a special interpretation’ (Halliday &
Matthiessen 2004: 207) or relates to different meanings of emotion
terms. For example, admire (1 ‘to respect sb for what they are or what
they have done’; 2 ‘to look at sth and think that it is attractive and/or
impressive’, OALD) in the progressive will usually be used in meaning
2 rather than 1. Here are some examples from the fiction sub-corpus of
the BRC:

(3) And I noticed Otley was admiring himself in the mirror a lot these
days (BRC, ACK 490)
(4) He was admiring the dexterity of a young street orderly (BRC,
ANL 738)
(5) Anneliese was admiring the cut and style with a practised European
eye for fashion (BRC, GUE 2066)

Both emotion nouns and emotion verbs are thus restrained grammat-
ically in terms of plural occurrences and the progressive form, even
though there are exceptions, with some emotion nouns more frequent
in the plural, for example leaning.
With adjectives, the situation is slightly different in that there is no
restriction on their gradability: emotion adjectives are in fact poten-
tially gradable (happy – happier – happiest) but are not often used as
graded forms (happier, happiest). Again, this reflects a general tendency
for adjectives to be used in the positive, rather than in the compara-
tive or superlative form, with the comparative being more frequent than

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 57

the superlative. Here are some figures (Leech et al. 2001: 295) from the
whole BNC:

Grammatical word classes Frequency per million tags

general adj 56336

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

general comparative adj 1674
general superlative adj 751

Perhaps the fact that there is little inter-register variation in terms of

comparative and superlative forms is partly explained by the specific
nature of emotion adjectives: since such adjectives relate to ‘private’
states into which only the emoter has insight, it would be very marked
to say I’m happier/angrier/more surprised than he is (whereas it is possible to
say I’m happier now than I was then). Therefore, there is less opportunity
for using graded forms with emotion adjectives.
The close correspondence in syntactic variation in the four registers
as far as emotion nouns and adjectives (and the -ing form of emotion
verbs) are concerned is thus only at first sight surprising, taking into
account the fact that it was not the same subset of lexical items that
were analyzed in the different corpora. There is simply something in
the nature of emotion terms as such that explains these findings: the
similarities between the registers become explicable by the nature of the
lexical items investigated, and partly by general language tendencies.

Table 2.21 Variation in verb form

Conversation News reportage Fiction Academic discourse

Base 41.9% 32.1% 28.9% 36.7%
-ed 26.9% 44.3% 53.3% 42%
-s 19.7% 9.9% 4% 5.9%

With verbs there is more inter-register variation, as we have seen. For

reasons of convenience Table 2.21 above lists the percentages again. If
we assume that the base form and the -s form are associated with the
present tense or with tense-less (modal and so on) usages,11 and that the
-ed form is associated with the past tense or the passive voice, we can
make the following hypotheses:

• In conversation speakers use emotion verbs less frequently to refer to

their emotions in the past (that is, in narratives), and more frequently
‘to refer to current [for example emotional] states’ (Biber et al. 1999:
458). Conversation in general is characterized by the present tense, by

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

58 Emotion Talk Across Corpora

‘speakers’ general focus on the immediate context’ (Biber et al. 1999:

457). Compare also Shimanoff’s (1985) finding that college students
have a tendency to refer to their present emotions more than to their
past ones.
• In news reportage, which uses the past tense to report narrated events

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

as well as to represent indirect speech, and the present tense when
quoting news actors, it seems that emotion verbs are more frequently
used in indirect speech or descriptions of news actors rather than in
direct speech (though news actors’ speech may also include emotion
terms in past tense). Some -ed forms might also be associated with the
passive voice, since news is characterized both by the present and past
tense as well as the occurrence of passives (Biber et al. 1999: 456, 476).
• In fiction, which is associated with the past tense in narration, and
where the present tense often only occurs in the direct representation
of characters’ speech (Biber et al. 1999: 458), many emotion verbs are
used to describe characters rather than in characters’ speech, though
the latter also occur. Fiction and news reportage appear similar in this
respect.
• In academic discourse, it is hypothesized that the base form is used
mainly in tense-less forms (with modal verbs and so on modifying the
emotion verbs) rather than in quoted speech. Even though academic
prose, like conversation, is characterized by the present tense, this is
here used ‘to imply a lack of time restriction, with the present subsum-
ing past and future time’ (Biber et al. 1999: 458). Of course, academic
discourse is also associated with the passive voice (Biber et al. 1999:
476). Thus it may be that both tense-less use of emotion verbs and
the use of emotion verbs for descriptions of past emotions or in the
passive voice are perhaps equally common.

These assumptions must necessarily remain hypotheses because with-

out further investigation it is simply not clear how many base forms are
associated with the present tense rather than non-tense, and how many
-ed forms are associated with the passive voice rather than past tense.
Such investigation was beyond the scope of the analyses undertaken in
this book. The hypotheses above are made on the basis of what is known
about the registers already, as well as from a general impression formed
by the analysis of the concordance lines, but will be confirmed to some
extent by the analyses of patterns for 15 emotion terms presented in
Chapter 4.
In order to complement these analyses, and to look into the differ-
ence between authorial and non-authorial affect (Section 1.4), a rough

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 59

Table 2.22 Distribution of emoters

Emoter Conversation News reportage Fiction Academic discourse

Raw freq % Raw freq % Raw freq % Raw freq %
‘I’ 211 47.7 67 12.1 419 14.4 35 6.2

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

‘you’ 59 13.3 7 1.3 156 5.4 15 2.7
‘we’ 5 1.1 14 2.5 21 0.7 32 5.7
‘3rd’ 154 34.8 408 73.5 2,248 77.3 392 69.5
Total 442 – 555 – 2,907 – 564 –

count for different emoters (for emotion verbs) was also undertaken.12 As
can be seen in Table 2.22, ‘I’ and ‘you’ references are most frequent in
conversation, whereas in news reportage, fiction and academic discourse
third person references occur most often. This confirms the assumptions
made above: that conversation is the only register in which speakers
frequently refer to their own and, to some extent, to interlocutors’
emotions, whereas in news reportage and fiction verbs are much less
frequently used in dialogue (that is, involving the usage of I and you).
The results of this analysis can also be compared with the findings by
Scheibman (2002: 65–105) for subject types of feeling verbs in American
English conversation, based on a smaller corpus and hence lower fre-
quencies overall. She found that first person singular subject types are
most frequent (42.2 per cent), followed by third person subject types
(33.3 per cent if singular and plural are added) and second person singular
subject types (20 per cent). And Shimanoff (1985) as well as Anderson &
Leaper (1998) note that college students refer more frequently to their
own emotions in conversations.
The findings are also not surprising given what we know about these
registers as such, namely that the pronouns I, me and you are far more
common in conversation than in the other registers ‘because both par-
ticipants are in immediate contact, and the interaction typically focuses
on matters of immediate concern’ (Biber et al. 1999: 333). It must be
noted however, that emotion verbs in conversation are in fact used more
frequently to refer to a third party’s emotion than to refer to the hearer’s
emotion. This is because emotions are conceptualized as private states
(see above), and belong to the so-called A-events (as do mental states
in general) – events that are ‘[k]nown to A [the speaker], but not to B’
(Labov & Fanshel 1977: 100). A-events have a special status, since they
cannot easily be challenged or attributed to someone in ordinary con-
versation. For instance, it would be somewhat marked or noticeable to

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

60 Emotion Talk Across Corpora

say You are not sad. It seems very challenging indeed to attribute a private
state to someone other than yourself, in particular when that someone
is your interlocutor. It is much less challenging to attribute a private
state to a third party (who is not present and who cannot object) than
to claim that the hearer feels/felt a certain emotion at a certain time. But

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

such attributions can of course occur in questions (examples from BRC
conversation sub-corpus):

(6) Were you disappointed when you put it back on? (BRC, KCP 2728)
(7) it’s, I mean like Tracey said are you disappointed? (BRC, KBF 12205)
(8) Was you disappointed? (BRC, KCT 828)
(9) Were you frightened trying to get in? (BRC, KBL 1266)
(10) Are you frightened you will er swear? (BRC, KD2 12)
(11) Who d’ya hate really really hate? (BRC, KP3 1751)
(12) You hate her? (BRC, KNY 1778)
(13) You hate it don’t you? (BRC, KCT 4506)
(14) What animal is it that you hate? (BRC, KBF 731)
(15) Oh dear you were pleased to get off weren’t you? (BRC, KB0 388)
(16) Bet you were surprised to hear Monsignor on the phone weren’t
you? (BRC, KE2 8659)
(17) Are you surprised? (BRC, KB1 4509)
(18) Yeah why, why are you surprised? (BRC, KBX 522)

This does not mean that such attributions do not occasionally occur
in statements since the frequencies for this were not systematically
examined:

(19) You hate carrot cake or whatever it is. (BRC, KC4 370)
(20) I told him you hate Chris. (BRC, KNY 1441
(21) Oh, you hate her hair (BRC, KNY 1780)
(22) You hate that. (BRC, KDH 12)
(23) Well I’ll tell you what er, Liz Taylor was on, you know the Oprah
Winfrey show that you hate? (BRC, KD8 513)
(24) Last time he flew out, you were frightened. (BRC, KB1 2042) (from
BRC conversation sub-corpus)

The basis of such attributions may be the assumption of feeling rules (Sec-
tion 1.3), the projection of one’s own experience, or one’s knowledge of
the other (Fiehler 1990: 141). Also, we may of course choose to share our
emotions with others, and they can pick up on this. Communication in
this sense can be seen as ‘a way of bridging the gap between the personal

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 61

and the social sides of emotion’ (Planalp 1999: 134; on the social sharing
of emotions see also Rimé et al. 1998). However, this is up to the emoter:
Wowk (1989) notes, with respect to the counselling of breast cancer
patients, the existence of ‘certain acknowledged rights of the recipient
to define and disclose his or her personal state’ (Wowk 1989: 60).

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

The status of emotions as A-events also helps to explain the fact that
‘we’ references are very rare in all four registers. Since emotions are con-
ceptualized as private states, it becomes difficult to talk about more than
one person’s experienced emotion at the same time. To talk about our
love/hate of linguistics implies both an insight into someone other than
yourself, and furthermore claims that that person experiences the same
kind of emotional reaction – it de-individualizes the subject to a certain
extent. It claims solidarity, coupledom, unity. In academic discourse
such references are more frequent because of its specific usage of we
as authorial voice, either referring to a single researcher, to a group of
collaborating researchers or to the writer and the reader/readers.

2.5.3 Summary of syntactic variation

If we assume that language is at least partly responsible for con-
struing reality, the usage of emotion terms in the English language
with respect to syntactic variation seems to contribute to the Western
conceptualization of emotions as prototypically:

• non-comparable (see adjectives)

• non-countable (see nouns)
• stative (see verbs)
• private rather than public/shared (see emoters)

As Biber et al. have noted, the contrast between noun types (for example
count/non-count) ‘is not a simple reflection of reality, but rather reflects
how we choose to conceptualize the entities which we want to talk about’
(Biber et al. 1999: 242), and similar assumptions have been made by
explicitly constructivist approaches to language: ‘experience is the reality
that we construe for ourselves by means of language’ (Halliday & Matthiessen
1999: 3, original emphasis).
Research has in fact noted that cultures can be described as indi-
vidualist (Western) or collectivist (East Asian and South American): see
Planalp (1999: 223). Individualist cultures emphasize personal indi-
viduality, whereas collectivist cultures value family and community
structures. Members of individualist cultures place high value on their
independence, including the subjectivity of their emotional experience

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

62 Emotion Talk Across Corpora

as supporting their unique identity, whereas in collectivist cultures emo-

tional experience is less subjective and more shared (Gordon 1990: 153,
Mees 2006: 10). The similarities in the syntactic variation of emotion
terms in all registers seem to derive from this specific conceptualization
of emotion terms (as well as general language tendencies), whereas the

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

differences can be explained in terms of the general characteristics of the
respective register (Section 1.6.2).

Notes
1. The results that are given in Chapter 2 (and Appendix E1 online) show key
tendencies in variation, without necessarily representing exact and stable fig-
ures that will always be found in the different registers (even if exact figures are
reported rather than rounded ones). This is because of (1) the limitations of
the corpus, (2) classification errors, (3) the ‘failings’ of the human analyst (see
Appendix A 1.2 online). Unfortunately, ‘when working with large corpus files,
there will always be some variability due to error, depending on occasional
human mistakes or inaccuracy of methodological tools’ (Biber et al. 1999: 39).
Furthermore, there may be difficulties in comparing the distribution of lexis
across corpora of different size. A discussion of this issue is provided by Biber
(2006: 251).
2. All results are statistically significant using the χ2 test unless otherwise noted.
I used the chi-square statistic (χ2 ) because this seems to be the most frequently
used in corpus linguistics, but am myself skeptical as to the relevance of such
tests for linguistic analysis. Although Leech et al. (2001: 16) argue that the G2
log-likelihood-statistic performs better than χ2 , Moore (2004) points out that
‘Agresti (1999, p. 246) cites studies showing ‘X2 is valid with smaller sample
sizes, and more sparse tables than G2,’, and either X2 or G2 can be unreli-
able when expected frequencies of less than five are involved, depending on
circumstances’ (Moore 2004: 1). For further discussion, see Dunning (1993),
Oakes (1998), Leech et al. (2001), Moore (2004), and Kilgarriff (2001). χ2 was
calculated with the help of a perl program kindly written by Michael Oakes;
any mistakes in its application are mine.
3. Planalp (1999) cites Meštrović (1997) as arguing that emotions are shaped
to a large extent by the media, which tell us how to feel about war, crime
or people. The strong view is that ’emotions have been transformed from
personal, moral passions into “synthetic, quasi-emotions” manipulated by
the culture industry’ (Planalp 1999: 239, referring to Meštrović 1997: xi).
4. In Table 2.6 and all following
indicates that the emotion term is among
the ten most frequent emotion terms in the respective corpus. Where
is
not present, this does not mean that the emotion term does not occur at all
in the corpus; merely that it is not amongst the ten most frequent emotion
terms in the given category. This also relates to the discussion of occurrences.
5. It is worth mentioning that previous research into emotion words supports my
findings to some extent. For instance, Stubbs found that happy and happiness
are both frequent and evenly distributed in the LOB corpus (Stubbs 1996: 89).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Emotion Profiling 63

My analysis showed that this is true also for three of the four BRC sub-corpora
even when meaning distinctions are taken into account (which Stubbs does
not): happy is very frequent in conversation (119.8 ppm), news reportage
(147.7 ppm), and fiction (172.2 ppm), but much less frequent in academic
discourse (22 ppm). Precht’s (2000: 130–48) analysis of stance found no regis-
ter variation concerning factor five (‘Opiniated affect/person-oriented’: afraid,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

fear, drawn, interested, love, lucky, should have, wrong in a variety of grammati-
cal frames), but concluded that factor eight (‘Adjectival affect/social ties’: nice,
glad in different grammatical frames) occurs primarily in spoken language
and more in female speakers, and that factor ten (‘Intimate affective expres-
sion/private’: funny, god, hate, have to, know, like, predAFF1, want in a variety
of frames) is typical for American conversation (more females and more
younger speakers than in other registers). Even though the findings cannot be
compared directly in terms of methodology, corpus design and investigated
features (see Section 1.5), it is interesting to note that some of my findings (the
high frequency of glad, like and hate in conversation) support her analyses.
6. Appendix A 2.4 online shows the classification of emotion terms as ‘positive’,
‘negative’ or ‘neutral’ (also Section 5.3.4).
7. Consequently, it was not the same subset of nouns/adjectives/verbs that were
investigated in each sub-corpus, because different emotion terms required
a manual analysis in the four registers. The only lexical items that were
analyzed in all four corpora are: admiration, affection, desperation, gratitude,
impatience, jealousy, regret, sadness (nouns), annoyed, anxious, appalled, appre-
hensive, delighted, disgruntled, dismayed, ecstatic, enthusiastic, envious, puzzled,
surprised, terrified (adjectives), admire, delight, despise, disapprove, dislike,
frighten, humiliate, intimidate, panic, resent, surprise, terrify (verbs). However,
to make the analysis more comparable, the same percentage (subset) was
used in all registers. For instance, only 46 nouns in the conversation corpus
fulfil the three criteria above and could therefore be included in the anal-
ysis of syntactic variation in conversation. This in turn means that in the
three other corpora only 46 nouns were analyzed, even if more nouns ful-
filled the three criteria. In such cases, the least frequent nouns were excluded
from the analysis to make the findings more representative, some randomly
whenever necessary. A list of all items analyzed for syntactic variation can
be found in Appendix A 2.5 online.
8. The reason why the percentages for singular and plural nouns (and so on)
do not add up to 100 per cent is that some occurrences are classified as the
wrong part of speech by the automatic POS tagger. These were excluded in
the manual analysis.
9. Some occurrences of more preceding on adjective, which did not indicate a
‘true’ comparative, were not counted (for example all the more grateful).
10. But note that the -ing form can also be used in non-finite structures rather
than in the progressive.
11. Incidentally, although the -s form is associated with third person subjects,
this does not automatically indicate non-authorial affect, since usages with
a ‘dummy’ it can also express authorial affect: it surprises me, it annoys me, it
contrast to the non-authorial X admires me, X hates me.
12. It must be noted that in this rough calculation (which involves both pro-
noun usages and lexical NPs), no meaning-based analysis of the data was

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

64 Emotion Talk Across Corpora

possible because of time constraints. That is, occurrences classified as ‘you’

also include those that mean ‘one’ (in general) rather than the interlocu-
tor(s) and so on (see Biber et al. 1999: 328–32 on different usages of personal
pronouns). It must also be mentioned that the ‘I’ is not necessarily the
speaker/author (when the I occurs in quoted/attributed speech), and that in
fiction issues of point of view (Simpson, 1993, Toolan 2001, Rimmon-Kenan

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

2002) can be quite complex. There is hence no direct correlation between ‘I’
and authorial affect, especially in the written registers.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

3
A Local Grammar of Affect

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

This chapter is about the analysis of lexico-grammatical patterns that are
associated with emotion terms, that is, about words and word classes that
precede or follow them. As in Section 2.5 (syntactic variation), a subset
of emotion terms is used for this purpose because it was not feasible
to analyze the patterns of more than 1000 emotion terms. Altogether,
15 emotion terms were analyzed in detail in the four sub-corpora: nine
emotion adjectives (four ‘positive’, three ‘negative’ and two ‘neutral’),
three emotion nouns (one ‘positive’, one ‘negative’ and one ‘neutral’)
and three emotion verbs (one ‘positive’, one ‘negative’ and one ‘neutral’).
Table 3.1 lists the terms that were analyzed.
These terms were chosen from the list of 146 emotion terms that
were used for the analysis of syntactic variation in Section 2.5 (see
Chapter 2, Note 7), on the basis that they occurred reasonably fre-
quently in all four sub-corpora. This chapter gives an overview of patterns
that can realize affect, mainly involving the attempt to sketch a ‘local
grammar’ of affect, which is based on a combination of Hunston’s
(for example 2003) local grammar approach and Fillmore’s FrameNet
approach (http://framenet.icsi.berkeley.edu). Though space does not
permit an extensive discussion of both approaches in this chapter, brief

Table 3.1 Analyzed emotion terms

Emotion terms ‘Positive’ ‘Negative’ ‘Neutral’

Adjectives delighted, enthusiastic, anxious, disappointed, surprised,
impressed, pleased frightened willing
Nouns affection hate surprise
Verbs admire hate surprise

65

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

66 Emotion Talk Across Corpora

overviews will be given in the following Sections (more detailed infor-

mation can be found in the cited publications by Hunston and her
colleagues as well as on the FrameNet website).

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

3.1 The local grammar approach

As many linguists have pointed out, the relationship between linguis-

tic form and meaning is far from arbitrary (see Hunston 2003: 342).
For instance, the different meanings of a polysemous word are distin-
guished by the various grammatical patterns in which the word occurs,
whereas words that have similar meanings can share grammatical pat-
terns (Figure 3.1). This behaviour can either be explained as ‘meaning
causes behaviour’ or as ‘behaviour causes meaning’ (Hunston 2003: 342),
and is examined in more detail in Hunston & Francis (2000) and Fran-
cis et al. (1996, 1998). It is also the basis on which the concept of local
grammar was developed (Hunston 2003: 347–8) – a concept that I have
found useful for describing the behaviour of emotion terms in the BRC.
The concept of local grammar goes back to the finding that software
for the grammatical analysis of corpora (parsers), which are based on
traditional grammatical models, find it hard to deal with specific areas of
language, and that ‘there is still a lot left over’ (Hunston & Sinclair 2000:
76). Hunston & Sinclair (2000), following Gross (1993) and Barnbrook &
Sinclair (1995), have suggested the use of parsers that are based on local
grammars for such cases. Local grammars are essentially descriptions of
particular areas of language (rather than the language as a whole), such as
dictionary definitions, newspaper headlines, the language of cause and
effect or the language of evaluation. In other words, a local grammar
describes ‘one meaning only’ (Hunston 2002: 178), even if meaning is
defined rather broadly. Local grammars typically work with transparent
category labels referring to functional categories that are characteristic for
the area of language that is to be described (Butler 2004: 158), for example
Definiens, Definiendum for definitions, Evaluative category, Thing evaluated
for evaluative language. Local grammars are then used to describe and

pattern 1 word 1

word pattern 2 word 2 pattern

pattern 3 word 3

Figure 3.1 Words and patterns

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 67

analyze sentences. The resulting description ‘is “functional” in a different

way from the tradition of functional grammars, and incorporates some
valuable pragmatic parameters’ (Hunston & Sinclair 2000: 79).
The approach has similarities with the FrameNet project (http://
framenet.icsi.berkeley.edu/) developed by Fillmore and others (these

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

similarities are investigated by Hunston 2003), and there are also con-
nections to other theories such as construction grammar (for example
Fillmore 1997), which are discussed by Stubbs (2001). Both FrameNet
(FrN) and local grammar are corpus-based, and involve the seman-
tic mapping of categories onto sentences. Taking two examples from
FrameNet, let’s very briefly compare the two approaches:

(1) She subsequently attended the genetic counselling clinic, and was
very anxious about the situation. (FrN)
(2) George felt anxious that he had no opportunity to be alone with
Tamar. (FrN)

A local grammar analysis would presumably parse these sentences as

indicated in Table 3.2 (with categories merged from Hunston & Sinclair
2000, and Hunston 2003). A local grammar approach would collect other
examples of words that use these patterns and sort them into meaning
groups (Table 3.3). It would also identify other patterns (for example ADJ
at n) that are used to attribute emotions to experiencers (Hunston &
Sinclair 2000, Hunston 2003), that is, patterns that share this function.
Local grammars are thus motivated by ‘the function of the sentence, not
the words it has in it’ (Hunston 2003: 345).
FrameNet on the other hand, involves more than patterns: it iden-
tifies semantic–syntactic frames that are associated with a particular
sense of a word, notes their elements, and how they are realized lexico-
grammatically in the corpus. It also identifies senses of other lexical
items that share this frame. Ultimately, ‘[t]he aim is to document the

Table 3.2 Local grammar analysis

Experiencer Hinge Emotion Cause/Target

noun group link verb adjective group prepositional phrase with about
She . . . was very anxious about the situation (FrN)

noun group link verb adjective group that clause

George felt anxious that he had no opportunity
to be alone with Tamar (FrN)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

68 Emotion Talk Across Corpora

Table 3.3 Patterns and meaning groups

Pattern Meaning groups

ADJ about n ‘passionate & cool’
‘happy’

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

‘unhappy’
‘philosophical’
‘nervous’
‘angry’
‘curious’
‘cynical & serious’
ADJ that ‘surprised’
‘angry’
‘horrified’
‘glad’
‘anxious’

(from Hunston 2003: 354–5)

range of semantic and syntactic combinatory possibilities – valences – of

each word in each of its senses, through computer-assisted annotation of
example sentences and automatic tabulation and display of the annota-
tion results’ (http://framenet.icsi.berkeley.edu/book/book.html#intro).
For example, anxious is described by FrameNet as belonging to the
Emotion_directed frame, which is defined as follows on the FrameNet
website:

The adjectives and nouns in this frame describe an Experiencer who is

feeling or experiencing a particular emotional response to a Stimulus
or about a Topic. There can also be a Circumstances (sic) under which
the response occurs or a Reason that the Stimulus evokes the particular
response in the Experiencer.

The entry for anxious also lists the structure of core and non-core frame
elements in sentences in the corpus (including a quantification). The
parsing of a sentence with anxious thus involves a lot of detail about
the structures in which it occurs, with extra elements such as Degree,
Circumstance, State, Reason and the listing of typical copula (Cop) or
other support (Supp) verbs (examples from FrameNet):

(3) She [Experiencer] subsequently attended the genetic counselling

clinic, and was [Cop] very [Degree] anxious [Target] about the
situation [Topic].

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 69

(4) George [Experiencer] felt [Supp] anxious [Target] that he had no

opportunity to be alone with Tamar [Stimulus].

FrameNet hence provides a description of particular frames in which

different lexical items participate, whereas a local grammar tries to pro-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

vide a description of one area of meaning (based on function) realized
by different lexical items. Local grammars involve patterns in the CCED
(Collins Cobuild English Dictionary) sense (Hunston 2003: 356), whereas
FrameNet includes other types of contextual information.

3.2 Combining local grammar and FrameNet:

affect patterns

For the analysis of emotion terms undertaken here I combined some

aspects of the local grammar and the FrameNet approach. Like local
grammar, my starting point is one area of meaning: the system of affect,
in particular how words denoting affect/emotion are used to attribute
emotions (to self or other) in discourse.
As a first step, the semantic elements that are involved in affect were
identified and labels for them were chosen. The crucial question is
‘Which elements of emotional experience are important and should be
identified in an analysis of affect?’ For example, Hunston (2003) includes
the labels Experiencer, Emoter, Cause, Target, and Action/Phenomenon,
without, however, explicitly defining them (but they are relatively self-
explanatory). FrameNet describes 12 different frames that use an overall
Emotions Frame (see Table A.10 in Appendix A 3.1 online),1 most of which
include an Experiencer (who is in some frames called a Cognizer or a Judge),
and some element that provokes the emotion or tells us what the emo-
tion is about or at what/whom the emotion is directed. The name of
this element differs, and is called Stimulus, Topic, Event, Content, Eval-
uee, Action, Reason or Means (Appendix A 3.2 online). Even though there
may well be reasons for distinguishing between these elements within
the framework of FrameNet, for our purposes it is better to focus on the
similarities between them, namely that they all describe an entity, situ-
ation, state-of-affairs and so on that causes an emotion or at which an
emotion is directed. It then seems reasonable to suggest treating these
different terms (Reason, Topic, Stimulus, Content . . .) as belonging to the
same overall category (trigger). The coding of diverse elements as trig-
ger emphasizes the similarities between emotion terms, and examples
that are classified differently in FrameNet with respect to their frames

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

70 Emotion Talk Across Corpora

Table 3.4 Coding of affect

Emoter Emotion Trigger

I admire you for your intellect (FrN)
Everyone loves compliments (FrN)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

He worries … about the house (FrN)
God forgives repentant sinners (BRC, A07 388)
He desires that you take over the keepership of the
royal castle of Berwick (BNC, CD8 1292)
[O]ne Soviet repented their actions against the Soviet Union
and one Afghan (BRC, A95 103)
citizen . . .
Mr. Whiskers is upset that there are no more cat treats (FrN)
Modigliani was anxious because of the difficulty of finding models
(BNC, ANF 1504)

(see Table A.11 in Appendix A 3.3 online) and frame elements would
consequently all be described equally as in Table 3.4.
Thus, I distinguish between three main elements for describing emo-
tional experience: the one experiencing the emotion, the emotion itself
and the cause or target of the emotion. Galasiński similarly notes that
only three aspects of emotional experience are important: the emoter,
the emotion and the object of the emotion (Galasiński 2004: 48). In
accordance with appraisal theory, I will use the following terms to label
these three categories:

• Emoter: the one to whom an emotional response is assigned: who is

said to ‘feel’ an emotion;
• Emotion: the particular emotion involved;
• Trigger: what causes the emotional response; what is ‘evaluated’ emo-
tionally; what the emotion is about; the cause, reason or target of an
emotion.

As a second step (after the identification of the three main affect cate-
gories), corpus examples from Francis et al. (1996, 1998 – from the Bank
of English), Hunston (2003 – from the Bank of English), and FrameNet
(from the BNC) as well as the detailed analysis of the 15 emotion terms
in the BRC were used to establish how emoters, emotions and triggers
are realized lexico-grammatically in patterns. Some exploratory corpus
research of other emotion terms in the BRC was also undertaken to prove
or disprove the existence of certain patterns for affect.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 71

This approach led to the inclusion of structures that are non-obligatory,

and can also occur with other patterns, structures that are what Hunston
has labelled ‘ “free-floating” [elements] that [need] to be labelled as a
local grammar or frame element wherever [they occur]’ (2003: 357). For
example, in many but not all cases when-clauses realize triggers:

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(5) I jump with surprise when a dog suddenly barks at me (BRC, AM6
1032)
(6) he was pleased when you agreed to this (FrN)
(7) He seemed disappointed when the dispute was settled at higher levels
(FrN)
(8) We were confused and anxious when we were not understood (FrN)

The patterns N when-cl and ADJ when-cl (emotion when trigger) were
hence included to take into account the fact that when-clauses following
emotion terms often describe the trigger of an emotion – information
which would be useful for a parser to have.
Similarly, I included elements even when they are dependent not
on the emotion term but on one of its co-occurring lexical items. An
example of a pattern that is included because of a co-occurring item
(here bestow) of an emotion term is N on n (emotion on trigger): bestow
affection on his slightly wayward children (BRC, CHG 1570). This takes
into consideration that his slightly wayward children are clearly the trigger
for the emotion.2
Additionally, it was found helpful in some cases to employ three
further category labels in the description of sentence elements. These
are:

• Expressor: ‘a body part, gesture or other expression of the Experiencer

that reflects his or her emotional state’ (FrameNet);
• Empathy target: ‘the individual or individuals with which the Experi-
encer identifies emotionally and thus shares their emotional response’
(FrameNet);
• Action: an action contextually associated with an emotion term.
There is no clear dividing line between expressors and actions.

Section 3.3 describes the generalized patterns that make up the local
grammar of affect. Even though this local grammar needs to be elabo-
rated by more corpus-based research regarding the patterning of addi-
tional emotion terms, the detailed analysis of the 15 chosen emotion
terms was complemented by corpus material from Francis et al. (1996),

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

72 Emotion Talk Across Corpora

Biber et al. (1999), and Hunston (2003) so that the description applies to
more than just the analyzed emotion terms.

3.3 Patterns and meaning

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

In a local grammar approach, pattern elements are described with the
help of labels that reflect their function. Different patterns can also be
grouped together if they realize similar functions. Thus, in the following
sections, patterns of affect are grouped first according to function, then
according to part of speech, illustrated with one or two corpus examples.3
The examples are from all the corpus-based sources mentioned above,
including but not limited to the BRC. Rather than listing each specific
pattern individually (for example ADJ about, ADJ with) I will generalize
as far as possible (for example ADJ prep).4

3.3.1 Unemoted affect

In a number of examples, emotion nouns seem to occur without an
emoter:

(9) When the two women set off to Berlin together in search of
Roswitha’s past, the film becomes an exploration of the power of
affection to transcend boundaries of age, background and sexual
orientation. (BRC, AJF 36)
(10) Books: The sorrows of love and the joys of political hate (BRC,
AK4 521)

Such ‘unemoted’ affect can be realized through a (usually abstract)

noun followed by a prepositional phrase (Table 3.5). Emotion is here
conceptualized as an abstract entity (Lutz 1990: 84) rather than as an
emotional response of specific emoters at specific triggers. However, the
same pattern may also be used to attribute affect to an emoter:

(11) Yeats watched, with an appalled fascination, as she [emoter] threw

herself into ‘a joyous and self-forgetting condition of political
hate’ [emotion]. (BRC, AK4 547)

Table 3.5

Pattern Elements Examples

(abstract) n prep emotion the power of affection (BRC, AJF 36)
(of ) N the joys of political hate (BRC, AK4 521)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 73

Whether or not we deal with unemoted or emoted affect thus crucially

depends on the context, and the pattern n of N can in fact realize emoted
affect. Since in appraisal theory affect is concerned with ‘how people
express their feelings in discourse’ (Martin & Rose 2003: 25, emphasis
mine) rather than with how emotions are referred to in discourse, it is

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

perhaps best not to consider such unemoted instances as affect in a strict
sense. In any case, such examples have a very weak evaluative power.
That is, the emotion terms are not used to evaluate emoters or triggers,
but are part of more general statements. For instance, in the examples
above, they are used to outline characteristics of human artefacts. The
film is described as ‘an exploration of the power of affection to tran-
scend boundaries’ and the book is said to deal with ‘the joys of political
hate’ – the emotion terms are used to describe entities rather than to eval-
uate emoters or triggers. Examples of such abstract references can also
be found in much academic literature (for example psychology), where
statements are made about emotions as such.

3.3.2 Undirected affect

More interesting for affect are patterns that do involve the attribution of
emotions to particular emoters, even if they do not involve triggers. We
might call this category undirected affect (see also Section 5.3.2.1). Patterns
in this group specify emotion and emoter but not trigger. They can realize
both authorial (attribution of emotion to self) and non-authorial affect
(attribution of emotion to other).

Verb patterns
Starting with verb patterns, only some emotion verbs can be used in undi-
rected affect patterns. Some examples are presented in Table 3.6. While
in utterances such as I don’t mind or Don’t panic the trigger is certainly
to be retrieved from the context, an utterance such as I don’t scare eas-
ily makes a statement about a general emotional disposition (no matter
what the trigger).5

Table 3.6

Pattern Elements Example

V emoter emotion I don’t mind. (BRC, KCS 2157)
Don’t panic. (BRC, B20 3174)
I don’t scare easily (V adv) (Francis et al.
1996: 137)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

74 Emotion Talk Across Corpora

Adjective patterns
With the adjective pattern ADJ n, the emotion adjective can directly
modify a head noun describing a person (the emoter), as in Table 3.7.
But in some cases the emoter is only metonymically related to the modified
head noun. This is the case with adjectives modifying an expressor (body

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

part, gesture and so on of the emoter – see above), with a part–whole
relation between emoter and expressor (Table 3.8).

Table 3.7

Pattern Elements Examples

ADJ n emotion emoter In one room, a frightened man bared his back
which was a mass of scabs and bruises.
(BRC, CH5 4286)

Table 3.8

Pattern Elements Examples

ADJ n emotion expressor She smiled at Meredith’s surprised expression.
(BRC, CEB 933)

A metonymic relation also exists in examples where the emotion adjec-

tive does not directly modify an emoter or an expressor, but rather an
event or process. For example:

(12) According to a disappointed editorial in The Economist (BRC, A6F

1390)

Here readers have to infer the emoter by establishing the contiguity rela-
tion existing between the head noun (editorial) and a potential emoter,
here the unnamed author of the editorial in the Economist.
This relates to the more general question of how emoters can be identi-
fied. The distinction between metonymic and non-metonymic emoters
may be of interest here (Table 3.9), with further sub-distinctions within
metonymic emoters (compare also Halliday & Matthiessen 2004: 202–3
on ‘Sensers’ in mental clauses).
The final example (a disappointed editorial) might be argued to include
a fusing of both affect (an emotional response is attributed to the author
of the editorial) and appreciation, since a semiotic phenomenon is
evaluated (compare Section 1.4), though to me affect seems to be what is
really at issue here, with the emoter having to be inferred by the reader.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 75

Table 3.9

Pattern Elements Examples

Non-metonymic conscious participants a frightened man
emoter (persons, animals) (BRC, CH5 4286)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Metonymic expressor (body part, Meredith’s surprised
emoter gesture) expression (BRC, CEB 933)
their anxious faces
(BRC, FAJ 2626)
human collectives/ there was surprise in
institutions/countries Whitehall (BRC, A4X 247)
and so on television is admired
abroad (BNC, HMK 3)
semiotic phenomenon: a disappointed editorial
product of emoter (BRC, A6F 1390)

In a second pattern of undirected affect – ADJ prep – the preposi-

tional phrase (with for) indicates an empathy target, that is, a person
with whom the emoter identifies (Table 3.10). The prepositional phrase
refers to a second hypothetical emoter: the empathy target. The speaker
assumes that this empathy target shares his/her emotion, for example in
the examples in Table 3.10 that ‘he’ is happy and that ‘Jimmy’ is disap-
pointed. In a sense, two emoters are construed: one actual (I) and one
imagined (him, Jimmy).

Table 3.10

Pattern Elements Examples

ADJ prep (for) n emoter is emotion for I’m happy for him (Hunston
empathy target 2003)
I’m very disappointed for
Jimmy (BNC, AHC 662)

Noun patterns
It is usually the non-count usages of emotion nouns that are used to
attribute emotions to emoters, as shown, for instance, in Table 3.11. As
previously mentioned (Section 2.5), emotion nouns are actually almost
always non-count, and if they can be used as count nouns they have
different meanings. In some cases, however, (for example with the noun
affection), even count usages of emotion nouns do realize emotional
responses (Table 3.12). With these two basic noun patterns, the emoter is
realized by a pronoun (my, his), adjective (for example filial, parental) or

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

76 Emotion Talk Across Corpora

Table 3.11

Pattern Elements Examples

adj N (non-count) emoter’s emotion his childlike affection (BRC, HGL 1451)
his not unappealing surprise (BRC,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

CJF 619)
his murderous hate (BRC, B1X 2559)

Table 3.12

Pattern Elements Examples

adj N (pl) emoter’s emotion my affections (BRC, B1X 3669)
men’s affections (BRC, KCA 421)

Table 3.13

Pattern Elements Examples

N prep (from, emotion from/of affection from her dear old Agnes
of ) n emoter (BRC, ASE 996)
the frank astonishment from his
agent (BNC, HGV 4809)
the hate of the evil terrorist minority
(BRC, AC2 149)

Table 3.14

Pattern Elements Examples

N prep (in/on) n emotion in/on genuine surprise in his voice (BRC,
expressor FRF 2234)
a lot of hate in their eyes (BRC,
KC6 2023)
the look of surprise on his face (BRC,
B20 1201)

genitive noun (men’s). The emoter can also be realized by a prepositional

phrase following the emotion noun (as in Table 3.13). We can also find
patterns where the emoter is metonymically realized in the form of an
expressor through a prepositional phrase (with in or on) following the
emotion noun (as in Table 3.14).
In patterns where the preposition precedes rather than follows the emo-
tion noun, it is the prepositional phrase that realizes the emotion, with a
preceding noun or verb indicating an expressor or an action (Table 3.15).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 77

Table 3.15

Pattern Elements Examples

n prep (of) N expressor of a look of clownish surprise
emotion (BRC, CJF 2159)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

an expression of absolute astonishment
(BRC, FPM 2700)
v/a prep action/expressor Giuseppe Patana conducts the score
(with/in) N with/in emotion with obvious affection (BRC, A35 307)
He squealed in . . . surprise (BRC, FPX 1301)
Gideon Eddy laid an arm across the
shoulders of Charlie Smith in mock
affection (BRC, HHC 2031)
was frozen with fear (Hunston 2003)

The relation that is expressed by the preposition may vary as in the

example in the table: in a look of clownish surprise, the look is specified
in more detail; in squealed in surprise and frozen with fear, the emotion is
the cause for what happens; in laid an arm across the shoulders of Charlie
Smith in mock affection and in conducts the score with obvious affection the
specification is one of manner. The preposition seems to depend on the
adjective, noun or verb preceding the emotion noun, rather than the
emotion noun itself (for example frozen with fear is classified as the pat-
tern ADJ with n by Hunston 2003), but the emotion noun is here taken
as the node, since my concern is not with valency/dependency as such
(compare my comments above). As Hunston & Francis point out, ‘the
same pattern can be described in terms of any one of its major elements’
(Hunston & Francis 2000: 45). It is worth emphasizing that adj with N is
therefore not a pattern in Hunston’s definition (though ADJ with n is),
but that it nevertheless can be included in a local grammar (S. Hunston,
email communication). Typical actions and expressors that are involved
in this pattern are:
• cognitive processes (remember, notice, reflect)
• linguistic processes (reiterate, greet, say, squeal, gasp)
• descriptions of eyes/face (face was scarlet and twisted, eyes blazing, eyes
smouldering, smile)
• descriptions of eye contact (looking, staring)

3.3.3 Directed affect

Patterns that specify explicitly the emotion and its trigger, and usually
also the emoter, can be classified as directed affect.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

78 Emotion Talk Across Corpora

Table 3.16

Pattern Elements Examples

Vn (emoter) emotion trigger He hates days when he can’t get
straight into his workshop.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(BNC, HA6 1149)
If you’d like a copy of those recipes,
we can easily let you have one (Francis
et al. 1996: 51)
(trigger) emotion emoter These stories surprised and moved me
(Francis et al. 1996: 53)
Cats could easily panic the birds
(Francis et al. 1996: 486)
It was a sight that never failed to thrill
her (Francis et al. 1996: 506)

Verb patterns
Starting with verbs, the first, basic pattern is V n (with the emoter realized
through a preceding or following noun phrase) as listed in Table 3.16. The
verb can also be followed by an -ing clause (V - ing: He likes walking his
dogs, He resented her doing well) or a wh-clause (V wh: Ihated what Horatia
Manners did to Andrew and Virginia). Depending on the meaning group of
the verb, the mapping of affect categories onto the sentence elements is
emoter emotion trigger or trigger emotion emoter (compare also Ortony
et al. 1987: 357–8 on English causative and noncausative affective verbs).
The passive variation of this pattern (n PV) changes the order of the
elements, but is used for similar purposes. With passives, the emoter
need not be expressed but may remain implicit (often retrievable from
the context):

(13) But emotionally, the non-people are hated or at least detested.

(BRC, A07 243)

If the emoter is expressed, it is either realized by an adverbial:

(14) Another Early music reviewer has criticized an Austrian performer

(much admired in France) (BRC, J1A 1547) [PV prep (in) n:
metonymically → ‘people in France’]

or through a by-structure:

(15) He . . . was respected and admired by all of his colleagues in the

forces (BNC, FR0 4141) [PV prep ( by) n]

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 79

We can also find prepositional verb patterns, with the PP expressing the
trigger. There are two variations of prepositional patterns: one where the
preposition follows the emotion verb directly (V prep n), and one where
it follows a noun phrase following the emotion verb (V n prep n). With
V prep n patterns the mapping is emoter emotion trigger (Table 3.17).

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Table 3.17

Pattern Elements Examples

V prep n emoter He still hankers after high office (Francis
(after, about, emotion et al. 1996: 152)
at, for, in, trigger Science fiction fans in Britain have been rejoicing
of, on, at the return of ‘Thunderbirds’ to their TVs.
over, (Francis et al. 1996: 167)
towards,
with) I yearned for something new (Francis et al.
1996: 180)
He did not care for the place (Francis
et al. 1996: 181)
I despair of her (BNC, CEE 288)
Marie’s parents dote on her (Francis et al. 1996: 218)
Students . . . are encouraged to strive towards a
high level of achievement (Francis et al. 1996: 257)
David thought she had cooled towards him
(Francis et al. 1996: 257)
I really sympathize with the two officers (Francis
et al. 1996: 263)

The preposition can also be followed by an -ing clause (V prep -ing: he still
worries about me being the youngest, He delights in stirring up controversy
and strife), a wh- finite clause (V prep wh: Maureen decided to live for the day
and not worry about what the future would bring for them), or a wh- infinitive
clause (V prep wh-inf: Many agonized over whether to take the offer).

Table 3.18

Pattern Elements Examples

V prep n trigger He has impressed as stand-in for the injured
emotion Tommy Wright (BNC, CBG 498)
trigger

A special case of the V prep n pattern is where the trigger is split, and
the mapping of the elements differs from that listed above. For instance,

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

80 Emotion Talk Across Corpora

Table 3.19

Pattern Elements Examples

V n prep n
V n about n emoter emotion trigger that’s what I hate about it (BNC,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

trigger CAE 1087)
V n as n emoter emotion trigger Rotha Lintorn Orman admired
trigger Mussolini as a man who had dealt
firmly with the socialist menace
(BNC, CS6 364)
YASSER Arafat, the chairman of the
PLO, long scorned as a proponent of
terrorist tactics (BNC, K35 1476)
V n by n emotion emoter trigger Lady Dawkings . . . surprised herself
by her reaction (BNC, EW1 428)
V n for n emoter emotion trigger I hated Mandeville for his
trigger arrogance (BRC, H90 606)
V n in n emoter emotion trigger We admire this characteristic in
trigger emotion emoter others (BNC, B19 1079)
trigger Determined but polite efforts were
made to interest her in the purchase
of tomatoes (BRC, ASN 2500)
V n on n emoter emotion trigger The white-bearded campaigner
prided himself on blunt, outspoken
views. (BNC, A1W 35)
V n with n trigger emotion emoter The compact circuit . . . has surprised
trigger owners Ladbrokes with its robust
evening trade (BRC, CH7 2099)

with impress (in Table 3.18). With V n prep n patterns, the mapping is
either emoter emotion trigger (trigger) or trigger emotion emoter (trig-
ger), as shown in Table 3.19. Again, the preposition can also be followed
by an -ing -clause (V n prep -ing: You probably hate me for saying that,
Some salesmen tried to interest me in buying property here) or wh-clause
(V n prep wh: But he has hated Paul for what happened to Johnny). The
mapping of the affect elements depends on the meaning group of the
verb (for example ‘surprise’ vs. ‘hate/admire’), and often involves two
triggers: for example someone is hated for doing something, someone
is admired for having a certain characteristic. In other words, the emo-
tion is directed at someone (trigger1) because of something (trigger2).
Arguably, in an example such as I admire you for your diligence the speaker
is saying that s/he admires the fact ‘that you are so diligent’ (that is, both

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 81

you and for your diligence can be analyzed as making up the phenomenon
that is admired). This occurs with other patterns, too (see below).
Again, there are passive patterns:

(16) [Taylor] [trigger1], admired [emotion] as the author of Holy Living

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

and Holy Dying [trigger2] (PV as) (BNC, CFF 1233)
(17) waterlilies [trigger1] have been admired [emotion] for generations
due to their exquisite blooms [trigger2] (PV due to) (BNC, CLT 177)
(18) Hazel [trigger1] was admired [emotion] for her cheerful efficiency
[trigger2] (PV for n) (BNC, HBH 774)

Moving on to clausal patterns, a finite clause following an emotion

verb can realize the trigger of the emoter’s emotion. Again, two pat-
terns can be found – V fin cl and V n fin cl. Let’s first look at V fin
cl (Table 3.20). The pattern V that seems to occur often with affective
mental process verbs (for example hope, fear) that project clauses (on
projection with mental clauses see Halliday & Matthiessen 2004: 206).
This also occurs commonly in the passive (Table 3.21). The pattern V n
fin cl with the elements emoter, emotion and trigger has a number of
variations (Table 3.22).

Table 3.20

Pattern Elements Examples

V fin cl (V emoter emotion I hate that I cause her so much pain (BNC,
that) trigger HUA 112)
The students fear that the government
does not intend to fulfil this demand
(Francis et al. 1996: 99)

Table 3.21

Pattern Elements Examples

it be V-ed that emotion trigger it was feared that a bomb had caused the
blast (Francis et al. 1996: 527)

Again, we can find two triggers with verbs such as admire (with the
object of admire as trigger1 and the adverbial clause with because, cos and
so on, as trigger2), whereas it in I hate/appreciate it functions as a dummy
object rather than as a second trigger, with the finite clause realizing the
notional trigger. The emotion may not have occurred yet, as in the last
example, where ‘we’ will only be surprised if ‘Trevor’ does in fact ‘stay on

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

82 Emotion Talk Across Corpora

the park until the end of the season’. This pattern seems to function in a
way similar to an if-clause, and can be paraphrased as ‘we’ll be surprised
if Trevor stays on the park. . .’.
The trigger can also be realized by a non-finite clause following an
emotion verb and a noun group (V n to inf ) or simply an emotion verb

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(V to inf ) (Table 3.23).

Table 3.22

Pattern Elements Examples

V n fin cl
Vn emoter emotion I admire her because she is an actress who
because trigger trigger can also sing (BNC, ADR 1686)
V it emoter emotion I hate it when people talk about me
fin clause trigger rather than to me. (BNC, G36 1624)
(that, I really appreciate it that you raised me in
when, if)6 such a warm and happy family (Francis
et al. 1996: 543)
Li Yuan would hate it if she spoiled his
surprise. (BRC, G04 1351)
V n and cl emotion emoter Let us hope Trevor can surprise us all and
trigger stay on the park until the end of the
season (BNC, FR9 813)

Table 3.23

Pattern Elements Examples

V n to emoter emotion They would prefer the truth to remain
inf trigger untold (Francis et al. 1996: 290)
He wanted her to go and buy some
presentable clothes she could wear to the
sort of smart restaurants he was taking her
to. (BRC, CH1 1840)
V to inf emoter emotion He hated to disappoint her (BNC, ACW 340)
trigger
She hoped to find an English audience
receptive to her watercolors (Francis et al.
1996: 92)
Douglas preferred to do his own driving
(Francis et al. 1996: 92)
. . . surprising himself to find that he actually
believed it. (BRC, EVG 2184)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 83

The final example in Table 3.23 involves a projected trigger: that is,
the surprise is really caused by ‘his’ believing (he actually believed it), not
by ‘his’ realization (find) of this belief.
Finally, a number of verb patterns have an empty subject it preceding
an emotion verb with a finite or non-finite clause following the emotion

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

verb, which indicates the trigger. The emoter is usually the object of the
emotion verb. Again, there are patterns where a noun (It V n cl) or a
prepositional phrase (It V to n cl) is between the verb and the following
clause, in contrast with (rarer) patterns where the clause directly follows
the verb (It V cl); see Table 3.24 on page 84.
In some cases, ‘to n’ realizes the emoter rather than ‘n’ (for example It
matters to everyone that the killer is still around, It didn’t matter to her what
happened to us). Again, the trigger may be projected with a verb such as
find as in the last example of Table 3.24, with the trigger being the fact
that the nail was vacant rather than the realization of this fact. With
if-clauses the emotion need not have happened yet, for example it
wouldn’t surprise me if. Whether the pattern It V fin cl occurs with
many emotion verbs (other than hurt indicating a general emotional
pain rather than labelling a specific emotional response) is doubtful, as
is the question if the pattern It V non-fin clause occurs commonly and
exists with other emotion verbs. (It hurts followed by an -ing or to-inf
clause is only attested very rarely in the BRC but, as an example, a search
for the phrase It hurts knowing came up with 547 google™ hits on 7
November 2006.)

Adjective patterns
Moving on to adjective patterns, the first pattern is ADJ n. Rather
than the noun indicating the emoter of an emotion (as above), the
head noun (or -ing form) here indicates the trigger of the emotion
(Table 3.25). Although the noun also points metonymically to the emoter
(for example Jon Hallworth, as the one facing the battle), the battle to
be fit in time for Saturday’s big Premier League kick-off seems to be what
is causing the anxiety and hence functions as a trigger. Similarly, in the
unutterably sad sound, the sound seems to cause sadness, is a trigger.
Directed affect can also be realized through adjective complementa-
tion patterns with prepositional phrases (Table 3.26). Alternatively, the
preposition can be followed by a wh-clause (ADJ prep wh: The next morn-
ing I awoke anxious over what had happened, I . . . was worried as to how
my death would affect them, pleased with how it’s going) or -ing clause
(ADJ prep -ing: pleased about continually having to clean his porch roof, the
wary fox, who is frightened at seeing all the footprints pointing towards the

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

84

Table 3.24

Pattern Elements Examples

It V fin cl emotion trigger It hurts that you cannot see this (Francis et al.
(that) 1996: 520)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

It V fin cl emotion trigger it hurts, being around someone who’s
(-ing, to) just killed someone (BNC, BN1 1333)
I know from personal experience how much
it hurts to lose one so close. (BNC, HD2 492)
It V n fin emotion emoter It hardly surprised her that the goldfish pool
cl (that, if, trigger had spread so much (BNC, HA0 2505)
wh, when) It annoyed me that I didn’t have more time
to do more ironing (Francis et al. 1996: 530)
. . . it wouldn’t surprise me in the least if . . .
you discovered it had nothing to do with
being at school (BNC, HUL 433)
It surprised her how little public knowledge
of Jonathan’s behaviour affected her.
(BNC, H8H 2222)
It amazes me how many plastic shopping
bags are given out by cashiers in large
supermarkets (Francis et al. 1996: 534)
It saddened me how these children have
accepted life in detention as normal (Francis
et al. 1996: 534)
It irritates me when I’m asked to do things
that are not part of my job (Francis et al.
1996: 535)
It V n emotion emoter It disturbs me to see you unhappy
non-fin cl trigger (Francis et al. 1996: 532)
(to, -ing) It worries me seeing him so helpless (Francis
et al. 1996: 536)
It scarcely surprised him to find . . . that the
nail beside it was also vacant (BNC, G0M
1327)

Table 3.25

Pattern Elements Examples

ADJ n emotion trigger OLDHAM keeper Jon Hallworth is facing an
anxious battle to be fit in time for Saturday’s
big Premier League kick-off. (BRC, CH7 869)
The unutterably sad sound of the sparrow
lingered in his ears (BNC, G0Y 2062)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 85

Table 3.26

Pattern Elements Examples

ADJ prep emoter emotion You needn’t sound so surprised about it
n (about, trigger (BRC, F9C 1038)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

as to, at, Modigliani was plagued and anxious
because of, because of the difficulty of finding models
by, for, in, (BNC, ANF 1504)
of, on, over,
He was definitely impressed by Karen’s
to, towards,
lovely grey mare (BNC, KS9 487)
with)
So Regan, anxious for an alternative
theoretical platform from which to put into
orbit his conviction (BRC, CM8 491)
General Noriega tried to isolate opposition
within the military after a similar coup
attempt last year by . . . retiring those thought
to be less enthusiastic in their support of
him. (BRC, A2X 306)
I was never more disappointed in anyone
(BNC, CDS 944)
they’re not so frightened of people (BNC,
K1E 1972)
It looks as if he’s not very keen on the price
we want to pay (BRC, KB7 1605)
General Haig was wedded to his profession
(Hunston 2003)
I’ve always felt very affectionate towards
Karen (Hunston 2003)
I am writing to say how delighted I am with
your magazine (BNC, HAC 5840)

den and none the other way). With these prepositional patterns the usual
order is emoter emotion trigger, with the trigger expressed by whatever
follows the preposition. As usual, the relation between trigger and emo-
tion varies according to emotion and the patterns can be used both to
attribute emotions directed at triggers to others (non-authorial affect)
and to oneself (authorial affect). With the pattern ADJ with n for n
(I was very disappointed with Trevor for retaliating) the trigger is again
split into two, the person at whom the emotion is directed (trigger1:
Trevor) and the cause for provoking this emotion (trigger2: retaliating). It
is an alternative for the structure ADJ that (see below), paraphrasable as
‘I was very disappointed [emotion] that Trevor retaliated [trigger]’.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

86 Emotion Talk Across Corpora

Table 3.27

Pattern Elements Examples

ADJ fin cl emoter emotion And they were all very impressed because
(because, trigger she’d done twenty different dinners, in

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

if, lest, twenty days. (BRC, KPV 5705)
that, when, She was continuously anxious lest Mitzi
wh/how) upset coffee all over them (BRC, AEA 1515)
He was very angry that she had spoken to
people about their private affairs (Francis
et al. 1998: 401)
[T]he people are terrified that . . . they may
be killed (Francis et al. 1998: 402)
He seemed disappointed when the dispute
was settled at higher levels (BNC, GT6 1413)
They are afraid what their neighbours and
children will think (Francis et al. 1998: 404)
Cathy was surprised how ill she looked
(BRC, HWP 259)
. . . your father and mother . . . may well be
pleased if your boyfriend takes the time to
have a chat with them (BNC, BPF 740)

Table 3.28

Pattern Elements Examples

ADJ non-fin emoter emotion You’ve got to be very thankful to win
cl (to, -ing) trigger once (Hunston 2003)
you would be surprised to find an STE
actually on sale. (BRC, A8R 185)
she had been happy working for Graham
(BRC, A6J 404)

As with verbs, there are a number of cases where a finite clause follow-
ing the emotion adjective indicates the trigger (Table 3.27). Depending
on the emotion adjective, the trigger may be something that has not
happened yet but which is feared or desired (for example anxious lest,
terrified that, afraid what) (cf. the discussion of irrealis triggers in Section
5.3.2.1). Depending on the clause type, the emotion may also not have
been experienced yet (if-clauses). There are also adjective patterns with
non-finite clauses (Table 3.28).
The trigger may be an event that has already happened/is happening
(We are pleased to see the Canadians do so well in the World Cup, telling

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 87

him how delighted she was to meet him), or not (she became frightened to
be left alone, The Democratic leadership in the Senate is anxious to reverse
the handsome victory by Mr Bush in the House of Representatives last week).
The to-infinitive clause often involves projection (cognition/perception):
They were puzzled to find the kitchen door locked; you would be surprised

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

to find an STE actually on sale; I was disappointed to find that accommodation
was scarcer . . . than I thought it would be; We are pleased to see the Canadians
do so well in the World Cup. As suggested above, this means that the
projected events actually cause the emotion (that the door is locked, that
an STE is on sale, that accommodation was scarcer, that the Canadians
did so well) rather than the cognitive process.
Finally, there are adjective patterns with an empty it subject. Even
though these all involve the verb make and could be classified as verb
patterns, it is the adjective that carries the affect meaning and is therefore
taken as the node. I follow Hunston (2003) in including such cases as
adjective patterns (Table 3.29). One of these examples (It makes me sad to
see all the good work we have done devalued) yet again involves a projected
trigger, with the ‘real’ trigger the ‘devaluing of the good work’ (rather
than the ‘seeing’).

Table 3.29

Pattern Elements Examples

It v n ADJ fin cl emoter emotion It makes me sick that anybody should
trigger doubt my commitment (Hunston 2003)
It v n ADJ non-fin emoter emotion It makes me sad to see all the good
cl (to, -ing) trigger work we have done devalued in
this way (Hunston 2003)
It makes me nervous not knowing
what they are using it for and to
potentially harm others7

Noun patterns
Finally, let’s briefly look at noun patterns: as with adjectives, there
are noun complementation patterns with prepositions, with the PP
realizing the trigger (Table 3.30). The preposition may depend on the
collocating verb: the preposition to is used with show, give and so on,
whereas on is used with bestow; other prepositions depend on the emo-
tion noun (surprise at/about, hate for). The preposition may also be
followed by an -ing clause (N prep -ing: his surprise at becoming Prime
Minister, She had given up all hope of ever bringing Oreste over) or wh-clause

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

88 Emotion Talk Across Corpora

(N prep wh: For a time the eagles were speechless with surprise at what had
happened ).8
Looking at prepositions preceding the emotion noun, a specific pattern
in this group is to N (‘astonishment’) used as an adverbial inserted in a
clause at different positions:

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(19) to my surprise about 99 per cent of them came back in (BRC, AK4
1023)
(20) By 27 April, the problem was solved, to the surprise of the
Americans (BRC, A64 1174)
(21) Ricket, to the surprise of the gallery, turned the match round
(BRC, A9H 56)
(22) Sonia and Helen flung themselves into my arms, to Joe’s great
astonishment (BNC, AC6 1306)

This pattern functions as a ‘stance adverbial’ (Biber et al. 1999: 856),

and is used to indicate the emoter’s surprise at the propositional content
of the clause, as an alternative to the structure X was surprised that p.

Table 3.30

Pattern Elements Examples

N prep n emoter emotion surprise . . . about her reference to a
(about, at, trigger referendum (BRC, APE 122)
for, on, to, full of hate for England and your church
towards) (BRC, B1X 1376)
bestow affection on his slightly wayward
children (BRC, CHG 1570)
showing . . . affection to members of the
opposite sex (BRC, A8M 64)
they . . . express affection towards Pamela
(BRC, B30 762)

Another example involving a preceding preposition is shown in

Table 3.31. This is similar to undirected noun patterns (see above); with
an additional specification of a projected trigger.

Table 3.31

Pattern Elements Examples

v prep (with) action with She realised with astonishment that he was
N that emotion trigger actually pleading (BRC, K8R 1649)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 89

There are also noun patterns with finite and non-finite clauses, with
the clause realizing the trigger (Table 3.32).

Table 3.32

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Pattern Elements Examples
N fin cl (that, emoter emotion astonishment that so many hotels border
when) trigger on Fawlty Towers (FrN)
a degree of surprise was expressed when
McEnroe showed up here after a knee
injury (BRC, A33 498)
N non-fin cl emoter emotion with one of her smallest smiles she
(to, -ing) trigger glanced at Caroline as if affecting surprise
to see her there (BRC, GUE 1348)
the initial wave of guilty surprise,
finding that the beautiful girl she’d seen
at the market had been Roman’s younger
sister (BRC, GUE 2012)

Undirected vs. directed affect patterns

It must be noted that many patterns for undirected affect can simply be
turned into directed affect by adding more elements, for example I jump
with surprise can become I jump with surprise when a dog suddenly barks at
me and the frank astonishment from his agent can be turned into the frank
astonishment from his agent at/that. I also assume that in many cases of
presumably ‘undirected’ affect, the trigger can in fact be found in the con-
text of the pattern or can be retrieved by the reader/hearer via inference.9
Listing a pattern as ‘undirected’ hence does not mean that the emotion
is undirected – it only means that the trigger is not explicitly realized
in the affect pattern. The difference from directed affect patterns is that
a potential slot for the trigger of the emotion is not filled. (Vice versa,
some ‘directed’ affect patterns can be shortened and turned into ‘undi-
rected’ affect: I was shocked at his lie – He had lied. I was shocked [trigger
mentioned in preceding co-text].) Real ‘undirected’ affect can occur:

(a) when an ‘undirected affect’ pattern is used and

(b) a trigger is not explicitly mentioned in the co-text and
(c) a trigger cannot be inferred from the co-text or world knowledge.

That is, in terms of the presence/absence of triggers, distinctions have to

be made as in Figure 3.2.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

90 Emotion Talk Across Corpora

trigger inferable

directed trigger in co-text

trigger in pattern

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

direction

undirected no trigger (mood)

Figure 3.2 Directed and undirected affect

‘True’ undirected affect is probably rare, though Martin & White sug-
gest that with feelings that are construed as undirected mood, ‘one might
pose the question “Why are you feeling that way?” and get the answer
“I’m not sure” ’ (Martin & White 2005: 47). And one of the factors that
distinguish moods psychologically from emotions is that they do not
have an object (for example, Planalp 1999: 13). However, the very fact
that one actually is provoked to pose that question (‘Why are you feeling
that way?’) is a clue as to the normal expectations of people that emo-
tions are usually caused by triggers. This is an area delineated for future
research: when and how often do speakers explicitly refer to triggers (use
‘directed affect’ patterns), when and how often are the triggers explicitly
mentioned in the context (anaphorically or cataphorically) rather than
in the pattern, and when and how often do they need to be inferred by
readers from the context or co-text? And, finally, when and how often do
we refer to ‘true’ undirected emotions, and how often does this indicate
some sort of deviation from the norm, such as a disease or illness (for
example depression, PMT)? Does the very fact that the emotion is not
caused by a trigger mean that the emoter and/or the emotion is evalu-
ated as ‘not normal’ in some way? For example, patients suffering from
temporal lobe epilepsy experience emotions that are not attached to any
kind of trigger (Oatley et al. 2006: 146, citing research by Maclean 1993).

3.3.4 Covert affect

From the outline above, three basic patterns emerge for the expression of
directed affect that are valid for adjectival, nominal, and verbal emotion
terms (ET):

ET PP: emotion trigger

ET fin/non-fin cl: emotion trigger
It ET fin/non-fin cl: emotion trigger

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 91

Table 3.33

Pattern Elements Examples

It v N (sg) non-fin emotion trigger It comes as a surprise to find him
cl (to) boasting (BRC, AHG 582)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

it was as much of a surprise to find
they had sausages (BRC, ACK 2921)
it was really a pleasant surprise to
have your letter (BRC, H9N 2559)
It v N (sg) fin cl emotion trigger It was no surprise when . . . they
(when, if, that) created a fine goal (BRC, CH3 7390)
It came as a very pleasant surprise to
me when I learned that (BRC, HHC
1994)
It will be a considerable surprise if
they fail to win (BRC, A9H 987)
It is no surprise that parliament
should have approved a conservative
plan (BRC, AAK 453)

In fact, as observant readers may have noted, no examples have so

far been given for the pattern It N fin/non-fin cl (that is, noun pat-
terns with a dummy it subject). The patterns can be found in Table 3.33.
These have not been discussed so far because surprise as a count noun
does not name an emotional response but rather denotes ‘an event,
a piece of news, etc. that is unexpected or that happens suddenly’
(OALD). Surprise in such examples can thus be paraphrased as ‘sur-
prising thing/event etc’, and is similar in its usage to the adjective
surprising:

It was as much of a surprise to find they had sausages – It was surprising

to . . .
It was no surprise when they created a fine goal – It was not surprising
when . . .
It will be a considerable surprise if they fail to win – It will be very
surprising if . . .
It came as no surprise that parliament should have approved a
conservative plan – It was not surprising that parliament . . .

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

92 Emotion Talk Across Corpora

But such usages arguably do imply the emotional response of an

emoter and can also be paraphrased by using the adjective surprised or
the verb surprise:

It comes as a surprise to find him boasting – I was surprised to find

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

him boasting – It surprised me to . . .
it was really a pleasant surprise to have your letter – I was pleasantly
surprised to have your letter
It will be a considerable surprise if they fail to win – I will be
considerably surprised if . . . – It will surprise me if . . .

Even though it looks at first glance like the emoter is absent in such
patterns, s/he is implied, and can in fact be explicitly referred to with
the help of a PP in the pattern a N to n, as in Table 3.34.

Table 3.34

Pattern Elements Examples

N (sg) prep emotion to that is not a surprise to me (BRC, AE0 2656)
(to) n emoter the triumph came as a surprise to many fans
(BRC, CH6 381)
they can hardly come as such a surprise to Daum
(BRC, CH3 7112)
his attitude to sex will come as a surprise to most
unenlightened adults (BRC, CH1 7466)

But patterns involving surprise (and some other emotion nouns) as

count noun (a/the surprise, surprises) denote events, or situations that
cause or trigger surprise in emoters, and thus have a different mean-
ing from the non-count patterns described above denoting emotional
responses. I therefore propose to treat these as only indirectly indicating
emotional responses of emoters, and will call this type of affect ‘covert
affect’ (cA) in contrast to ‘overt affect’ (oA).
Apart from the patterns already noted, other patterns also involve such
covert affect. With some, the trigger is the entity that is referred to as a
surprise (endophorically or exophorically), and the covert emoter can be
explicitly realized with a PP with for in the pattern N (sg) for n:

(23) What a lovely surprise/what a surprise [covert emotion of S] (BRC,

KDE 239, KDA 7560)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 93

(24) fans of his macho Dylan in Beverly Hills, [emoter] are in for a
surprise [covert emotion] (BRC, CH1 2055)
(25) you’ [emoter] re sure to meet a fiendish surprise [covert emotion]
(BRC, CH1 2666)
(26) she [emoter] had an even bigger surprise [covert emotion] in store

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(BRC, CH1 8250)
(27) the surprises [covert emotion] of the ‘alternative’ Nobel prizes
(BRC, A2M 198)
(28) there were no surprises [covert emotion] (BRC, A22 231)
(29) research has revealed some great surprises [covert emotion] in the
natural world (BRC, A1M 155)
(30) the analysis produces one further surprise [covert emotion] (BRC,
H8K 1430)
(31) the surprise [covert emotion] is the lack of drama [trigger] (BRC,
A3G 630)
(32) His axing [trigger] is a surprise [covert emotion] (BRC, CH7 2429)
(33) the surprise [covert emotion] of key change [trigger] (BRC, J1A 809)
(34) it was a lovely surprise [covert emotion] for everyone [emoter]
(BRC, CH6 4233)
(35) that would be a surprise [covert emotion] for your da [emoter]
(BRC, KDN 4008)
(36) tomorrow is a nice surprise [covert emotion] for you [emoter]
(BNC, KDE 90)

Finally, there are patterns where the emotion noun is used to

modify another noun (the pattern N n), for example a surprise
attack/defeat/visit/decision or a hate figure. The relations between the emo-
tion noun and the modified noun may differ, but often the modified
noun realizes the trigger of the emotion. For example, a surprise attack
is an attack that causes surprise, is surprising (similarly: a surprise buy-
ing rush/consolation goal/defeat/visit/ending) and a hate figure is someone
who causes the emotion of hate in many. But there are also more
lexicalized N n combinations such as surprise party (a party that is sur-
prising/unexpected), surprise bag (a bag which is filled with unknown
goodies), surprise element (which causes surprise). With hate, there are
also different relations: hate propaganda, hate campaign and hate mail
are motivated by hate, a hate trap is a trap caused by hate (you are
trapped by hate), hate comedy can refer to comedy that is made up of
hateful remarks about other people or social groups, hate symbols are sym-
bols of hate (symbols showing someone’s hate), a hate group advocates

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

94 Emotion Talk Across Corpora

hate and so on. There may also be two triggers in sentences with N n
patterns:

(37) Robert Owen [trigger1] . . . became a surprise [covert emotion] semi-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

finalist [trigger2] (BRC, A9R 126)
= ‘It was surprising that Robert Owen became a semi-finalist’
(38) East Germany’s embattled government resigned last night [trigger
1] in a surprise [covert emotion] step [trigger 2] (BRC, A7W 90)
= ‘the . . . government’s step in resigning last night was surprising’

This is very much newspaper language, and lexical items like step or
move appear delexicalized, almost like thing: a surprising step ∼ a surpris-
ing thing ∼ surprising. Arguably, it would also be possible to paraphrase
this example as ‘the government’s resignation was surprising’ (with one
trigger only). And we also find patterns involving projection:

(39) the surprise [covert emotion] announcement of British entry into

the Exchange Rate Mechanism [trigger] (BRC, APE 60)
= ‘the announcement that Britain would enter the Exchange Rate
Mechanism was surprising’

where it is perhaps not the announcement as such but rather the pro-
jected event (Britain’s entering the exchange rate mechanism) that is
surprising.
I propose to analyze such (N n) combinations according to their mean-
ings and contexts. Combinations like surprise attack and hate figure do
seem to imply the caused emotion of emoters, and should be included
in the analysis of affect (as cA). But lexicalizations such as surprise party
and surprise bag do not truly indicate affect, and should consequently be
excluded from the analysis of affect. Lexicalization can ‘bleach’ evalua-
tive (here affective) meaning (Martin & White 2005: 85). To investigate
the relation between different emotion nouns, the head nouns they can
modify, and the relations between the two would fill a book on its own.
Clearly, more research is needed in this area of word formation and syn-
tactic analysis: what is the role of appraisal here? On relations between
nouns in noun + noun sequences, see also Biber et al. (1999: 590–4), who
note (p. 590) that such sequences ‘are used to express a bewildering array
of logical relations.’

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 95

3.3.5 Summary
To briefly sum up the findings so far, we can describe affect patterns
according to three factors: (1) is an emoter present or absent (emoted
versus unemoted affect); (2) is a trigger present or absent (directed versus
undirected affect) and, (3) is an emotional response implied or directly

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

expressed (covert vs. overt affect)? We can also use these three factors
to describe sentences with affect patterns, for example:

(40) His axing is a surprise: emoted, directed, covert affect

(emoter = speaker; trigger = his axing; a surprise = something that
causes surprise) (BRC, CH7 2429)
(41) She realised with astonishment that he was actually pleading: emoted,
directed, overt affect (emoter = she; trigger = that he was actually
pleading; astonishment = emotional response) (BRC, K8R 1649)
(42) Books: The sorrows of love and the joys of political hate: unemoted,
undirected, overt affect (no emoter; no trigger; hate = emotional
response) (BRC, AK4 521)
(43) Oh mummy I’m sad: emoted, undirected, overt affect (emoter = I;
no trigger; sad = emotional response) (BRC, KC3 2183)

However, it must be kept in mind that undirected affect patterns may

in fact be used in the context of triggers or that the triggers can be
deduced by the reader/hearer. That is, the ultimate decision of what
kind of affect is present in discourse depends on more than just the
pattern. Furthermore, the description of affect patterns needs to be com-
plemented by additional corpus research.10 (In particular, adverbs were
excluded from this account because of lack of data. These have no com-
plementation patterns, but are nevertheless patterned in certain ways
(Hunston & Francis 2000: 42).) More theorizing is also necessary on pat-
terns and associated theories, as there are still many problematic areas
(Stubbs 2001: 460).
However, importantly for the purposes of this book, local grammars
can be used to ‘[quantify] ways of expressing meanings in different reg-
isters’ (Hunston 2002: 178). The advantage of such an approach lies in
the fact that the resulting frequencies of patterns can be related to the
meanings or functions of these patterns. This allows us to look at affect
patterns that are most frequent in a given register, and make informed
guesses about the typical usage of emotion terms in that register. Rather
than, say, looking at hundreds of concordance lines individually, the
frequent presence of a pattern as deduced from the collocates of an

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

96 Emotion Talk Across Corpora

emotion term will tell us that a particular function is most typical for
a given register. For example, let’s imagine that the adjectives happy,
delighted, pleased, frightened, disappointed frequently occur with the right-
hand collocate for in a given register, indicating the presence of the
pattern ADJ for. This pattern, as we know now, is used in sentences

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

such as:

He’s had a great career and I’m happy for him.

I am delighted for you, Alexandra.
I’m pleased for you, that I am.
I’m frightened for him.
I’m very disappointed for Jimmy.

With these adjectives the pattern ADJ for is used to express empathy
with someone, and this function appears particular important in our
(hypothetical) register. The point is that we can assume this without
needing to look at all concordance lines individually, though there is of
course a margin for error. This type of analysis proceeds from the assump-
tion that ‘meaning belongs not to the word but to the phrase’ (Hunston
2003: 349), and that focusing on the word as unit is not enough. This
means moving beyond the analysis of individual emotion terms and
concentrating instead on patterns of affect. This is the aim of Chapter 4,
where I will look at patterns and their functions in the different registers
in more detail.

Notes
1. The following discussion omits the frames Predicament, Emotion_Heat, and
Subject_Stimulus. The words participating in the Predicament frame (for
example bind, fix, jam, mess, misfortune, pickle, pinch, plight) do not gener-
ally denote emotions, and are not part of affect. Strictly speaking, the words
in the frame Emotion_Heat (boil, burn, chafe, fume, seethe, simmer, smoulder,
stew) also do not denote emotions but are metaphorical or metonymic ref-
erences to emotions, and also have other non-affect related meanings (water
boils). Furthermore, they often occur with supporting emotion terms. Look-
ing at the examples given in FrameNet, we can find that boil occurs 12 times
out of 12 with a supporting emotion term, burn 15 out of 15, chafe 1 out of
8, seethe 11 out of 21, simmer 15 out of 18, smoulder 13 out of 17, and stew 0
out of 6. With the exceptions of stew and chafe (the latter included in my list
of emotion terms), the majority of these terms usually seem to co-occur with
emotion terms, which means that the reference to an emotional response is
captured in the analysis in any case. Finally, almost all words participating in
the Subject_Stimulus frame (for example alarming, disturbing) were excluded

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 97

since they are classifiable as appreciation: reaction (Martin & White 2005: 58).
This means that affect patterns involving these words are not described.
2. I have excluded from the description word-specific patterns (be a rival/a com-
petition/connive for someone’s affections, replace/supersede someone in someone’s
affections, hold in affection, shower with affection, spring a surprise, take
by surprise, catch by surprise, and patterns involving the additional Degree

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

element mentioned by FrameNet (for example too ADJ to-inf; so ADJ that).
This means that the outline of affect patterns in this chapter does not offer a
complete description of all patterns of the individual lexical items. However,
I do mention even relatively rare patterns if they are needed for the sake of
completeness.
3. Annotation: capital letters refer to emotion terms, that is, N n means emotion
noun followed by another noun. They indicate that this is the term that I am
focusing on, rather than signalling that this is the term that the other elements
in the pattern depend on. So n = noun phrase/group (often including -ing
forms), v = verb phrase/group, adj = adjective phrase/group, pv = passive verb
phrase/group, prep = prepositional phrase/group, to-inf = to-infinitive clause,
that = that-clause, -ing = ing-clause, wh = wh-clause, fin cl = finite clause, non-
fin cl = non-finite clause (similar to Hunston & Francis 2000). Patterns that
involve ‘n’ often have alternatives with nominal/complement clauses (that,
-ing, wh-, to-infinitive). According to Biber et al. (1999: 8.12–14, Chapter 9)
that-clauses occur commonly with emotion verbs (for example hope, wish,
fear, worry that), adjectives (for example annoyed, glad, sad that), and certain
emotion nouns (for example hope, fear that). Wh-clauses also occur with emo-
tion verbs (for example like, hate, loathe, respect) and adjectives (for example
amazed what they do, amazed at how). To-infinitive clauses appear in structures
with emotion verbs (for example hope, wish, love, want, hate to), adjectives
(for example reluctant, keen, afraid, glad to), and nouns (for example desire,
wish, willingness to). Finally, -ing -clauses occur mostly with emotion verbs
(for example hate, prefer, worry about, admire, delight in) and adjectives (for
example afraid of, interested in, sorry for).
4. On prepositions used with emotional states in English from a cognitive lin-
guistic point of view, see for example Dirven (1997), Osmond (1997) and
Radden (1998). The kind of preposition used depends both on the nature of
the emotion term and on the nature of the trigger (Osmond 1997: 112) or on
the image schemata associated with prepositions (Radden 1998).
5. Another possible verb pattern candidate is n PV (She has . . . the look of a woman
surprised), but the -ed participle of surprise might also be classified as an adjec-
tive. There are cases in which it is difficult to differentiate between adjective
and V-ed patterns (compare also Martin & White 2005: 47 and Halliday &
Matthiessen 2004: 224).
6. The finite clause and the it are co-referential in such patterns (Biber et al. 1999:
662). Other affect patterns with co-referential it are:
(i) I love it here. (V it adv) (BRC, CK9 1666)
(ii) My family hated it in Southampton (V it prep) (Francis et al. 1996: 557)
7. This example is from a googleTM search of the internet, because a BRC
search (using the node ‘it makes’ and looking for it makes + n + emotion
adjective + -ing -clause) came up with no results, pointing to the limitations
of the corpus where rarer patterns are concerned.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 98 Emotion Talk Across Corpora

8. There are also patterns where a PP following the emotion term realizes the
emoter and the trigger is realized by a preceding noun, a following PP or a
that-clause:
(iii) The move (trigger) caused some surprise in Whitehall (emoter) (N in n)
(BRC, A4X 247)
(iv) there is surprise among outsiders (emoter) at his rapid rise (trigger) (N

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

among n) (BRC, AHJ 897)
(v) Within Whitehall (emoter), therefore, there was surprise and concern
that evacuation should have revealed so many symptoms of inner-
city poverty (trigger) (N within n) (BRC, HP2 270)

This seems to occur most often with the phrase THERE BE emotion as well
as passive verb patterns.
A special case involving PPs as trigger is also:

(vi) real affection between foster parents (emoter/trigger) and the children
(emoter/trigger) (N between n) (BRC, HP2 312)

Here we are dealing with a reciprocal emotion, meaning that the affection
is directed by the foster parents at the children, and by the children at the
foster parents. Both are thus simultaneously emoters and triggers. Between . . .
and thus construes a reciprocal emotion, a construal which is presumably
possible also with other emotion terms like animosity and love that can involve
reciprocal emoters.
9. Since speakers’ knowledge of emotions (emotion schemas) includes knowledge
about prototypical and potential antecedent events or eliciting conditions that
cause a specific emotion (see Section 5.3.1), triggers can be inferred by readers
with the help of co-textual (and other) information. White puts it like this:

Once the emotion portion of a scenario is instantiated, the listener may

draw inferences about how the events leading to or following from that
emotion are to be interpreted. In other words, the listener fills in other
portions of the schema through inference, even though the speaker has
not been explicit about those aspects of the events (White 1990: 60).

10. Some other patterns, which seem to be specific to individual emotion terms,
are presented in Table N.1 below:
There are also instances of V n in n and n PV realizing a very distinct
meaning of surprise:
(vii) She was quite sure she had surprised them in the act of carrying out the
next stage of the deception (BRC, GV2 3526)
(viii) No one . . . likes being surprised in the shower (BRC, BMR 1223)

The verb feel additionally involves patterns such as V n towards, V adj about,
It V adj to-inf (He feels no bitterness towards the British, He felt good about the
show, It feels good to have finished a piece of work), though it is not the verb
itself that carries the affect meaning but the noun or adjective functioning

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 A Local Grammar of Affect 99

as subject complement/predicative (bitterness, good). Such patterns are hence

listed as adjective or noun patterns rather than verb patterns. Furthermore,
particle verbs (fall for, look down on) are to be included in all patterns (patterns
are V particle . . . or V particle particle . . . rather than V). And some verb pat-
terns can involve a substitute (I hope so/I hope not). Finally, bear in mind that
the grammatical context can change patterns (Francis et al. 1996: 611–15,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Hunston & Francis 2000: 59–66).

Table N.1
Pattern Elements Example
Vnn (emoter) emotion She envies him the opportunities he will
trigger have to become big and powerful (Francis
et al. 1996: 276)
V n to n (emoter) emotion The goalkeeper seemed to prefer
trigger to trigger dribbling the ball up the field to defending
his goal (Francis et al. 1996: 427)
an N for n emotion for trigger She felt an affection for
Alison (BRC, APM 2598)
N of n emotion of Baby Sousan had been a gift from Allah, a
for n emoter/trigger for blessing, a reward for the affection of two
emoter/trigger people for each other (BRC, CEC 2017)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

4
Patterns of Affect Across Corpora

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

In the emotion profiles described in Chapter 2, the focus was solely
on individual emotion terms. This means that these profiles include
occurrences for all sorts of different types of affect (emoted, unemoted,
directed, undirected and so on), as there is no way of automatically
knowing how an emotion term is used in discourse by looking at its
frequency. As suggested in Section 3.3.5, it is only by looking at con-
text and patterns that we can make some hypotheses about the meaning
and functions of emotion terms. This is the aim of this chapter, which
reports some of the findings of an analysis of the lexico-grammatical
patterns of 15 emotion terms (see Table 3.1) in the BRC. The descrip-
tion focuses on L1 and R1 patterning, and is limited to detailing
those patterns that are most frequent in each sub-corpus.1 For exam-
ple, ’m is the most frequent L1 collocate of surprised in conversation,
but not, be and was are also common. Nevertheless, the main focus
is on ’m surprised. Furthermore, I only mention functions of emotion
terms that seem to cut across the usage of several terms – meaning
that for example the noun affection is not discussed in detail in news
reportage and academic discourse because it seems to have a distinct
patterning.
Three aspects of this analysis should thus be kept in mind:

• The patterns/functions that are outlined as particularly important for

a given register are not the only patterns/functions to occur in that
register.
• A pattern/function that may be characteristic for a given register may
also occur (more infrequently) in one of the other registers.
• Patterns/functions that occur very commonly across all corpora are
mostly disregarded.

100

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 101

Additionally, the findings represent only tendencies, since in some cases

the frequencies are not high enough to warrant general conclusions. In
the following, I describe each of the four sub-corpora in turn in terms of
its affect patterns: conversation (4.1), news reportage (4.2), fiction (4.3)
and academic discourse (4.4). I also compare affect triggers across the

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

four corpora (4.5).

4.1 Conversation

The first common function of emotion terms in conversation is to express

overt authorial affect (usually in the present tense). This is either realized
by a directed or by an undirected affect pattern (with undirected affect
patterns, the trigger has to be retrieved from elsewhere). This function is
especially common with four emotion terms: the adjectives surprised and
disappointed, and the verbs surprise and hate.2 Thus, the most frequent L1
collocate of surprised is ’m, followed by not. Typically, we find:

• I’m surprised + that-clause (usually without that)

• I’m surprised
• I’m not surprised

(1) David Dad! You’re not allowed to <pause> to talk like that to
people’s dog. And especially dogs we know.
Jane I’m surprised the boys let it run their garden [sic].
David What <pause> well it was David <pause> David’s idea, don’t
blame me. (BRC, KCH 6395)
(2) Albert he had a <pause> Honda Civic engine in it! <voice quality:
laughing> Oh, oh dear <end of voice quality>!
June I shouldn’t think anybody’d insure it!
Albert I’m surprised! (BRC, KB1 4164)
(3) well I mean I’m not surprised they don’t wanna take a picture of
our garden cos it’s a mess but er I’d have thought he would have
taken a picture of yours and sold you one. (BRC, KBG 1094)
(4) None My cat scares the dogs. <laugh>
Cherrilyn I’m not surprised! (BRC, KBL 3446)

It becomes clear that some of these imply some sort of negative evalu-
ation, as shown in (1) by the answer ‘don’t blame me’, but others seem
more neutral in this respect, for example (4).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

102 Emotion Talk Across Corpora

Another emotion adjective where authorial affect is common is disap-

pointed, which collocates frequently with (a) bit at L1. The most common
pattern with a bit is:
⎞ ⎛
I’m (in)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

⎟ ⎜
⎟ a bit disappointed ⎜ (that)
⎠ ⎝
I was (with)

That is, the pattern realizes either directed or undirected overt present
or past affect, with the past more frequent than the present. Some
examples are:

(5) I was a bit disappointed in those Laura Ashley dresses, they were
to [sic] tight for a start. (BRC, KCD 2234)
(6) I’m just a bit disappointed that th there really is a lot of money’s
worth of Lego and you’re not playing with it. (BRC, KCH 2338)
(7) Hm, I was a bit disappointed with that dress. (BRC, KCD 1127)
(8) I was a bit disappointed. (BRC, KC9 4828)

The collocation with a graduation resource (Martin & White 2005)

or evaluative modulator (Bednarek 2006a: 44) here downscales (rather
than upscales) the amount of disappointment that is attributed to the
self in conversation, that is, it mitigates the negative evaluation that is
expressed with regards to the trigger.
Authorial affect is also common with the verb surprise, where the most
frequent R1 collocate is me (and the most frequent L1 collocate is n
t,
followed by it). The following patterns are important for conversation:

• X (often it) SURPRISE (lemma) me (because/cos)

• It/that wouldn’t surprise me (at all)
• X (that/which/it) doesn’t surprise me

(9) He surprised me, he was the first one down them great big shutes.
(BRC, KBG 2205)
(10) But you know it surprises me I mean I know the church is, is closed
and it’s been closed for a couple of years now (BRC, KCC 272)
(11) That wouldn’t surprise me at all, the things I’ve seen them try to
eat. (BRC, KB8 2912)
(12) I mean my mother’s trying to book my holiday which doesn’t
surprise me, but then again she usually tries to anyway, where
you going?, what you doing? (BRC, KD6 4444)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 103

Again, note the (negative or positive) evaluation that is implied with

some of these examples, for instance (9) and (12).
Connected to this function is the usage of the noun surprise in conver-
sation, which collocates at L1 most frequently with the indefinite article
a. This indicates a covert affect pattern, namely a surprise as something

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

that causes surprise in implied emoters. The exclamation What a sur-
prise is very frequent, and clearly realizes authorial affect (with the self
as implied emoter):3

(13) Amanda Is one of those yours?

Kathleen No wa–, the bottom one’s Jenni’s? But what a surprise!
(BRC, KCW 1301)

Authorial affect is also important with the verb hate, which collocates
at L1 with I (and at R1 with it/to). That is, hate is clearly used to make
statements about the speaker hating something or hating to do some-
thing, overwhelmingly in the present tense, with various triggers (often
it with the trigger to be retrieved from the context):

(14) I hate the police (BRC, KBE 6520)

(15) I hate this babysitting job (BRC, KDN 3995)
(16) I hate you (BRC, KB8 250)
(17) I hate Vegemite (BRC, KPH 1)
(18) I do hate to cough (BRC, KBE 4947)
(19) Paul What do you do with the onion then?
Arthur Throw it at the cat <unclear>
A. loves the onion and she always eats the onion
Arthur It happens when you’re <unclear>
Arthur when you’re brought up in the war you see, waste not,
want not
A. I hate it (BRC, KP1 2513)

In some cases the trigger is conceptualized as a re-occurring event or

activity that keeps happening, to the annoyance of the emoter, when I
hate it is accompanied by a when-clause specifying the trigger:

(20) That is it and I hate it when you keep on and on nagging. (BRC,
KDT 366)
(21) I hate it when we sit around and sit around. (BRC, KC6 224)
(22) I hate it when I have my hair permed. (BRC, KBC 4586)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

104 Emotion Talk Across Corpora

I hate to can also be used with verbal processes as a disclaimer accom-

panying statements that are thereby construed as negative to the hearer
(sometimes jokingly):

(23) Dad they’re not gonna be looking at your legs, I hate to tell you

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

this. (BRC, KP6 1833)
(24) They’re out for <-|-> a laugh, I mean Charlie [. . .] he said you’ve got
to be really stupid to tell <pause> anyone anything in this place
and I said yeah but I’m not a proper Haileyburian I’ve just come
in, you know <pause> he goes yeah but you’re gonna get that way
I mean you are, you know, you are now <pause> I hate to te– I
hate to break it to you but you are. (BRC, KP6 2300)

However, hate to also occurs frequently in the phrase I’d hate to, with
the to-infinitive clause realizing hypothetical triggers:

(25) I’d hate to live that close to, to the river (BRC, KB8 7885)
(26) I’d hate to end up like that. (BRC, KBC 6330)
(27) I’d hate to have a broken jaw (BRC, KCE 6542)

In general, I hate (authorial affect) is clearly related to expressing neg-

ative affect with respect to triggers (though it is a dislike rather than a
‘hate’ as such that is expressed).
A second common function in conversation is the intensification of
affect, signalled by the collocation of an emotion adjective with an
amplifier – a resource of graduation (Martin & White 2005) or modula-
tion (Bednarek 2006a: 44). This is the case with the adjectives impressed
and pleased which both have very as their most frequent L1 collo-
cate in conversation (in contrast to disappointed which collocates with
(a) bit):

(28) And they were all very impressed because she’d done twenty
different dinners, in twenty days. (BRC, KPV 5705)
(29) Yes mind you I had to admit, you know me, I know, never have
been very impressed with him (BRC, KE2 6804)
(30) I’m very pleased about that. (BRC, KB8 11760)
(31) He didn’t look very pleased did he? (BRC, KB8 4085)

The third usage of emotion terms that is important in conversation is

the expression of authorial and non-authorial past affect – a collocation

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 105

with was at L1. This occurs with both delighted and frightened. On the
one hand, was delighted is used to refer to authorial and non-authorial
(mostly directed) affect:

(32) so yes yesterday I was delighted I was only nine stone thirteen

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(BRC, KD8 8761)
(33) And he was delighted that he learnt how to say, I’m having a lie-in
tomorrow morning. (BRC, KPU 2940)
(34) Oh her mum was delighted to see me. (BRC, KD9 1808)
(35) She was delighted with it. (BRC, KPU 143)
(36) She, she was delighted. (BRC, KB3 454)

As can be seen, directed affect with delighted can be realized by a that-

clause, a prepositional phrase (with) or a to-infinitive clause. However, as
the collocation analysis shows, to-infinitive clauses are most frequent in
conversation (the most frequent R1 collocate of delighted is to). Contrary
to expectation, this is not only used in the conversational formula I’m
delighted to (for example meet you), but also occurs with second and third
person emoters, in the past and in the future:

(37) I know that Bob and Glynis will be delighted to have some
because he’s been out of work. (BRC, KBF 5714)
(38) Good I’m delighted to hear that. (BRC, KPY 1012)
(39) Oh her mum was delighted to see me. (BRC, KD9 1808)

Was frightened realizes more often non-authorial (15 instances) rather

than authorial (7 instances) affect. With non-authorial affect this is often
mitigated (signalled here by underlining):

(40) I think he was frightened of <unclear> a little bit too far on the
edge I think. (BRC, KD0 4519)
(41) Course he was frightened to death weren’t he (BRC, KBE 2319)
(42) I felt really guilty about that, because I, the thought that he was
frightened of me shouting at him (BRC, KDW 6513)
(43) I think she was frightened that she was gonna get <pause> be
bitten. (BRC, KBF 12444)
(44) Perhaps she was frightened that <pause> she’d gotta (BRC, KCT
9247)

When non-authorial affect is not mitigated, it is assumed that the

speaker must have some knowledge concerning the third party emoter.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

106 Emotion Talk Across Corpora

In fact, such references can allude to a common experience in the past

that involve the speaker and the emoter:

(45) Well from when you first had the operation and you was fright-
ened keep bumping into you, and it’s, I mean that’s taken an awful

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

lot of confidence away and you still got <unclear> <pause> (BRC,
KC8 1710)
(46) Last time he flew out, you were frightened. (BRC, KB1 2042)
(47) Raymond I loved it! <pause> Perversely! <pause> But, to me
Margaret Mm.
Raymond it was er er er exciting <unclear>
Margaret Terrifying, yeah of course it was! It was
Raymond And my
Margaret different, yeah.
Raymond my sister didn’t, she was frightened (BRC, KDM
16630)

In conversation people are also typically said to be frightened of

something (realizing overt directed affect):

(48) But, eh listen, if, don’t be frighten if there’s anything that you
don’t know don’t you be frightened of asking him how to do it
(BRC, KB9 686)

Whereas authorial affect in conversation seems most often related to

the positive or negative evaluation of triggers on the part of speakers,
the other usages have more varied functions, for example the sharing of
emotional experience (on this aspect of social behaviour see Rimé et al.
1998), contributing to the evaluative part of a narrative (Labov 1972),
explaining people’s behaviour in terms of their emotion or reporting on
other people’s evaluations (in terms of their emotional responses). We
will look at these functions in slightly more detail in Chapter 6.

4.2 News reportage

As in conversation, the usage of emotion terms in news reportage is

associated with a number of different functions. Six main functions can
be identified. Firstly, emotion terms frequently realize first person affect.
Although this is an authorial affect pattern, more often than not this
occurs in quotations, and is hence attributed to emoters other than the
news writer. Emotion terms that have I as most frequent L1 collocate are

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 107

the verbs hate and admire but first person also occurs commonly with
other emotion terms (see below).
Concerning hate, this is used to realize overt and directed affect with
the trigger often expressed by definite noun phrases (for example hate
the film/programme/word/Mogadishu government/ANC/literary critic) or a

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

to-infinitive clause (often with see):

(49) ‘I love animals and hate to see them killed,’ said Clara McKay
(BRC, CH2 2138)
(50) I’d hate to see throw-away goals like that on a regular basis (BRC,
CH7 3820)
(51) I hate to see them do that (BRC, CH1 2298)
(52) I would hate to see Midland go (BRC, AKL 333)

Admire in news reportage is most often followed at R1 by the, indi-

cating the trigger of the emotion (for example admire the play/the
old classical musicals/the courage/the portrait/the ingenuity), that is, with
directed affect.4 But other structures also occur as triggers:

(53) Husband Anthony, 47, said: ‘I admire her guts.’ (BRC, CH2 1171)
(54) Javier Clemente, Spain’s new manager, says: ‘Everybody knows
how I admire English football . . .’ (BRC, CH7 4421)

First person affect is also frequent with was surprised (a common pattern
in news reportage), which occurs 15 times in the news sub-corpus. Nine
of these occurrences collocate with I, usually in quoted speech (and often
in sports reporting),5 for example:

(55) In an astonishing outburst on the Barnet Clubcall line, Flashman

said: ‘I was surprised at the stupidity of people having a go at
me. [. . .]’ (BRC, CH3 4972)
(56) Angry Jemson suffered the embarrassment of coming on as substi-
tute and then being substituted himself at Carrow Road. Now he
is gunning for Wednesday boss Francis for showdown talks about
his on-off move back to Nottingham Forest. ‘I was surprised to be
taken off,’ said fans’ favourite Jemson. (BRC, CH3 3918)

Secondly, emotion terms also occur commonly in news reportage

realizing non-authorial (non first person) affect; that is, statements are
made by news writers about emoters’ emotions (sometimes in reported
speech). For example, the verb surprise is most often followed at R1 by

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

108 Emotion Talk Across Corpora

the (indicating a definite noun phrase realizing the emoter). The emoter
is often metonymic:

(57) NAMIBIA’S seven political parties in the Constituent Assembly

surprised the country by announcing their agreement on a consti-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

tution with ‘only technical and minor amendments’ outstanding.
(BRC, AAK 247)
(58) The Chancellor said the British people had responded to privat-
isation, trade union reform, deregulation and tax reform with a
dynamism that had surprised the world. (BRC, A59 161)
(59) He surprised the nation with the announcement, made in a radio
and television broadcast on Saturday night, that his government
has refused registration to any of the 13 political associations that
applied for it. (BRC, A3U 163)
(60) MICHELIN, the world’s largest tyre maker, which sounded a warn-
ing of the depths of recession when it plunged £527 m into the red
last year, yesterday surprised the market with an operating profit
throughout 1991. (BRC, AKL 196)

Clearly, this contributes to the news value of unexpectedness (Bednarek

2006a: 171) as well as intensification: emoters are not just some news
actors, but rather the whole country, world or nation.6
Concerning the verb hate, he hates/d is also a relatively frequent
collocation in news reportage (often in reported speech):

(61) Eubank never stops telling us he hates boxing. (BRC, CH3 4470)
(62) Greenaway has said he hates the idea of a shot in a film being only
a preposition, linking what went before and after (BRC, A5B 56)

In other (non-reported) instances, the reader has to assume that the

news writer’s (usually unmitigated) attribution is based on some sort of
evidence. Often, this can be found in the wider context:

(63) He had fled the invading Germans, only to find himself in a society
he hated for its philistinism and prudery. ‘Here there is no such
thing as art, only business,’ he wrote to a friend. (BRC, AJX 658)
(64) What a blinding man, and I’m so very pleased that Michael is seem-
ing to make an amazing recovery. ‘I’ve spoken to him a few times
and he’s the epitome of a true warrior. He still wants to fight,
believes he still has it in him to fight . . .’ ‘He’s still sharp, has a
sense of humour, he’s witty. He’s bitter about what happened – he
didn’t want to lose. He hated that . . .’ (BRC, CH3 3732)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 109

Where this is not the case, readers have to take the writer’s word for it:
instances of she hates and he hates are not preceded by mitigators such as
seem/appear.
With delighted, the most frequent L1 collocate is was and the most
frequent R1 collocate is to, though with occurs also reasonably frequently

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

at R1. This indicates most often non-authorial affect (21 occurrences)
(sometimes in indirect speech). Here are some examples for was delighted:

(65) He told them: ‘I was delighted and over the moon to hear that
I was the father of a blue-eyed, black-haired baby girl. (BRC,
CH2 131) [authorial, direct speech]
(66) Mr Kinnock said he was delighted at Mr Prescott’s election, adding
he was ‘a dear and close friend’. (BRC, A1Y 291) [non-authorial,
indirect speech]
(67) Francis was delighted at landing Bright –; even if it was at the
third time of asking. (BRC, CH3 2186) [non-authorial]

With the collocation delighted to, no pattern clearly dominates. Instead,

delighted to occurs both as authorial (first person) and non-authorial
affect, in the present, past, future and hypothetical:

Authorial: Non-authorial:
I am/I’m delighted to (3) X was delighted to (5)
We are delighted to (2) X were delighted to (2)
I’d be delighted to (1) X are delighted to (2)
I was delighted to (1) X would be delighted to (1)
I shall be delighted to (1) X will be delighted to (1)

With authorial affect (I/we), non-past is more frequent than past, but
with non-authorial affect past is more frequent than non-past.
Non-authorial affect is also common with anxious, where the most
frequent L1 collocate is is, and the most frequent R1 collocate is to.7
Concerning is anxious, instances are also always used with a following
to-infinitive clause (apart from one exception, anxious for). A common
pattern is thus X is anxious to-infinitive clause, realizing non-authorial
overt and directed affect, making statements about news actors’ present
(not past) wishes. Typical examples are:

(68) the Soviet Union is anxious for matching changes on the Nato
side. (BRC, A88 79)
(69) So President Bush is anxious not to have a crisis. (BRC, A2X 133)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

110 Emotion Talk Across Corpora

(70) The Democratic leadership in the Senate is anxious to reverse the

handsome victory by Mr Bush in the House of Representatives last
week (BRC, A1G 457)
(71) The striker is anxious to shake off a thigh strain and said: ‘I trained
today, and I shall see how it reacts. (BRC, CH7 1102)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Finally, frightened collocates commonly at L1 with are:

(72) Russian souls are ‘unsure, doubting, vulnerable, continually on

the look out for deception’ and are frightened by choice. (BRC,
AHT 422)
(73) What people are frightened of is buying a property, finding the
tenant goes bankrupt and leaves them with the running cost (BRC,
AHJ 585)
(74) Blackpool tourism boss Barry Morris said: ‘People are frightened
of spending money.’ (BRC, CH6 9462)
(75) ‘Grizzlies are much faster than men, and when they are frightened
there are very few more dangerous creatures.’ (BRC, CH2 5864)
(76) ‘My greatest impulse was to hug them –; but you are afraid to do so
because they are so fragile, you are frightened they might break.’
(BRC, CH2 10468) (referring to the speaker as emoter as well as
people in general)

It is worth pointing out that those that are truly non-authorial (72–75) are
all based on ‘experts’ comments: (72) is based on a study after 18 months
of research of the market, (73) is an analyst’s comment, (74) is a tourism
expert’s comment, (75) is uttered by Jasper Park warden Gerry Israelson.
A common function, too, in news reportage is the usage of emotion
nouns to premodify another noun (the pattern N n), for instance with
surprise and hate. In fact, the noun–noun pattern is characteristic of news
discourse in general: ‘News has by far the greatest number of premodi-
fying nouns that are productive in combining with many head nouns’
(Biber et al. 1999: 592).
As in conversation, a is most frequent at L1 position with surprise,
indicating covert rather than overt affect. However, in contrast to con-
versation, this occurs most often in the pattern a surprise N, for example:
a surprise addition/ attack/ choice/ entry/ gift/ move/ selection/ defeat/ visit/
victory, that is, as a premodifying noun.

(77) THE IDEA that the Warsaw Pact would be able to mount a surprise
attack on Nato will shortly be ‘barely plausible’ (BRC, A2X 501)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 111

The noun hate occurs relatively infrequently in news reportage, but

among the 14 instances that do occur the use of hate to premodify
another noun is common: hate symbols, hate comedy, hate mail, hate trap,
hate figure, hate campaign occur. (Less frequently, hate occurs as unemoted
affect or as emoted directed/undirected overt affect.) This is similar to the

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

usage of surprise above and realizes covert affect. Examples can appear in
headlines in newspapers as an emotional ‘flash’, signalling to the reader
that the following text is interesting to read in terms of some emotional
significance:

(78) THE HATE TRAP

Homes can’t be sold
Couples can’t split
Families live in hell
ROGER TODD
THOUSANDS of couples who hate each other are trapped together
in a living hell because of the slump.
The tormented partners have had to put off their divorce actions
as they can’t sell their homes.
Now they are stuck with each other, locked in violence and misery
with no end in sight, lawyers revealed yesterday.
MEN explode in frustration and batter their wives.
WOMEN strike back with anything that comes to hand – including
knives and rolling pins.
SOME have nervous breakdowns and suffer stress and high blood
pressure.
(BRC, CH2 9181–9191)

Here the affect indicated in the headline (hate trap) is picked up several
times in the text, for example: families live in hell, couples who hate each
other, a living hell, tormented, misery, frustration, nervous breakdowns and
so on. In fact, both parts of the noun phrase hate trap re-occur in some
form in the body of the text, with trap referred back to with modality
(Homes can’t be sold, Couples can’t split, can’t sell their homes) and lexical
items (trapped together, stuck with each other, locked). This textual pat-
terning confirms White’s finding that news reports typically have two
phases: ‘an opening nucleus containing the text’s core informational
and interpersonal meanings; a subsequent development stage which
acts not to introduce new meanings but to qualify, elaborate, explain
and appraise the meanings already presented in the opening “nucleus”.’
(White 1997: 111).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

112 Emotion Talk Across Corpora

The intensification of affect, realized by the collocation of emotion

adjectives with an amplifier (noted with respect to the conversation
sub-corpus above), also occurs in news reportage, here with the adjec-
tives disappointed, impressed, and pleased. Disappointed in fact occurs
equally frequently at L1 with be and the adverb very. All of the mod-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

ulated/graduated affect instances occur in quotations (mostly with first
person and present tense), for instance:

(79) Maggie Mukasa, 39 and a schoolteacher, said: ‘Although I will

be better off, I am very disappointed by the outcome . . .’ (BRC,
AJX 336)
(80) ‘I’m obviously very disappointed for the players,’ said the man-
ager, Bobby Graham. (BRC, A9R 116)
(81) He said: ‘Naturally I am very disappointed, especially as I felt that
the evidence and explanations produced would be accepted . . .’
(BRC, CH7 404)
(82) Hammam said: ‘Vinny is like an adopted son to me and I’m very
disappointed . . .’ (BRC, CH3 6439)

The adjective impressed similarly collocates with an amplifier (so), which

is part of a pattern: someone is reported to be so impressed that they do
something:

(83) Milton ward Tories were so impressed by his la-de-da-accent and

gold-plated walking stick that they made him social secretary.
(BRC, CH1 6180)
(84) Manchester United manager Alex Ferguson revealed last night he is
so impressed by Wallace’s comeback that he is considering making
room for him in his team’s title push. (BRC, CH3 4688)

In other words, the attribution of overt affect is used to give a moti-

vation for news actors’ behaviour towards the trigger or the person
metonymically responsible for the trigger.
Finally, pleased also collocates most commonly at L1 with very. Typical
examples are:

(85) ‘I am obviously very pleased about winning but it was also great
that my parents were here . . .’ (BRC, A9R 660)
(86) The officer added: ‘We’re very pleased he has been caught . . .’
(BRC, CH2 5397)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 113

(87) He is very pleased with the ‘good meaty topics’ departments have
been suggesting for scrutiny this year (BRC, A1J 167)
(88) Downing Street said the Cabinet were ‘very pleased’ that interest
rates had been reduced to 10 per cent. (BRC, CH2 5797)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Another common function in news reportage – hedging, hypothe-
sizing or predicting affect – is realized by the colligation of affect with
modality, often becoming apparent through a left-hand collocation with
be (surprised, disappointed, willing). Looking at the larger patterns in which
be surprised occurs we find as most frequent the following phrases:

• I would be surprised if. . .

• I would not/wouldn’t be surprised if. . .
• I would not/wouldn’t be surprised to see/hear. . .

Some of these are similar to conversation (if-trigger) and in fact occur in

quotations, often judging the likelihood of events happening:8

(89) ‘I would be surprised if it doesn’t test 2,200,’ says Mr Tora, a view

shared by others. (BRC, A3J 18)
(90) Park – like Gloucester, Bristol and Saracens – are unbeaten in the
league so far. ‘Don’t be surprised if we get a good result,’ Colin
Horgan, a spokesman for the club, said yesterday. (BRC, A5U 400)
(91) ‘I would not be surprised to see Footsie (FT-SE 100 index) off
another 50 points,’ says Fred Carr (BRC, A3J 19)

Similarly, the collocation of disappointed with be, which (in addition to

very) is also frequent at L1, results from a colligation with modal verbs,
most frequently will (’ll, won’t) (other L1 expressions are can’t, only to
be, want to be, was not to be, is likely to be, would). That is, the most
common pattern here is: [second or third person] WILL be disappointed,
for example:

(92) Cheap-thrill seekers will be disappointed: the result is realistically

dull and unpleasant. (BRC, AHA 768)
(93) But I fear she . . . will only be disappointed by what Mr Davy has
to offer. (BRC, CH1 1679)
(94) But if you fancy a spot of luxury, too, you won’t be disappointed.
(BRC, CH1 1118)
(95) EVEN if you haven’t read the enchanting novel on which THE
POWER OF ONE (Cert PG; General) is based, I think you’ll still be
disappointed (BRC, CH1 846)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

114 Emotion Talk Across Corpora

Most of these seem clearly to be used to evaluate things, for example

the film in (95), in terms of the emotional responses they will produce
in emoters (for example the reader).
With be willing, in the majority of examples it is used in cases where a
speaker says that an emoter would be willing to do something:9

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(96) Shah declined, but said he would be willing to help finance it for
a fee. (BRC, A2S 492)
(97) Earlier, Geoffrey Shaw, for Mrs Sutcliffe, said she would be willing
to accept whatever damages figure the court substituted for the
£ 600,000 award. (BRC, A30 409)
(98) His friends have also suggested that he would be willing to sell on
a stake in MK Trafford Holdings (BRC, A33 64)
(99) Peter Ashman, legal secretary of Justice, said: ‘. . . When you talked
to judges in the Sixties, they might have told you they didn’t
happen at all or would be willing to concede that a miscarriage
might take place once every 10 years.’ (BRC, AHM 126–127)

Other instances of be willing are also modalized (X were unlikely to be

willing to, X will eventually be willing to, X may be willing to). That is, a
frequent use of willing is to make statements about what news actors are
(un)likely to be willing to do or will be willing to do (in reported speech,
with time shift realized by would).
Finally, the pattern A n is also common in news reportage, realizing
undirected overt affect. This occurs for example with the adjectives anx-
ious, delighted and enthusiastic (but not with impressed and pleased), with
the noun often denoting an emoter directly:

(100) BIG SHOT: delighted Andrew clinched the match (BRC, CH6
5747)
(101) ‘. . . you’d better have one to remind you what a foetus looks like
at 12 weeks,’ an enthusiastic PR said. (BRC, A1Y 423)

Anxious (while also occurring with human participants as emoters, for

example anxious Hong Kong citizens) has a semantic preference for nouns
indicating ‘time duration’ (for example moment, time, week), which seems
to be particularly important in news discourse (though anxious moment(s)
also occurs in fiction):

(102) The 26-year-old midfielder, who played in all three of England’s

European championship games, now faces an anxious wait for
Taylor’s verdict. (BRC, CH7 2804)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 115

(103) JONATHAN SPEELMAN gave his supporters an anxious day in

the sixth round of the Pilkington Glass World Chess Champi-
onship semi-finals yesterday. (BRC, A4K 729)

What these examples are saying is that these times (moments, days and

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

so on) are spent by emoters in a state of anxiety because of something
that happens during them. It is not the days themselves that cause the
anxiety, but the happenings during these days.
Enthusiastic also occurs frequently with nouns (though this is in fact
common in all the written registers not just in news reportage):

(104) an enthusiastic 15-handicap golfer (BRC, CH3 33)

(105) an enthusiastic American super-star (BRC, AA9 217)
(106) an enthusiastic bargain hunter (BRC, AJP 234)
(107) the enthusiastic and emotional crowd (BRC, A9M 5)

Here, there is a difference between examples such as (106) and (107).

With the latter, the noun only realizes the emoter (the crowd), with the
former (106) the noun simultaneously realizes the emoter and the trig-
ger: enthusiastic bargain-hunters are people [emoter] who are enthusiastic
about bargain-hunting [trigger]. The noun can also realize an emoter
metonymically:

(108) an enthusiastic response (BRC, AKB 355)

(109) the enthusiastic admiration (BRC, CH1 5835)
(110) the enthusiastic applause (BRC, AA9 184)

Since there is a contiguity relation between the pre-modified noun and

the one ‘responsible’ for the action described by the noun, the emoter is
realized metonymically (Section 3.3.2).
Concerning the function of emotion terms in news reportage in gen-
eral, a safe speculation is that this is related to the news value of
colour. This stipulates that newspaper articles should highlight emotion-
ally relevant aspects (Ungerer 1997: 318). We can safely assume that
many statements of authorial and non-authorial affect are the result of
reporters’ questions concerning news actors’ feelings:

We ordinarily take our naïve desire to know the emotions of others as

natural. In American popular culture, conscious emotional experience
is a focus of interest that shows up in the themes of TV talk shows,
melodramas, and even broadcasts of so-called hard news. ‘Can you
put into words what you are feeling right now?’ Such are the urgent

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

116 Emotion Talk Across Corpora

questions with which nosy news reporters badger victims of tragedies,

harass bereaved relatives, and fawn on winners of sporting events
(Palmer & Occhi 1999: vii).

This has also been nicely captured in a Doonesbury cartoon by Garry

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Trudeau (December 13, 1985), which cannot be reproduced here for
copyright reasons. This features a horde of journalists at a press con-
ference asking about a character’s (Zonker) feelings:

Journalist 1: Zonker! How did you feel when you heard the good
news?
Journalist 2: Yeah, what were your feelings? Could you describe
them?
Zonker: Well, at first I didn’t feel anything. I just went numb.
Then I felt a rush of giddiness, followed by feelings
of disorientation, queasiness, shortness of breath . . .
hunger, rage, sexual longing, vertigo, boredom, and
finally, a tingling sensation.
Journalist 3: What about after the news sunk in?

Additionally, negative emotions contribute to the news value of neg-

ativity (the negative makes the news) and emotion terms relating to
(non-negated) surprise have to do with the news value of unexpected-
ness (Bednarek 2006a: 171). The intensification of affect (via amplifiers)
might be related to the news value of superlativeness, that is, the fact that
‘the bigger, the better, the more x, the more newsworthy’ an event (Bell
1991, Bednarek 2006a: 17) as well as the importance for intensification
in news reportage in general (White 1997: 109).

4.3 Fiction

Fiction uses emotion terms mainly to describe character emotion in nar-

rative and dialogue (mimicking conversation), and is also characterized
by the frequency of references to expressors.
Describing character emotion can be achieved through overt author-
ial (either with a first person narrator or in the depiction of character’s
speech) or non-authorial affect. Such descriptions are very much associ-
ated with the past tense: eight of 15 emotion terms have was as their most
frequent left-hand collocate: surprised, delighted, anxious, disappointed,
frightened, impressed, pleased, and willing.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 117

Starting with was surprised, this occurs most often in the patterns:

• and emotion adjective (surprised and touched)

• at/by/that/to-inf clause (with mental process verbs as a large sub-class)
• . . . and was surprised . . .

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

• emoter (name or pronoun) was surprised

For example:

(111) Anne was surprised and fascinated by Penny’s earthy language,

delivered in the same plummy voice. (BRC, G16 1428)
(112) In the full light she was surprised to see he was much younger
than his bulky, overcoat-clad figure had led her to believe. (BRC,
BMW 535)
(113) On the day of which we are speaking, the mother came home
from work and was surprised to find her daughter not yet in
from school. (BRC, H9G 158)
(114) I was surprised, but she said Bill had given her a list to be a help
with messages. (BRC, BP9 805)

Moving on to delighted, was delighted is used mostly realizing non-

authorial affect (with third person emoters). It occurs both in directed
and undirected affect patterns, with a variety of different patterns real-
izing directed affect (PP with at, by, with, that-clause, to-infinitive clause,
when-clause). Examples for both directed and undirected affect patterns
in fiction are:

(115) She was delighted at having achieved her aim and made her
mother notice her. (BRC, APU 234)
(116) Elizabeth Mowbray was delighted by the marked change in her
daughter that followed Joan’s arrival. (BRC, CCD 170)
(117) Jay was delighted with the compliment and intrigued to see
inside Lucy’s flat. (BRC, A0L 149)
(118) He was delighted that it was already, even before he did this,
addressed to him. (BRC, AR2 858)
(119) Mother Francis was delighted to see the new friendship develop-
ing (BRC, CCM 606)
(120) She had to admit . . . that she was delighted when old Lady
Lassiter moved to the Dower House (BRC, C98 2032)
(121) The whole community was delighted for her. (BRC, CCM 204)
[undirected, with empathy target]

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

118 Emotion Talk Across Corpora

(122) There was no reason to be afraid; on the contrary, he was

delighted. (BRC, A6N 2048) [undirected]
Concerning was anxious, this occurs predominantly realizing ‘volition’
and directed affect, but also (rarely) realizing ‘anxiety’ and undirected

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

affect:

(123) She was anxious that Swire Sugden should not be allowed to
spread himself around like a malignant tumour. (BRC, ACK 24)
(124) As he drove through the night, heading back to London, New-
man was anxious to get clear of Hauser’s ‘province’. (BRC, CN3
2694)
(125) Her face was anxious and sweet. (BRC, GUK 90)
(126) Vi was anxious. (BRC, CEH 1910)

Some common patterns with the ‘volition’ meaning (anxious to) are:

• X was anxious to
• . . . , anxious to
• and anxious to
• I’m anxious to know how/when
• X BE so anxious to
• X link verb very anxious to
• X were anxious to

Looking at was disappointed, was frightened, was impressed, and was

pleased, these are also used simply to describe characters’ emotions (often
directed at triggers) – that is, for overt, directed or undirected affect. They
also occur in character dialogue. Some typical examples are:

(127) Anne was disappointed that her plan was dismissed but sug-
gested inviting Chris for meals. (BRC, G16 2521)
(128) Taking care not to damage the fish, Yanto withdrew it from the
trap. He was disappointed, ‘just a babee’ he muttered to himself;
still, it would keep him in cigarettes and cider for a week. (BRC,
B3J 52–53)
(129) Joey was disappointed not to be able to wear a ‘monkey suit’ as
he called it, but Carmella was adamant. (BRC, ATE 3165)
(130) ‘I don’t know where she is!’ he admitted, reluctantly. ‘I was
disappointed when I found you didn’t either. . .’ (BRC, AB9 199)
(131) She was frightened and she was sick. (BRC, FSF 1690)
(132) He was frightened of touching her. (BRC, GWF 66)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 119

(133) Suddenly George was frightened that the man would think him
mean. (BRC, FAB 1579)
(134) ‘Sir John, I was frightened. . . .’ (BRC, K95 3401)
(135) As he entered she was impressed by his appearance. (BRC, C98
2430)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(136) Preston was impressed. (BRC, F9C 1746)
(137) I could see what went on through the two front windows despite
the 4p off Whiskas stickers, and I have to admit I was impressed.
(BRC, HW8 471)
(138) She was pleased an’ all. (BRC, CK9 1541) [undirected]
(139) He was pleased for his Dad, and things were better between
them. (BRC, ALJ 1964) [undirected, with empathy target]
(140) He could see that Elizabeth was pleased by his attempts to
describe it all (BRC, C98 1851)
(141) Athelstan was pleased he had judged the moment right (BRC,
K95 3567)
(142) In answer to one such question, he replied that he was pleased
to say that he was now being given full support by the unions
(BRC, AC2 572)
(143) Miss Kenton did not attempt to introduce further flowers into
my pantry, and in general, I was pleased to observe, she went
about settling in impressively. (BRC, AR3 618)
(144) Elinor was pleased to see him. (BRC, ASS 1422)

Finally, let’s have a look at willing. Almost all instances of was will-
ing co-occur with a to-infinitive trigger (was willing to), and are used to
make statements about the extent to which characters are willing to do
something:

(145) She intended to stay there, and was willing to retrench in other
ways so that she could continue to live in Thrush Green among
her friends (BRC, ASE 1723)
(146) Cecilia was willing to do anything to make up to Tina for the
deprivations of her childhood, though what these deprivations
were she hardly knew. (BRC, EDN 1336)
(147) He was willing to be persuaded. (BRC, GV2 288)
(148) Even Rachel was willing to share the subterranean joke. (BRC,
A6J 351)

A small subset of occurrences can be characterized as a grammati-

cal metaphor of (epistemic) modality, expressing the subject’s certainty

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

120 Emotion Talk Across Corpora

that a proposition is true. The lexical item is: (X was) willing to bet
(that) p:

(149) She was willing to bet Hugh Puddephat had never suffered
similar doubts. (BRC, HTR 2372)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

These statements are not predominantly about characters’ wishes,
rather they are about their certainty that something is the case.
Apart from these emotion adjectives, which collocate with was at L1,
other emotion terms are also strongly associated with the description of
character emotion, for instance the verbs surprise, hate and admire, and
the noun hate.
Firstly, analyzing the R1 collocates of surprise we find me and
him, followed by her. Surprise is thus clearly used to describe charac-
ters’ surprise with emoters as objects of the verb. Frequent patterns
here are:

• X (had ) surprised him/her

• It/that/X did not surprise him/her
• that surprises me
• It/which/this didn’t/doesn’t surprise me (that)
• It wouldn’t/would not (adv) surprise me (if)
• Surprise me

Again, this occurs in narrative description or dialogue, for instance:

(150) Yet one aspect of the speech had surprised him (BRC, EF1 2662)
(151) For some reason this did not surprise him at all. (BRC, ASN 2402)
(152) They were also well bred, and that surprises me. (BRC, H9C
2029)
(153) ‘It wouldn’t surprise me,’ Masha said, ‘if Likud carried out the
massacre.’ (BRC, AE0 471)
(154) ‘. . . Take me somewhere I’ve never been before. Surprise me.’
(BRC, C8S 1788)

Be is also frequent as an L1 collocate of surprise, with similar patterns

as in conversation – You’d be surprised, I’d be surprised, I shouldn’t/wouldn’t
be surprised (if ) – occurring in character dialogue.
Secondly, considering hate, although the is most frequent at R1, to and
it also occur often. At L1, I is most frequent, but she and he are not as
far off as in conversation. In other words, hate is used for describing

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 121

the emotions of characters, and also occurs in quoted speech. Common

triggers with the are:

• groups of people: the Communists, the queers, the Germans, The English,
the Scots

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

• mental processes: the thought that, the thought of, the idea of
• other: the pills, the city, the work, the things, the authorities, the telephone

The usages of hate with to and it are similar to the ones described above
for conversation and news reportage:

(155) He hated it when she was tactful (BRC, AC3 740)

(156) Maire Carroll, serving in the shop and hating it (BRC, CCM 2505)
(157) ‘I hate it, I hate it, I hate it!’ (BRC, FRS 1550)
(158) I’d hate to end up in a French jail (BRC, GV2 814)
(159) And more than half of them, I hate to say, are mine. (BRC, HGN
2448)
(160) And I hate to see you unhappy (BRC, APU 2226)

Thirdly, moving on to admire, we find to at L1 and the at R1. A common

phrase (an idiom schema in Moon’s 1998: 161 sense) is:
⎞ ⎛
STAND BACK the effect
⎟ ⎜
⎟ to admire ⎜ his handiwork
⎠ ⎝
STEP OUTSIDE the sights

Stand back to admire the effect is most frequent among these variations.
Other common phraseologies are:

• pause/come (out) here to admire + concrete noun

• ‘willingness/ability’ (able to, eagerness to, happy to, wanted us to,
willingness) + admire

Some examples from the corpus:

(161) She buckled a stiff hard belt around Alexandra’s waist and stood
back to admire the effect. (BRC, H8X 542)
(162) He didn’t pause to admire the car (the car seemed like new:
great!) but hurried inside (BRC, FYV 459)
(163) Lydia had to admit that Betty’s eagerness to admire and
approve of people, while annoying, was a good characteristic
(BRC, G0X 1724)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

122 Emotion Talk Across Corpora

Finally, looking at the noun hate, its most frequent R1 collocate is for.
That is, the noun is used to describe an emoter’s hate for someone, that
is, realizes overt, directed affect, for example:10

(164) Then Johnny saw and understood the reason for Luke’s attitude

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

towards Una O’Malley, the reason for his murderous hate for any
man that desired her. (BRC, B1X 2559)

The second function that is very common in fiction is the descrip-

tion of characters’ expressions, actions and behaviour in terms of their
emotion (realized by the pattern prep N), with the nouns surprise and
affection. For instance, surprise occurs most often at L1 with in. As we
know from Section 3.3.2, in surprise either modifies expressors or actions.
For example:

(165) He looked at her in surprise, spoke brusquely. (BRC, CB5 3403)

(166) She shrugged, and then giggled in surprise as the rounded shoul-
ders of the spacesuit ascended like blunt-nosed rockets on each
side of her helmet. (BRC, F9X 2825)
(167) Rostov saw the Adjudicator’s eyes widen in surprise as he recog-
nised the same colouring in the skin around the admiral’s high
cheekbones. (BRC, FSE 56)
(168) The high, arched eyebrows in the long bony face rose slightly in
surprise, but no more. (BRC, AB9 410)
(169) Then the door clanged open and she thought in surprise, We’re
down ! (BRC, FP0 2486)
(170) Li Yuan’s lips parted slightly in surprise. (BRC, GUG 3638)
(171) Cleo’s jaw dropped open in surprise. (BRC, GW2 3033)
(172) ‘We’re the laundry,’ he said in surprise. (BRC, H9N 481)

Considering left-hand collocation with affection, of and with affection

are very common patterns. With of affection, a frequent pattern is N
of affection, most nouns belonging to a number of distinct semantic
subsets:

‘showing’: ‘quantity’
display of affection his lack of affection
show(s) of affection a lot of affection
sign of affection (he felt) a murmuring of affection
token(s) of affection a finite quantity of affection
the hints of affection a scattering of affection

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 123

‘warmth’ ‘force’
the fire of affection surge of affection
warmth of affection welling-up of affection
a rush of affection

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

‘expressor’
his . . . gesture of affection
smile of affection
spark of affection

Concerning with affection we know from Section 3.3.2 that this pattern
is used to characterize actions or expressors. Particularly frequently we
find:

• verbal processes (for example say, speak, greet)

• mental processes (look back, remember, dream of )
• ‘look’ (for example look at someone, watch someone, observing his
pleasure)
• ‘smile’ (smile (at) with affection)

Finally, the pattern A n (common in news reportage, as noted above)

also occurs frequently in fiction. (Table A.12 in Appendix A 4.1 online
gives more information about this pattern in the BRC, allowing further
inter-register comparisons as far as semantic preference and types of pre-
modified nouns are concerned.) The modified noun either indicates a
conscious participant, an expressor or a linguistic/mental phenomenon.
Expressors are particularly characteristic for fiction (and occur rarely in
news reportage/academic discourse where emoters or linguistic/mental
phenomena prevail). For example:

• anxious + expression, eyes, face (5), faces (2), features, expression (3), neck,
tone, voice
• delighted + cackle, cascade [of laughter], countenance, face, kisses, laugh
(2), laughter, squeal, whoop
• enthusiastic + expression, laughter, pummel, tone
• frightened + crying, expression,eyes (4), face (2), faces, leap, look, voice
• surprised + expression, fingers, look, voice
• willing + body, hands

Most often, emotions in fiction seem to show themselves in the eyes,

the face, the expression or the voice, mirroring reality. The collocation

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

124 Emotion Talk Across Corpora

anxious face(s) is particularly frequent in fiction, indicating a metonymic

emoter and undirected, overt affect:

(173) Jim could nonetheless tell by his anxious face that something
was wrong. (BRC, ATE 2513)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(174) He walked slowly into Blue Horizons, longing for deep, knowing
that he had first to face six anxious faces. (BRC, H8A 1451)

Concerning nouns realizing emoters, two further tendencies in fiction

become apparent, relating only to the adjectives frightened and will-
ing. Frightened occurs frequently with ‘young’ or ‘vulnerable’ (including
female) emoters as well as ‘animals’:

• frightened + animal, boy (3), child (3), creature, horse (2), little scribe,
mother, mouse, nervous wretch, rabbit, servant, sheep, woman, young Pole

The collocation with animals is used for comparisons, with the phrase-
ology like a frightened [‘animal’: animal, mouse, horse, rabbit]. Willing
co-occurs with ‘female’ and ‘vulnerable’ emoters:
willing + females, land army girl, slave, subjects, wenches, young secretary

Moving briefly away from the analysis of patterns, it is interesting

to note that fiction is in general characterized by its frequent reference
to behavioural surge expressions (Section 1.4) – as has already become
apparent by looking at patterns involving expressors. Tables A.13–A.15 in
Appendix A 4.2 online give examples for bodily symptoms and behaviour
that are associated with 11 emotion terms in the fiction sub-corpus in
detail. Summing up, we can see that some (bodily) behaviour is shared
among different emotions, while other behaviour is more specifically
associated with one emotion only:

• kissing, hugging: delight, affection

• squeals: delight, surprise
• laughter, smiles: delight, pleasure, affection, surprise
• eye contact (looking/staring): affection, surprise, hate
• pale face: anxiety, fear
• wide eyes: fear, surprise
• screams, shouts: fear, surprise
• freezing: fear, surprise
• biting one’s lips, sighing, shaking head, downwards/turning away
movement: disappointment

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 125

• red eyes, clenched hands, shivering, rubbing fingers through hair:

anxiety
• crying, clinging/squeezing, high pulse: fear
• dark eyes, twisted mouth/face: hate
• blinking/flickering eyes, grunts, gasps, raised eyebrows, open mouth,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

shaking head, sudden movements: surprise

In fact, these descriptions partly reflect what actually happens in

human beings when they experience an emotion – the reactions of their
bodily system, including their blood vessels, somatic muscles, tear ducts,
facial muscles, breathing apparatus, sweat glands, dermal apparatus, and
vocal apparatus (Oatley et al. 2006: 6 and passim; for extensive studies
on facial expression see Ekman 1992, 1999b). For instance, the fact that
laughter and smiles are associated in fiction with a variety of positive
emotions (delight, pleasure, affection) mirrors the fact that all positive
emotions share what Ekman calls ‘the Duchenne smile’ (Ekman 1992:
190), and are not distinguished by their own distinctive signal, maybe
because ‘it has not been relevant to survival to know which positive emo-
tion was occurring, only that it was a positive emotion rather than anger,
fear, disgust, or sadness’ (Ekman 1992: 190).
The fact that at least some bodily behaviour is shared among different
emotions in linguistic descriptions, and that not all such behaviour (for
example smiles and laughter) relates to emotion at all (Oatley et al. 2006s:
84–5, 102, cf. also Martin & White 2005: 90) means that such descrip-
tions should perhaps not be considered (as Martin & Rose 2003: 26 seem
to do) as a direct reference to emotion. Rather, I agree with Dijkstra et al.,
who argue that such references are less explicit than emotion terms, and
involve ‘inferences . . . in order to reconstruct a character’s emotional
reaction’ (Dijkstra et al. 1994: 140). At the same time, it is true that non-
verbal cues play a large role in human attempts to interpret experience
(van Meel 1994: 166) and ‘[t]o give his characters real life the author
has to add a nonverbal dimension, and he does this, paradoxically, by
describing non-verbal behavior in a verbal medium’ (van Meel 1994:
166). Such descriptions then contribute to helping readers visualize and
mentally imagine fictional characters and their behaviour, making them
more lifelike, and either intensifying or contradicting already mentioned
characteristics (van Meel 1994: 171).
Descriptions of symptoms and behaviour associated with emotional
experience and using emotion terms are only two of many ways in
which fiction writers can describe characters’ emotional reactions (com-
pare 5.3.1). That these are considered extremely important in fiction can

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

126 Emotion Talk Across Corpora

also be seen from the fact that in interpreting literature, school chil-
dren are encouraged to ‘[focus] on the reactions of characters to events
and the reasons for their reactions is an important aspect of reading and
interpreting stories’ (Rothery & Stenglin 2000: 223). Such descriptions
of emotional reactions provide characterizations, create imagery, cause

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

empathy or suspense in readers (cf. Dijkstra et al. 1994: 142) and estab-
lish point of view or focalization (Toolan 2001: 59–63). Through the usage
of emotion terms fiction writers may influence us in our like or dislike of
characters, and manipulate the degree of identification that we feel with
them as readers (see Section 6.3 for more detail).

4.4 Academic discourse

Finally, concerning the use of affect in academic discourse, this is asso-

ciated with authorial (plural) overt and covert affect, as well as with a
colligation with modality.11
Authorial overt affect seems important for the adjective surprised
and the verb surprise. Surprised collocates at L1 with be, and is associ-
ated predominantly with the phrase we should not be surprised (at, by,
that):

(175) Furthermore, we should not be surprised by the extent of learn-

ing and imitation that can be found, particularly in complex
animals. (BRC, CMA 859)
(176) Signs in use for many referents reflect one or two of these prop-
erties, and thus we should not be surprised that comparisons of
sign lists result in such a high degree of similarity, as the appear-
ance, movement, and use of an object can be expected to be
similar across different cultures. (BRC, CLH 123)
(177) It is the same – and we should not be surprised at this – in
ordinary discourse. (BRC, J7G 229)
(178) If we consider Braybrook and Powell’s (1980) surprise at finding
22 children with acceptable, natural speech abilities, and apply it
to Conrad’s (1979) national study of school leavers, we find that
we should not be surprised at all. (BRC, CLH 1116)

This is clearly an expression of (directed and overt) authorial affect

on the part of the academic writers, paraphrasable perhaps as ‘it is not
surprising that’ whatever is reported in the research is the case. It is used
to evaluate the research in terms of what researchers already know about

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 127

the object of study, but does not seem to carry any particular positive or
negative evaluation.12
The verb surprise (with only 19 occurrences) has us as a frequent R1
collocate:

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(179) What must have surprised him, as it still surprises us, was the
sheer variety of court life, its different rhythms, its stately cere-
monial in buildings still primitive both by Carolingian and later
standards, and the unique character bestowed on each court by
the personality of its prince. (BRC, EA7 290)
(180) It might surprise us to find such superstitious attitudes in modern
society (though of course, conservatives set their face to the past).
(BRC, CGF 1524)
(181) That such traditions continued in the early 17th century not only
in Italy but also in France should hardly surprise us, but the
possibility that they might have persisted even to the end of the
century and beyond may seem sufficiently improbable not to
merit further investigation. (BRC, J1A 473)
(182) What did surprise us was that the blame was put on our diet and
in particular our greasy fish suppers and general lifestyle. (BRC,
FBH 195)

These are very similar to the occurrences of surprised, with the ‘us’ often
including the reader and commenting on research findings (or historical
facts). Related to this usage are patterns with mental process verbs:

(183) It would not surprise most linguists to find them in women’s

speech (BRC, CGF 182)
(184) It will probably not surprise the reader to learn that . . . (BRC,
ALP 134)
(185) It might surprise us to find such superstitious attitudes in
modern society (BRC, CGF 1524)

Covert authorial affect relates to the usage in academic discourse of

the noun surprise (which has no as most frequent L1 collocate). Academic
texts use the following patterns with no surprise:

• X COME as no surprise
• It BE/COME as no surprise that
• It COME as/BE no surprise to learn/discover/realize that

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

128 Emotion Talk Across Corpora

This realizes covert affect, with the implied emoter seemingly either
the writer or the writer and the reader (in particular through the phrase
it will be/come as no surprise):

(186) That a taxonomist from Mars, armed with a DNA hybridisation

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

machine, would classify men, chimpanzees, and gorillas in the
same family comes as no surprise. (BRC, HWW 1288)
(187) It comes as no surprise that Brazil’s rural credit policy actually
encourages farmers to follow bad farming practices and there is
no countervailing influence to change it. (BRC, APN 569)
(188) Given that Singer and his supporters deliberately cast the mani-
festo for animal liberation in the image of these two charismatic
originals, to benefit from their impetus, it is no surprise that
complaints about second-class citizenry, or unequal rights justi-
fiably voiced on behalf of blacks and women, surface in the other
arena of cats, dogs, apes, and dolphins. (BRC, CM8 807)
(189) It comes as no surprise to discover from another source that
in 1921 the Kursk guberniia party committee had tried to deal
with these problems; ‘the purge of the party was the first basic
step towards the strengthening of the personnel of the party and
yielded enormous results. (BRC, A64 705)
(190) It will, however, be no surprise to learn that the Government
is protected against that risk. (BRC, ASB 1225)
(191) It will come as no surprise to learn, however, that this experi-
ment has never been conducted. (BRC, B16 1259)

These can variously be paraphrased as ‘it is not surprising that’ or

‘it will not surprise you that’. Again, these are often used to evaluate
research results or facts. BE/COME AS no . . . surprise that also occurs,
for instance in:

(192) It therefore comes as no great surprise that these systems are

not easily linked up to talk to one another. (BRC, J0V 1006)
(193) Hence it should be no surprise that ordinary policemen and
women come to feel that the police management and the govern-
ment do not care that the risks associated with routine policing
in a divided society are borne primarily by them (BRC, A5Y 1775)
(194) It may come as no surprise that the Canadian Court turned to
an analysis of the game of ice-hockey to address this issue. (BRC,
HWW 1450)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 129

Thus, statements are frequently made about facts/research findings as

not being surprising. It seems as if evaluations of (non-)surprise carry
little positive/negative evaluation – instead, the evaluation is in terms of
the writer’s or reader’s expectations.
Let’s now move on to the colligation with modality, which becomes

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

apparent by the tendency of five emotion terms to collocate most fre-
quently with be at L1: surprised (discussed above), disappointed, impressed,
pleased and willing.
Modal verbs that occur with be disappointed are: will > may >
would/must/are going to. They are used to make predictions or mitigate
statements of non-authorial affect:

• [third person] will be disappointed

• [third person] may be disappointed

(195) The reader may be disappointed by the standard of what is

written (BRC, A04 933)
(196) Supporters of the Public Service Ideal will be disappointed by
our finding that television news is biased towards the incumbent
government (BRC, A62 1082)
(197) The class between the virtuosi will however not be disappointed
in finding good instruments to fit every taste and every pocket.
(BRC, GWM 259)
(198) anyone who expects this book to be a treatise on the history
of landscape painting will be disappointed; in spite of copious
rewriting, lectures these pages remain. (BRC, A04 648)

As already noted above, a common use is to provide evaluations of

triggers in terms of the emotional reaction they produce (195, 197, 198).
With impressed, the attribution of affect also often seems hedged in
some way, though refusing to and it is hard not to be also occur (pointing
to the unwillingness of a potential emoter to be impressed by something):

(199) Those who pursue comparisons of the kind I am referring to

are likely to be impressed by the staying-power of a literary
preoccupation (BRC, A05 945)
(200) Young intellectuals eager for knowledge, progress and a better
future, were bound to be impressed by new solutions to the
problems facing contemporary China. (BRC, CG0 269)
(201) Conversely, we would be impressed by another yogi who could
alter these processes in his intestines on a word of command
(BRC, CM2 881)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

130 Emotion Talk Across Corpora

Common patterns with be pleased are:

• Emoter must be pleased

• Emoter will be pleased
• Emoter would be pleased

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(202) Supporters of the Libertarian Ideal must be pleased by the degree
of public satisfaction with the press despite public perceptions of
press bias. (BRC, A62 1071)
(203) Nevertheless, the basic guidelines are sound and pupils with
visual handicap will for the most part be pleased to co-operate in
working out their own best visual environment. (BRC, CJG 1076)
(204) Of those interviewed 52 per cent reported that they would be
pleased to enter a general ward (BRC, ECE 1762)
(205) Any reader worried by the prospect of such formality will be
pleased to see how [. . .] the more informal notation to which he
is more accustomed is soon restored. (BRC, EV9 1183)
(206) The author would certainly be pleased to hear from any such
great grandfather! (BRC, EV9 1213)

Finally, of 22 occurrences of be willing, emoter (not author) may be

willing to occurs most frequently (8 occurrences), followed by emoter
(not author) would be willing to (6 occurrences), and emoter (not author)
might be willing to (3 occurrences). Here are some examples:

(207) A man may be willing to believe another who claims that a

particular woman has a liking for sado-masochism or has other
sexual preferences. (BRC, GW1 1069)
(208) Such an approach also assumes that management would be will-
ing to pay for a large project, accept the risks involved and get
little reward in terms of information systems for a number of
years. (BRC, HRK 536)
(209) president Kennedy [sic], interviewed by Jean Daniel of
L’Exproaves [sic], hinted that the United States might be willing
to make some concessions. (BRC, G1R 1090)

The remaining instances are mostly co-occurrences with deontic

modals (X must/should/needs to be willing to).
Summing up, a common function of emotion terms in academic
discourse is to evaluate research findings or provide other kinds of

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 131

evaluations. Non-authorial affect is commonly hedged, hypothesized

or predicted, rather than simply stated, presumably to maintain the
‘objectivity’ of scientific discourse.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

4.5 Trigger comparison

Most of the discussion so far has not systematically considered triggers of

emotion in the different registers. Although a comprehensive discussion
of this is beyond the scope of this chapter, some comments will be made
on five emotion terms (enthusiastic, frightened, impressed, pleased and will-
ing) which are strongly associated with directed affect, collocating in at
least three registers with about, of, by and to respectively.13
Triggers of enthusiastic are very varied, but predominantly appear to be
non-human things. For example, in the BRC people are enthusiastic:

(210) about AIX (its version of Unix) and OSI (N: A8U 172)
(211) about extensification (N: AAG 52)
(212) about the project (N: AHG 1194)
(213) about At Play in the Fields of the Lord (N: AJV 105)
(214) about a new kind of ‘pop’ music (F: CHG 119)
(215) about such lovely slippers (F: CKD 1025)
(216) about successful performances (F: H7H 157)
(217) about her job as interior designer (F: C8T 2382)
(218) about dealing with attempted suicide patients (A: B30 1673)
(219) about pressing ahead with further restructuring of the economy
(A: CG0 620)
(220) about the press (A: A62 126)
(221) about the quality of television (A: A62 209)

Similarly, with frightened of (common in conversation, news reportage,

and fiction), triggers can be things (for example school, heights, scan-
dal, the law), sentient beings (for example our spaniel, dentists, the press,
Charlotte) or activities/behaviour (losing, asking him, telling the truth).
Moving on to triggers with impressed by, triggers in news reportage are
often behaviour that is associated with people (though things and people
themselves also occur as triggers), for instance:

(222) you winning the order of merit (BRC, A2S 297)

(223) the fertility of his futuristic imagination. (BRC, AHL 101)
(224) the professionalism of the Jockey Club.” (BRC, A2E 431)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

132 Emotion Talk Across Corpora

(225) her portrayal of a fading actress –; even though critics were

lukewarm. (BRC, CH2 10857)
(226) the amount of space that the Independent devoted to the
exhibition (BRC, AK4 995)
(227) Wallace’s comeback (BRC, CH3 4688)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(228) the local music-types (BRC, AA9 603)
(229) the low rates (BRC, AKD 613)
(230) the launch of the organisation (BRC, AA5 665)

In fiction, typical triggers concern people (the Governor-General, him)

and their behaviour or character (his bravery, his skill, his ability, Charlie’s
big con trick, what she did, her honesty, his choice of wife), people’s per-
formance or appearance (the striking resemblance between his Uncle Julian
and his father, my appearance, the high-powered performance, my singing),
things (often associated with people) (the new apartment, the pullover, your
sketches, my long fax, the amount of information, those words, such events)
and states-of-affairs (how happy they always were together, what he’d heard
so far, the turn-out, what he had seen).
Finally, in academic discourse, triggers concern semiotic things (the
argument, this fact, that fact, a reply, this passage, the picture, sacred
images), – sometimes associated with their producers – people and their
behaviour (another yogi, the role of Mrs Thatcher as a dominant figure, politi-
cians’ energy), as well as things and qualities associated with things (the
work, the quality of the work, market prices, the first performance of At the
Hawk’s Well, new solutions).
In contrast to the three written registers where impressed by dominates,
conversation prefers impressed with. Triggers in conversation are most
frequently people and associated behaviour:14

• it (4)
• her (2)
• the warnings
• her immaculate assets
• Haweswell
• that
• Tammy Girl
• the man and his wife
• the Underground
• him
• Mick

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 133

• the engineering on that

• his ideas
• his how he didn’t disregard one you know

Pleased collocates with to in the three written registers but with with

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

in conversation. In news reportage, the following patterns are common:

• pleased to [semantic preference: ‘gain knowledge’]: know, learn, hear,

read, come across, be proved wrong
• pleased to see
• pleased to have

In fiction, common collocations are:

• pleased to find
• pleased to hear it/that
• pleased to meet you
• pleased to say (that)
• pleased to see so/sth

In terms of semantic preference, the following subsets can be found in

fiction:

pleased to [‘gain knowledge’]: find, know, hear, discover, learn, observe,

rediscover, read
pleased to [‘linguistic’] agree, announce, call, tell, say
pleased to [‘move’] abandon the country, be back, be as far away, be back,
come, enter
pleased to [‘help’] help, serve
pleased to [‘meet’]: see, meet, attend
pleased to [‘get rid of’]: getting rid off, get off their hands, have got this off
her chest

Finally, in academic discourse the most frequent semantic preference

is for ‘gain knowledge’ (find, explore, hear from, see how), though other
lexical items also occur (for example tell you that, call, co-operate, help you,
enter, print). In other words, the most varied triggers are clearly found in
fiction, with the semantic preference for ‘gain knowledge’ cutting across
the three registers.
Conversation differs from these three registers in that the emotion
is usually directed at entities: in other words, people are pleased with

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

134 Emotion Talk Across Corpora

something, rather than pleased to do something. More specifically,

pleased with often occurs with anaphoric reference items:

• it (11)
• that (5)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

• them (4)
• you (3)
• him (2)
• me (2)
• this

Other triggers are things (the house, her bathroom, t-shirt, my Citroen,
my tickets, that clock, that watch, that cupboard, the parcel, the com-
pact disc, my body, something), people (myself, herself, Sister), people’s
behaviour/actions (the amount of work I’d done, what I’m doing, what Ger-
ald’s done, the way the tiling’s been done), and results (the results, the result,
the result).
Finally, let’s look at willing. In all four registers to is the most fre-
quent R1 collocate, with a to-infinitive clause realizing the trigger of
the emoter’s volition. Some of the triggers can be grouped into different
semantic subsets such as ‘correct’, ‘give, help’, ‘try, risk’, ‘linguistic’, ‘cog-
nition/perception’, ‘pay’, ‘responsibility, participation’, ‘give up’, ‘go,
move’, and ‘admit, accept, agree’.15 Others are less easily classifiable.
Most of these semantic subsets occur across all four corpora, but some
semantic subsets seem more important to one register than to another.
For example, ‘go, move’ seems most important in conversation; ‘linguis-
tic’ and ‘give up’ are frequent in news discourse, ‘give, help’ and ‘give up’
are important to fiction, and ‘correct’ only occurs with some frequency
in academic discourse (which also likes ‘try, risk’, ‘give, help’, ‘linguis-
tic’, ‘cognition/perception’, ‘pay’ and ‘responsibility/perception’). Some
examples from the BRC are:

(231) There’s quite a lot of second hand if he’s willing to travel around
for them. (C: KD5 1962) [‘go, move’]
(232) . . . factors that may have influenced the findings, such as the
declining proportion of people willing to answer interviewers’
questions. (N: AKL 52) [‘linguistic’]
(233) . . . as evidence that the international community is not willing
to make the kind of sacrifices needed to confront the social and
economic roots of Latin America’s drug problem. (N: A2M 366)
[‘give up’]

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 135

(234) Six nappies, even wet ones, did not weigh that much, and there
were others who were willing to help. (F: A6J 212) [‘give, help’]
(235) But would Angharad Morgan be willing to give up her own
independence? (F: CKD 2084) [‘give up’]
(236) . . . this is a widely recognized deficiency which no research

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

funding body seems willing to rectify. (A: B1G 1258) [‘correct’]

These trigger differences are clearly associated with inter-register vari-

ation in terms of content: in conversation, people talk about different
things than do news writers (who are often concerned with reporting
news actors’ utterances) and academic writers (who often discuss mat-
ters related to truth). More differences between the corpora in terms of
semantic preference relating to triggers can be read from Table A.16 in
Appendix A 4.3 online. But there is also significant overlap between all
four registers, suggesting a strong semantic preference of willing to for
certain semantic subsets (often associated with ‘giving something up’,
either literally, linguistically or metaphorically).
In terms of triggers, while some differences may exist between regis-
ters, most triggers across corpora seem to be either sentient beings, their
behaviour, things, states-of-affairs or activities. With respect to mental
processes, Halliday & Matthiessen have pointed out that while Sensers
[∼emoters] are ‘highly constrained’ (Halliday & Matthiessen 2004: 203),
Phenomena [∼triggers] are not: ‘That is to say, the set of things that can
take on this role in the clause is not only not restricted to any particu-
lar semantic or grammatical category, it is actually wider than the set of
possible participants in a ‘material’ clause’ (Halliday & Matthiessen 2004:
203). It should be noted, though, that in some cases specific (semantic
subsets of) triggers are associated with particular emotion terms, which
means that affect triggers can be partly constrained and do not totally
freely combine with emotion terms. This is a statement about typicality,
rather than possibility: at least some emotion terms have a certain pref-
erence for occurring with specific sets of triggers (for example pleased to,
willing to), which does not necessarily mean that they cannot occur with
other triggers.

4.6 Summary

As I have shown in this chapter, each register has a certain num-

ber of patterns and usages that are particularly important (with some

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

136 Emotion Talk Across Corpora

functions occurring across certain sub-corpora). Let’s sum these up very

briefly:

Conversation
• expressing authorial overt affect (often in the present tense): for

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

example, I’m (not) surprised, x surprises me, I hate, I was a bit disappointed
• up-scaling affect (with very): for example, impressed, pleased
• expressing authorial/non-authorial past affect: for example, x was
delighted, x was frightened

News
• first person affect (authorial affect that is quoted): for example, I hate,
I admire, I was surprised
• third person affect (non-authorial affect, either in indirect speech
or as description of news actor): for example, surprise the
world/country/nation, he hates, x was delighted, x is anxious, x are
frightened
• covert affect (N n pattern): for example, surprise attack, hate trap
• up-scaling affect (with so, very): for example, disappointed, impressed,
pleased
• hedging, predicting, hypothesizing affect (colligation with modality):
for example, I would (not) be surprised if, x will be disappointed, x would
be willing to
• undirected overt affect (A n pattern): for example, delighted Andrew, an
anxious day, an enthusiastic American superstar, an enthusiastic response

Fiction
• describing character emotion in narrative and dialogue (in past): for
example, x was surprised, delighted, anxious, disappointed, frightened,
impressed, pleased, willing
• expressor/action plus affect (prep N pattern): for example, giggled in
surprise
• undirected overt affect (A n pattern): for example, anxious faces, willing
young secretary

Academic discourse
• overt and covert authorial affect (with ‘plural’ emoter): for example,
we should not be surprised, it might surprise us, x COME as no surprise

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 137

• hedging, predicting, hypothesizing affect (colligation with modality):

for example, x will/may be disappointed, x must/will/would be pleased, x
may/would be willing to

These functions are (among others) related to the expression of eval-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

uation and emotional sharing in conversation, the creation of imagery
and characterization in fiction, news values in news reportage, and the
evaluation of research findings in academic discourse. More specific com-
ments on the use of emotion terms within individual texts will be made
in Chapter 6. However, as a basis of this analysis we need to modify the
appraisal system slightly, which I will attempt to do in Chapter 5.

Notes
1. This analysis was done manually (but see Table A.17 in Appendix A 4.4
online for the results of an automatic collocation analysis) because this is
more accurate (the manual analysis can identify differences between to as
a preposition and to as infinitive marker as well as wrong POS tagging),
and more comprehensive (it can consider semantic preference), and because
I am interested more in the differences between corpora than in statis-
tical significance as such. In the manual analysis of the 15 emotion terms,
wrong POS classifications were as far as possible excluded from the analy-
sis. Collocations and colligations were only noted if they occurred with a
minimum frequency of three (two if the number of overall occurrences in
the given sub-corpus was particularly small). Punctuation was excluded from
the analysis of collocation, meaning also that collocations across punctu-
ation (especially sentence punctuation rather than clause punctuation) were
disregarded. All collocates were identified at L1 and R1 position; only for
the analysis of some patterns could L2, L3 and R2 and R3 be considered.
The basis of the analysis was the lemma rather than the word form (that
is, I looked for collocates for surprise as a lemma rather than analyzing sur-
prises, surprising, surprised, surprise individually). The maximum number of
concordance lines analyzed was 250 – if an emotion term occurred more
than 250 times in one of the corpora, 250 concordance lines were randomly
selected.
2. The lexico-grammatical patterns of the emotion terms hate (N), admire (V),
anxious (A), enthusiastic (A) and affection (N) in conversation will not be dis-
cussed in this chapter, because the occurrences are not frequent enough to
allow conclusions as to representative tendencies.
3. Statements about things being a surprise (to or for someone) are also common,
for example that’s a surprise, was it a surprise to him?, it’s gonna be a surprise.
People also occasionally talk about someone getting a surprise. It follows that
we find for as a common collocate at R1, realizing the implied emoter of the
covert affect pattern (something is a surprise for someone), for example, it’s
gonna be a surprise for her.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

138 Emotion Talk Across Corpora

4. Note that admire has two meanings, which both relate to emotion:

‘to respect sb for what they are or for what they have done’
‘to look at sth and think that it is attractive and/or impressive’

5. Was surprised frequently occurs in directed affect patterns, with the trigger

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

realized by a prepositional phrase with at (2 occurrences) or by (2), a to-
infinitive clause (2), a that-clause (2) or a when-clause (3). Three occurrences
realize undirected affect, with the trigger to be retrieved from the context:

(i) Still, Bruno has charm and stage presence. I was surprised, although
not as surprised as Mike Tyson when Frank got up and hit him back.
(BRC, AAH 24)
(ii) C apps: 1st time The United Arab Emirates qualified against expecta-
tions; even their Brazilian coach, Mario Zagalo, was surprised. (BRC,
A9H 179)
(iii) The cameraman who took the pictures at a French villa near St Tropez
told the Mirror: ‘I definitely heard the child call out the word ‘Daddy’
to that bald man. ‘You could quite clearly hear the noise from the pool
at times. Eugenie uttered the word while she was playing with Bryan.
‘Of course I was surprised. I just wondered what the Queen would say,
let alone the duchess’s husband.’ (BRC, CH6 8797–8800)

6. At L1 the most frequent collocate is not, negating the experienced surprise.

Excluding it is not surprising but including n’t (rather than just not), the
(quoted and not-quoted) occurrences are usually with people rather than
metonymic emoters:

(iv) And that didn’t surprise Graham one bit as he bluntly admitted . . .
(BRC, CH3 3821)
(v) The reason for their non-availability will not surprise anyone who has
dealings with an insurance company (BRC, A3J 211)
(vi) Diana’s gesture did not surprise her, merely bemused her. (BRC, CH1
9675)
(vii) ‘Although I have not seen the report, it does not surprise me that the
Rover Group was sold for a song. (BRC, A7W 552)
(viii) ‘. . . It would not surprise us to see something like this over the next
few months.” (BRC, A4F 524)

These do not relate to the news value of unexpectedness, but seem to

be used either to evaluate things positively/negatively or describe people’s
expectations.
7. In general, anxious can indicate either worry (that is, anxiety) or volition, with
an R1 collocation with to indicating ‘volition’ (examples from OALD):

• He seemed anxious about the meeting (‘worry’)

• There were a few anxious moments in the baseball game. (‘worry’)
• He was anxious not to be misunderstood. (‘volition’)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 139

8. Surprise terms can actually be used as what SFL has called a grammatical
metaphor of modality (Halliday & Matthiessen 2004: 613–17). Simplifying the
matter greatly, this refers to mental clauses that realize (for example epistemic)
modality. I believe that affective comments such as I wouldn’t be surprised can
similarly be regarded as relating to probability. Compare:

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(ix) NINA [1352] I’ll bet they hear of these mini breaks through
the wh I can’t remember
[. . .]
NINA [1354] her name.
[. . .]
CLARENCE [1359] I don’t know what her name is but . . . I know
the one you mean.
[1360] The travel agency girl?
NINA [1361] Mm.
CLARENCE [1362] Young woman, yeah.
NINA [1363] I wouldn’t be surprised.
CLARENCE [1364] Probably.
(BRC, KBP)

Here, Nina’s utterance that she wouldn’t be surprised (if it turns out that
they heard of the mini breaks through the travel agency girl) seems to be
paraphrasable as ‘it is possible/likely that’, and functions like an epistemic
modality marker. This also becomes apparent by Clarence’s answer, prob-
ably, similarly expressing epistemic modality, and mirroring Nina’s modal
assessment.
9. Remember that willing has two meanings, which both relate to emotion:

(a) [not usually before noun] ‘willing (to do sth) not objecting to doing sth;
having no reason for not doing sth’: They keep a list of people (who are)
willing to work nights.
(b) [usually before noun] ‘ready or pleased to help and not needing to be
persuaded; done or given in an enthusiastic way’: willing helpers/volunteers,
willing support, She’s very willing.
(OALD)

A collocation at R1 with to indicates meaning (a).

10. With hate the pattern prep N also occurs: of is the most frequent L1 collocate
of hate, occurring in the patterns full of hate, tide of hate, rictus of hate, that is,
in overt (directed or undirected hate):

(x) And you sit there like a dog chained to a stake fixed in the past, mouthin’
your worn-out shibboleths, still full of hate for England and your
Church, too arrogant to make your peace with God. (BRC, B1X 1376)
[directed]
(xi) The evidence that it was an accident brought about by bullies was over-
whelming, especially at a time when 9 out of 10 Britons were engulfed
in a tide of hate for Hitler’s British Mosley stooges. (BRC, CHG 309)
[directed]

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 140 Emotion Talk Across Corpora

(xii) Zulei hopped backwards, trying to regain her balance, her mouth a
rictus of hate. (BRC, FP0 406) [undirected]

In many other cases, hate is used in a variety of patterns to realize undi-

rected affect that is attributed to emoters or the relationship between them,
for instance:

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(xiii) A strange relationship, it was built on love and hate, uniquely tender
on one side, and unspeakably cruel on the other. (BRC, FPK 922)
(xiv) The girls, looking one to the other, darting hate, hardly three foot
tall, arose and walked down to the front of the class. (BRC, BNC 386)
(xv) She had begun to need her rage and her hate, even of late her fierce
cruel fantasies. (BRC, APM 941)
(xvi) All the hate welled up inside him as he thought of Carrie and Mikey.
(BRC, EF1 2234)
(xvii) Ellie saw the mockery in her father’s eyes, as she saw the hate. (BRC,
EEW 1149)
(xviii) He was high on the flooding wave from the bursting dam of his dark
nature, all the anxiety, guilt, hate and repressions sweeping out in a
reckless torrent, no longer containable. (BRC, ADY 1202)

On the one hand, hate thus shows itself in emoters’ eyes; on the other hand,
hate is construed as an uncontrollable force: hate welled up inside him, hate
sweeping out in a reckless torrent. Other usages include the pattern: expressor
with affect (for example his eyes blazing with hate), and more rarely, covert
affect (N n pattern: hate mail, N: their common hate was Kenny) and unemoted
affect (study of revenge, immortal hate). Adjectives preceding hate (often nega-
tive or referring to intensity) are common, too (absolute, common, dedicated,
immortal, murdering, murderous, pure, similar, sudden, undiluted) as well as facial
expressions (looking, darting, a look of, eyes, face(s), mouth, staring). Since
the noun hate is too infrequent in the other corpora, no comparisons can
be made.
11. The lexico-grammatical patterns of the emotion terms hate (V), hate (N),
delighted and frightened in academic discourse will not be discussed in this
chapter, because the occurrences are not frequent enough to allow conclu-
sions regarding representative tendencies.
12. Considering R1 collocation, that and by are equally frequent as triggers (but
to also occurs relatively frequently). Again, some refer to researchers’ (non)
surprise at research results, but others realize non-authorial affect.
13. With respect to this comparison it must be pointed out that negation was
not systematically considered; that is, triggers are listed whether or not the
pattern is negated AT trigger or AT trigger. For example, triggers that are listed
for impressed, include triggers that occur with not impressed.
14. It does seem as if impressed with might be associated more with people as
triggers than impressed by. A search of the whole BNC of the exact phrases
impressed with and impressed by (looking at 250 random hits), showed that
impressed with occurs 40 times with people and impressed by only 21 times.
(names, pronouns, and NPs like the young man but not the students’ enthusiasm

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Patterns of Affect Across Corpora 141

were counted, the latter referring to a quality associated with people rather
than people themselves) Impressed by is also more frequent overall in the BNC
than impressed with (757 vs. 383 occurrences).
15. This sub-categorization is subjective to a certain extent, and some of these
groups can be seen as a subset of others (for example ‘admit, agree, accept’ and
‘pay’ as subsets of ‘give up’ or ‘admit, agree, accept’ as a subset of ‘linguistic’).

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

5
Mapping and Analyzing Affect

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

After looking at emotion terms and patterns in a large corpus (the BRC),
the focus of Chapters 5 and 6 will be on a modification of appraisal
theory and its application in a more detailed analysis of a much smaller
corpus (a subset of the BRC). The following sections of Chapter 5 provide
an overview of previous, related research, the data, and methodology,
while Chapter 6 outlines the results of the analysis.

5.1 Appraisal research

There is a large amount of linguistic research on various aspects of the lan-

guage of conversation, news reportage, fiction and academic discourse,
but it is difficult to do justice to all. Instead, I shall comment mainly
on appraisal analyses. Because of a somewhat critical attitude towards
large-scale corpus analysis (Martin 2004b: 342), most appraisal studies
analyze either individual texts or only small-scale corpora (White 1998,
Coffin 2006), with many case studies (for example Humphrey 2006). As
Kaltenbacher writes, ‘SFL is traditionally qualitative, looking at individ-
ual pieces of discourse of relatively small size’ (Kaltenbacher 2007: 90).
In this respect, appraisal has been applied to the close analysis of many
different kinds of text, such as a national inquiry report (Martin 2004a),
a political speech, narrative, exposition, legal act (Martin 2002), stories,
songs (Martin 2004c), an Act of parliament, a music review (Martin &
Rose 2003), or a movie (Martin 1995, 2000a), to name but a few. Such
analyses have emphasized the importance of references to affect in con-
struing personae, negotiating sociality and solidarity (Martin 2004b) and
aligning readers/hearers ‘into a community of shared value and belief’
(Martin & White 2005: 95).

142

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 143

Conversation
Although there is some non-SFL research on the use of references to
mental (including affective) processes, and stance in conversation (for
example Chafe 1982, Shimanoff 1985, Anderson & Leaper 1998, Precht
2000, 2003), there is hardly any research within appraisal on con-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

versation. Exceptions are Eggins & Slade (1997), Martin (2000b), and
Adendorff & de Klerk (2005, on Xhosa English), but their data are limited,
and contain hardly any affect; further, the focus is on appraisal resources
in general.
Rosenberg notes that in emotion research there is ‘too little exem-
plification of how actual conversations proceed’ (Rosenberg 1990: 170,
similar observations are made by Fiehler 2002: 80), and Galasińksi
points out that there ‘are relatively few studies of discourse strategies
employed by people accounting for, explaining, or simply telling stor-
ies about their emotions’ (Galasińksi 2004: 2–3). Similarly, Anderson &
Leaper mention that there has been little research examining speaker’s
references to emotion in talk (Anderson & Leaper 1998: 419). Some
(non-appraisal, non-SFL) analyses that do investigate emotion references
focus on medical topics or types of discourse (for example Wowk 1989,
Edwards 1999, contributions to Fussell 2002, Pittam & Gallois 2002)
rather than ordinary conversation (on functions of emotion terms in
German conversation, see Fiehler 1990:136).

News reportage
A general overview of linguistic research on news discourse is given in
Bednarek (2006a); particularly relevant to the analysis of affect in news
discourse are Ungerer (1997), and Edwards (1999).
Within appraisal theory, White (for example 1997, 1998, 2004c,
2006) analyzes a number of aspects of appraisal in news reportage, but
does not focus in particular on affect. For instance, his research shows
that interpersonal meanings occur in the headline/lead (which is often
emotionally charged) and in following phases when ‘elements of the
Headline/lead nucleus are appraised, typically by some expert external
source, in terms of their significance, their emotional impact, or by
reference to some system of value judgement’ (White 1997: 115). Bell
(1991: 161–74) similarly analyzes textual structure, and notes that evalu-
ation can be involved in the news story’s Commentary part, providing
‘the journalist’s or news actors’ observations on the action’ (Bell 1991:
170). See also Iedema et al. (1994) on the construal of objectivity and
subjectivity in media texts, with a focus on judgement, and Körner &

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

144 Emotion Talk Across Corpora

Treloar (2003) on an appraisal analysis of four news reports with respect

to news actors’ speaking positions. Lassen (2006) focuses on appraisal in
bio-technical press releases and Caple (forthcoming) is interested in the
multi-modal/inter-semiotic construal of evaluative stance in a certain
type of news story.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Fiction
Though there is much stylistic research on point of view/focalization
in literature (which includes affect as one of several resources), such as
Simpson (1993), Toolan (2001), and Rimmon-Kenan (2002), appraisal
analyses remain case studies. Thus, Martin & White (2005) provide an
examination of extracts of Annie Proulx’s Shipping News, and Dorothy
Sayers’s Strong Poison, and Macken-Horarik (2003a) focuses on one writ-
ten narrative ( Judith Stamper, Click). Rothery & Stenglin (2000) deal with
interpretations of literary texts. On various aspects of emotion references
in literature, outside appraisal theory, see van Meel (1994), Dijkstra et al.
(1995), Watson (1999), and Oatley (1994, 2003).

Academic discourse
Outside SFL, there is a large body of research on evaluation (including
affect) in written (and more rarely, spoken) academic language, such
as Hyland (1999, 2000), Hunston (1994), Bondi & Mauranen (2003),
Sandlund (2004) – an overview is given by Hood (2006: 37–8). Within
appraisal theory, Coffin (2006; which was not available at the time of
writing) provides a study of the language of the Australian secondary
school history classroom, with an interest in the language of time, cause
and evaluation, and Hood (2005, 2006) analyzes appraisal in research
articles.

5.2 The corpus

The corpus that was used for the manual text analysis is a subset of
the BRC, consisting of 85,121 words (based on a Microsoft Word count;
includes the names of speakers in the conversation sub-corpus). In ana-
logy to the ‘BNC baby’ (www.natcorp.ox.ac.uk), I shall refer to this as
the ‘BRC baby.’ The BRC baby consists of about 20,000 words from each
register:

Conversation News reportage Fiction Academic discourse

22,613 18,164 20,563 23,781

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 145

For the conversation sub-corpus two random samples of 2–3000 words

were selected from each of the social categories AB, C1 and C2, and
one sample of about 6000 words from DE, the samples consisting of a
whole recorded conversation in each case (files KBC, KBD, KE4, KD0,
KE2, KD8, KBP). The news reportage sub-corpus contains samples of

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

about 3000 words from the (broadsheet) categories of social news, arts,
sports, science, political news and commerce (from files AL5, AJG, A7S,
A1N, A1M, A1G, A1E), and includes 30 news articles in total. For the
fiction sub-corpus the first 2000 words or so were randomly sampled
from ten files: AB9, AC2, BMW, C8T, CB5, CFY, FAJ, GOS, H9C and
HR9 (these files are themselves samples from the beginning, middle
or end of a novel). Finally, in the academic discourse sub-corpus we
find two samples of about 2–3000 words each from files from the cate-
gories of natural science, politics/law/education, technical engineering,
humanities and social science, and one sample of the same size from
medicine (files A6 U, ACJ, ALP, AS6, EA7, EWW, FC1, FEF, FPG, HWV).
More detailed information (including titles of books, type of sample,
circulation size, level of difficulty, author gender, speaker age and occu-
pation and so on) is listed in online Appendix A 6.1. Even though
the sampling was relatively random, only those samples were chosen
which contained the desired number of words (2–3000 words), which
made up a relatively unified whole (for example a whole conversa-
tion), and which were also contained in the publicly available BNC baby
corpus.

5.3 The analysis: theoretical and methodological

considerations

In their recent book The Language of Evaluation, Martin and White note
that the qualitative approach that they have taken up there should be
complemented by ‘a quantitative approach, which would focus on fewer
variables across a corpus of texts’ (Martin & White 2005: 260). This,
they hope would ‘encourage a reconsideration of evaluative meaning’.
They also note that ‘our maps of feeling (for affect, judgement and
appreciation) have to be treated at this stage as hypotheses about the
organisation of the relevant meanings – offered as a challenge to those
concerned with developing appropriate reasoning’ (Martin & White
2005: 46). In the following sections I am taking up this challenge, and
discuss some aspects of appraisal theory’s maps of feeling with respect
to findings from corpus linguistics and cognitive linguistics/psychology.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

146 Emotion Talk Across Corpora

The aim is to outline the mapping of affect that was applied in the textual
analyses presented in Chapter 6.

5.3.1 Portraying vs. creating emotion

Let’s start with a discussion of the difference between what I shall call

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

portraying and creating emotion. With respect to the analysis of affect in
text and discourse, it is worthwhile to distinguish between two research
questions:

• How can emoters’ emotions be portrayed (portraying emotion)?

and

• How do texts create an emotional response (in the reader) (creating

emotion)?

The first question relates to how speakers or writers can portray their
own (authorial affect) or others’ (non-authorial affect) emotions, with
a focus on the different ways of doing so. The second question relates
to the creation of an emotional response or atmosphere in the text or
reader. This may be connected to the distinction that is made in SFL in
the description of visual (but not verbal) interpersonal systems, between
affect (character depiction: facial/bodily affect) and ambience (colour
options, creating a certain atmosphere or mood, evoking an emotional
response in viewers) (Martin, Painter & Unsworth 2006). I want to focus
predominantly on the former here (portraying emotion); for research
that deals with the latter (creating emotion) see Ungerer (1997), Macken-
Horarik (2003a), Robinson (2005) or Humphrey (2006).
As noted in Section 1.4, one of the factors according to which affect
is classified in appraisal theory (Martin & White 2005: 47) concerns
whether feelings are realized as a surge of emotion involving para-
or extralinguistic manifestation or whether they are more mentally
experienced as an ongoing emotional state: behavioural surge (the
boy laughed,the captain wept) vs. mental disposition (the boy liked the
present/felt happy,the captain disliked the present/felt sad). In an earlier
approach to affect, Martin and Rose (2003: 26–7) in fact mention three
different ways of relating emotions: (1) writers can either use words
that label emotions (fear), or (2) use words that denote ‘behaviour that
also directly expresses emotion’ (Martin & Rose 2003: 26) (shrieks), or
(3) describe ‘unusual behaviour which we read as an indirect sign of
emotion’ (Martin & Rose 2003: 27) (be very quiet). With the last – what

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 147

Martin calls ‘behaviour which indexes emotion’ (Martin 2002: 203), the
reader needs to invest some cognitive effort in retrieving the emotion:

. . . from this unusual behaviour we know something is wrong but we

can’t be quite so sure about the exact emotion being expressed; we

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

need to use a bit of psychology perhaps. Read in context, however,
we do know what Helena’s on about, because these symptoms are
surrounded by explicit references to emotions which tell us what the
strange behaviour means. (Martin & Rose 2003: 27)

These different ways of portraying emotions can thus be placed on a cline

of implicitness, as in Table 5.1. In this table, the broken line between
behavioural terms and the description of unusual behaviour signals that
the latter is ‘[r]elated to [the former], and sometimes hard to distinguish
from it’ (Martin & Rose 2003: 27).
In order to elaborate on this three-fold distinction and to consider in
more detail and more systematically how emotional responses can be
portrayed linguistically, let’s briefly look to psychology. In Section 1.3 it
was noted that emotions are often defined as involving:

• an eliciting condition/antecedent event

• a cognitive evaluation1

Table 5.1 Cline of implicitness

Describing Examples (from Martin & Rose less implicit

emotions 2003, Martin & White 2005) (less inferencing)

the use of mental fear, worry, pain, ecstatic,
disposition terms wild consuming fear
the use of restless, terrible convulsions, shrieks,
behavioural wept, smiled, tremble, shudder,
surge terms cower, whimper, cry, wail, chuckle,
laugh, rejoice, shake hands, hug,
embrace, restless, twitching,
shaking, start, cry out, faint, fidget,
yawn, tune out, pat on the back,
withdrawn, shake uncontrollably
the use of description very quiet, staring, ice cold, drinking
of unusual (physical) too much, wander from window
behaviour to window, rolls this way, that
side of the bed


more implicit
(more inferencing)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

148 Emotion Talk Across Corpora

• a physiological response
• an action readiness/an action2

In this sense, ‘emotions are complex physiological–affective–cognitive

responses to the physical and sociocultural environment’ (Schrauf &

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Sanchez 2004: 267). If we assume that non-psychologists’ folk know-
ledge of emotions also includes these aspects, that is, that people are
at least dimly and sub-consciously aware of these factors (as proposed
by Kövecses 2000: 130; for a discussion see Radden 1998: 289–92),
their knowledge structures or schemas about particular emotions should
include some knowledge about:3

• the kind of cognitive evaluation associated with an emotion;

• prototypical and potential antecedent events (eliciting conditions)
that cause a specific emotion;
• psycho-physiological expressions that are prototypically and poten-
tially caused by a specific emotion;
• expressions of action readiness;
• prototypical and potential subsequent actions that are caused by a
specific emotion.

People need and use this knowledge to understand/interpret their own

and others’ emotions, to interact adequately with others, and to interpret
the representation of emotional experience (for example in art).
Emotion terms can then be regarded as ‘a shorthand, an abbreviated
way to refer to the various events and processes which comprise the phe-
nomenon of emotion’ (Ekman 1997: 3). In cognitive linguistic terms,
our folk model of emotion is structured in terms of schema knowledge
– involving categories, interrelations, default assignments (prototypes),
and expectations, and words can ‘trigger’ the ‘activation’ of such a
schema (see Bednarek 2005a for an overview of relevant research). These
assumptions are confirmed by a large body of research from cognitive lin-
guistics as well as cognitive and cognitive-social psychology, and by other
researchers on emotion (for example Shaver et al. 1987, Johnson-Laird &
Oatley 1989: 92–3, Russell 1991, White 1990, Gottman et al. 1996: 251,
Kövecses 2000, Mees 2006: 7, Oatley et al. 2006: 184). For instance, it
has been shown that speakers are aware of typical behaviour, situations
and cognitive evaluations of the environment that are associated with
emotions (Ortony et al. 1988: 3, Wowk 1989, Parkinson & Manstead
1993: 300). (Though it is in fact debatable if this knowledge corresponds
to actual emotional experience in real life situations – Parkinson &

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 149

Manstead 1993: passim, Shaver et al. 1987: 1062.) Many words in English
refer explicitly to parts of emotion schemas (antecedent event, evalua-
tive process, physiological state, action, situational circumstances and so
on; see Heelas 1986, Johnson-Laird & Oatley 1989, Bellelli 1995, Fiehler
2002), triggering inferences about the emotional response involved. It

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

seems as if, in English at least, the ways of referring to emoters’ emotions
as set out in Table 5.2 (on page 150) seem possible.4
Whereas mental disposition terms can refer to the whole schema, other
ways of portraying emotion make reference only to parts of the schema
(part–whole relationship). As Kövesces notes, there are many linguistic
expressions in English that describe physiological expressions of emotions,
which can be considered as metonymies. This metonymic relation can be
postulated as a general principle: THE PHYSIOLOGICAL AND EXPRESSIVE RESPONSES
OF AN EMOTION STAND FOR THE EMOTION (Kövecses 2000: 134). We have seen
examples of this in Section 4.3 and there is much research reporting
on this (for example Enfield & Wierzbicka 2002). On the other hand,
behaviour that has to do with emotions (for example hiding, bang-
ing things around, slamming doors, kicking things, hugging) – what
are sometimes called action cues – has not received much attention in
research on emotion (Planalp 1999: 47). These behaviours seem rela-
tively important for emotion talk, however. Bamberg notes that ‘people,
when asked to give emotion accounts (of how they or others once felt),
construe elaborate circumstances around happenings and events, that is,
seek refuge in the “world of actions” ’ (Bamberg 1997b: 25). Finally, causes
of affect become apparent only through the context, and are extremely
varied, as Planalp suggests for conversation:

It is especially difficult to determine the objects or causes of emo-

tion or feelings in conversations because emotions can be about or
caused by practically anything. In conversation, objects of emotion
can be verbal (jokes or even the topic of conversation), nonverbal
(gestures), people (your partner or yourself), thoughts (daydreams),
or even emotions themselves (guilty about enjoying the ethnic joke
or anger about your partner’s jealousy). They can be something as
microscopic as a compliment, an insult, an interruption, or a touch
or, alternatively, something as macroscopic as a stressful interview, an
exciting argument, or a lifetime of frustrating interactions.
(Planalp 1999: 17)

Generally speaking, causes are often unusual/exceptional events

(Fiehler 1990: 233), for example threats of social rejection, loss (for

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
150
Table 5.2 Ways of portraying emotions
Portraying emotions How Examples
(from Bamberg 1997b, Bednarek 2005b,
Martin & White 2005, TV)
referring to emoters’ emotions the use of mental disposition terms fear, sadness, love, hate
fixed figurative expressions My heart sank, He had a broken heart
referring to emoters’ psycho-physiological the use of behavioural surge terms tremble, cry, his voice broke
expressions of emotion
referring to emoters’ actions or behaviour describing mental behaviour more difficult to talk to, more tense,
(caused by emotion) more withdrawn
describing linguistic behaviour
speech act terms rubbish, abuse, revile, caution, scold, castigate,
compliment, complain
emotional talk devices fuck, shit, what the hell
(Chapter 1)
describing (physical) behaviour
actual reward, commit, entrust, comfort, flee, caress,
he seemed to have aged and lost weight
hypothetical I could kill you, I could strangle you, I could kiss
you, ‘you touch my kids and you fight me’
referring to causes of emotions describing elicitors/antecedent events my biggest sister got into a car accident so
she died; I moved to Worcester and I couldn’t
see my neighbors and their dogs; my Mommy
hit me she hit me in the eye

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 151

example of relationships), disapproval, insults, receiving esteem, love

and so on (Planalp 1999: 18). Causes (or situations) are often part of
comparisons, for instance: He felt like a man who has just had a tooth drawn
which has been hurting him a long time (from Tolstoy’s Anna Karenina,
quoted in Wierzbicka 1999: 12).

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

These different types of describing affect clearly interact in texts. As
an example, let’s look at an extract from the news reportage sub-corpus
(see also Section 4.2). Literal and figural emotion terms are underlined,
references to emoter’s behaviour is in bold face, and references to causes
of emotions are in italics:

(1) THE HATE TRAP

Homes can’t be sold
Couples can’t split
Families live in hell

THOUSANDS of couples who hate each other are trapped together

in a living hell because of the slump.
The tormented partners have had to put off their divorce actions as they
can’t sell their homes.
Now they are stuck with each other, locked in violence and misery with
no end in sight, lawyers revealed yesterday.
MEN explode in frustration and batter their wives.
WOMEN strike back with anything that comes to hand – including
knives and rolling pins.
SOME have nervous breakdowns and suffer stress and high blood
pressure.
And CHILDREN are damaged by the constant tensions.
(BRC, CH2 9181–9192, from The Daily Mirror)

This extract illustrates well the prosodic nature of affect noted for
example by Martin & Rose (2003) and Foley & Hood (2006), and that
different types of affect ‘often work together’ (Martin & Rose 2003: 28)
to create textual effects. This particular text starts with references to emo-
tions and causes of emotions in the headlines and lead. It continues with
this portrayal in its description of past events (The tormented partners have
had to put off their divorce actions as they can’t sell their homes). As we move
on to the present (Now) we encounter the first description of behaviour
caused by emotion (alongside references to causes and emotions), which
is then continued in the following three sentences, with the final sen-
tences referring to consequences of emotions (if we interpret tensions

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

152 Emotion Talk Across Corpora

as an emotion term). The patterning is as follows (E = referring to emo-

tions, C = referring to causes of emotions, B = referring to mental or other
behaviour caused by emotions):

E–C–C–E–E–E–C–E–C–C–B–E–E–B–B–B–B–E

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Different ways of portraying emotions are thus interesting in terms
of textual patterning and prosodic structures, and clearly deserve more
attention.
However, in the analyses of Chapter 6 the focus was on mental dispos-
ition terms (emotion labels) only, disregarding other ways of portraying
an emoter’s emotion. Shimanoff (1985: 19) reports an inter-rater agree-
ment of only 20 per cent for more indirect references to affect,5 whereas
the inter-rater agreement for emotion labels is 75.8 per cent. The focus
on labels thus means a more reliable and more valid analysis of affect,
but the disadvantage is that this ‘may not get at the richness of the verbal
expression of affect and emotion’ (Anderson & Leaper 1998: 439).

5.3.2 Analyzing emotion terms in discourse

Limiting our attention to emotion labels, the focus now shifts to their
analysis. Any such analysis will be subjective to a certain extent on
account of the fuzzy nature of emotion lexis (discussed below). However,
the text/discourse analyst must take either/or decisions, putting certain
emotion terms in certain categories, and deciding what to include and
exclude from the analysis. To provide an antidote, this section docu-
ments some important methodological decisions. All such decisions
were consistent across the data; each file was analyzed twice (with a
sufficiently large time interval between the analyses), and checked for
accuracy several times. In problematic cases, paraphrasing, translating,
and two dictionaries (an English–German dictionary and the OALD) were
used, with Martin & White (2005) also being consulted for help with the
coding of emotion terms where applicable. On account of the small cor-
pus size (see Section 5.2), the analyses are to be considered illustrative
rather than representative.
The data were analyzed and coded with the help of Altova XMLSpy
2007, an XML editor software (www.altova.com). This software allows
the user to tag data with a number of attributes, here linguistic variables
such as affect trigger, emoter, negation and so on. Image 5.1 shows an
example. The software also allows you to automatically transform the
results of the analysis into a tabular form, which we will see later. Each

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Image 5.1
XML Spy Analysis

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

153

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

154 Emotion Talk Across Corpora

emotion term was coded on nine linguistic variables, but only five are
relevant for the analyses in this book:6

1. emoter
2. trigger

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

3. affect type
4. covert–overt affect
5. valence

While the analysis of emoter and trigger is relatively straightforward,

some more detailed remarks are necessary for affect type, covert vs. overt
affect, and valence.

5.3.2.1 Classification of affect

As mentioned in Section 1.4, in current appraisal theory affect is classi-
fied according to a number of factors (numbered by me for convenience):

• Factor 1: emotions are grouped into three major sets: in/security

(the boy was anxious/confident) – dis/satisfaction (the boy was fed
up/absorbed) – un/happiness (the boy was sad/happy).
• Factor 2: the feelings are culturally construed as positive or negative:
positive affect (the boy was happy) vs. negative affect (the boy was sad).
• Factor 3: the feelings relate to future states (triggers) or existing ones:
realis (the boy liked the present) vs. irrealis (the boy wanted the present).
Irrealis affect is categorized as dis/inclination (fear/desire).
• Factor 4: the feelings are graded in terms of a cline of intensity: low
(like) – median (love) – high (adore).
• Factor 5: the feelings are construed as directed at/ reacting to some
external agency or as a general mood: reaction to other (the boy liked
the teacher/the teacher pleased the boy) vs. undirected mood (the boy
was happy).
(Martin & White 2005: 46–9).

While most of these factors are relatively self-explanatory, some more

detailed comments on factors 1, 3 and 5 are necessary:
Un/happiness feelings are ‘concerned with “affairs of the heart” –
sadness, hate, happiness and love’ (Martin & White 2005: 49), and can
be sub-divided as indicating happiness or unhappiness. Such feelings can
either be directed at a trigger or not (but compare the discussion of undi-
rected affect in Chapter 3); see Table 5.3 (all examples in Tables 5.3–5.6
are from Martin & White 2005).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
Table 5.3 Un/happiness
un/happiness
unhappiness happiness
misery [mood: ‘in me’] antipathy [directed feeling: ‘at you’] cheer affection
down, sad, miserable dislike, hate, abhor cheerful, buoyant, jubilant be fond of, love, adore

Table 5.4 In/security

in/security
insecurity security
disquiet surprise confidence trust
uneasy, anxious, freaked out startled, jolted, staggered together, confident, assured comfortable with, confident
in/about, trusting

Table 5.5 Dis/satisfaction

dis/satisfaction
dissatisfaction satisfaction
ennui displeasure interest pleasure
flat, stale, jaded cross, bored with, angry, sick of, involved, absorbed, engrossed satisfied, impressed, pleased,
furious, fed up with charmed, chuffed, thrilled

155
10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek
156 Emotion Talk Across Corpora

In/security feelings relate to ‘emotions concerned with ecosocial well-

being – anxiety, fear, confidence and trust’ (Martin & White 2005: 49),
with respect to our environment (including people), as visualized in
Table 5.4.
Dis/satisfaction feelings, finally, involve ‘emotions concerned with

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

telos (the pursuit of goals) – ennui, displeasure, curiosity, respect’
(Martin & White 2005: 49) (see Table 5.5). They have to do with:

achievement and frustration in relation to the activities in which we

are engaged, including our roles as both participants and spectators.
These oppositions take us to the borders of affect as it is popularly per-
ceived, as reflected in Star Trek characters like Spock (a human/Vulcan
hybrid who suppresses emotion) and Data (an android who feels
none) – who occasionally express their fascination with things (Martin
& White 2005: 50, original emphasis).

Additionally, let’s look at factor 3 more closely, namely the suggestion

that feelings can concern future (irrealis) or existing states (realis). This
involves either fear or desire as illustrated in Table 5.6.
Table 5.6 Dis/inclination

dis/inclination
disinclination inclination
fear desire
wary, fearful, terrorized miss, long for, yearn for

Dis/inclination is not listed in the table as part of the classification of

kinds of emotion (there are only three major groups of emotion listed:
un/happiness, in/security and dis/satisfaction) (Martin & White 2005:
49), but seems to be treated on its own terms. Presumably, this is because
it differs from these three sets of emotion in terms of its irrealis trigger
and its directedness (Martin & White 2005: 48). An opposition is set up
between realis and irrealis (for example Martin & Rose 2003: 60–1), that
looks as follows:


realis : un/happiness, in/security, dis/satisfaction
Affect
irrealis : dis/inclination

However, dis/inclination is also listed alongside un/happiness,

in/security and dis/satisfaction when talking about ‘kinds of unhappiness’
(Martin & White 2005: 51), and in example analyses desire is placed in
the same column as cheer, misery, disquiet and so on (Martin & White

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 157

2005: 80–1), so that it seems reasonable to talk about dis/inclination as

a fourth category or sub-type of affect.
As becomes apparent, all four categories of affect (un/happiness,
dis/satisfaction, in/security, dis/inclination) involve both ‘positive’ (hap-
piness, satisfaction, security, desire) and ‘negative’ (unhappiness, dissat-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

isfaction, insecurity, fear) emotions (factor 2). In contrast, Galasiński
argues that it is problematic to:

assign universal value to emotions and label them as positive or neg-

ative. Such an evaluation must be context-bound. I can imagine
contexts (for example, those of abuse) in which the supposedly neg-
ative anger will be positive, as well as ones in which the supposedly
positive love will, in fact, be negative (Galasiński 2004: 46).

However, the difference between positive and negative emotions is

secerned by many researchers of emotion (cf. Jahr 2000: 7, Downes
2000: 104, Turner & Stets 2005), and examples such as that given by
Galasiński can be disregarded in a more general framework: ‘We are not
concerned here with the value that a particular psychological framework
might place on one or another emotion (cf. “It’s probably productive
that you’re feeling sad because it’s a sign that …”)’ (Martin & Rose 2003:
59). In this book (for example Chapter 2), I have repeatedly used this
‘positive–negative’ classification of emotion terms (which we might label
valence as is frequently done in emotion research), while also noting
that a number of emotion terms do not easily fit into this classification
(consequently labelled ‘ambiguous/neutral’; see discussion below).

emoter emotion trigger

X ‘admire’ Y

Attribution of Evaluation of trigger

emotion to emoter in terms of emotion

Figure 5.1 Affect and evaluation

Finally, concerning factor 5 (directed vs. undirected affect), let us
assume that this distinction is theoretically valid, even though it is ques-
tionable how many references to affect really are undirected (Section
3.3.3). With directed affect we can then distinguish at least two elements
(Figure 5.1, above):

• the attribution of some emotion to an emoter through the usage of

an overt or covert emotion term;

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

158 Emotion Talk Across Corpora

• the evaluation of some entity (the trigger) on the part of the emoter
in terms of this emotion (by construing the emoter’s reaction with
respect to this entity).

Importantly, the emoter can either be the self (authorial directed affect)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

or the other (non-authorial directed affect) (Section 1.4). With authorial
affect, the self attributes an emotion to him/herself, and simultaneously
provides an evaluation of some trigger, whereas with non-authorial affect
both an emotion and an evaluation of some trigger are attributed to an
other (Figures 5.2 and 5.3).

I [emoter] admire [emotion] linguists [trigger]

Attribution of emotion Evaluation of trigger

to emoter

Figure 5.2 Authorial affect and evaluation

He [emoter] admires [emotion] linguists [trigger]

Attribution of emotion Attribution of Evaluation

Figure 5.3 Non-authorial affect and evaluation

If we only consider the speaker of such utterances, the speaker of

the utterance I admire linguists clearly expresses a positive evaluation of
the trigger (linguists). This evaluative meaning depends on the cognitive
evaluation associated with the relevant emotion schema (admire), and is
therefore readily understood by hearers. People know that the emotion
of ‘admiration’ involves a positive cognitive evaluation on the part of
the emoter, and assume that this is implied when someone says I admire
linguists.
On the other hand, the speaker of the utterance He admires linguists
only attributes a positive evaluation of the trigger (linguists) to an emoter,
but does not say whether s/he shares it. At the same time, non-authorial
affect can imply a certain kind of evaluation on the part of the speaker.
This may depend on the speaker’s attitude towards the trigger: if the
speaker dislikes linguists, saying that John admires linguists may imply

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 159

negative evaluation of John, but if s/he likes linguists, the same utter-
ance may imply positive evaluation of John. Furthermore, feeling rules
(Section 1.3), which tell us whether emotions are appropriate or inappro-
priate (Fiehler 1990: 85) may influence the evaluative meaning of non-
authorial affect. For example, a mother who hates her children implies nega-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

tive evaluation of the mother since the utterance refers to a socially unac-
cepted emotion (see also discussions in Bedford 1956/57: 294, White
1998: 105, Bamberg 1997b, Planalp 1999: 170–9, White 2004c: 240,
Bednarek 2006a: 54–6). In other words, the evaluative meaning of non-
authorial affect can depend on the reader/hearer’s attitude towards the
trigger or on social rules and expectations concerning emotional experi-
ence. These social expectations are extremely powerful. As the American
writer Lionel Shriver, whom I have already quoted in Chapter 1, puts it:

[…] we are all profoundly normative. We have explicit expectations

of ourselves in specific situations – beyond expectations; they are
requirements. Some of these are small: If we are given a surprise party,
we will be delighted. Others are sizable: If a parent dies, we will be
grief-stricken. But perhaps in tandem with these expectations is the
private fear that we will fail convention in the crunch. That we will
receive the fateful phone call and our mother is dead and we feel noth-
ing. I wonder if this quiet, unutterable little fear is even keener than
the fear of the bad news itself: that we will discover ourselves to be
monstrous.
(Shriver 2006: 92)

The difference in the evaluative meaning of authorial and non-

authorial affect can have results on linguistic construals of affect. Thus,
Bamberg shows that there are differences in how emotions are referred
to in first person accounts versus third person accounts of ‘anger’ in
terms of agentivity, individualization, probability and intention, to shift
blame and save face (Bamberg 1997b: 8–9). Precht’s large-scale corpus
analyses suggest that there is ‘a slight tendency for an expression about
one’s self to be more positive […], and comments about others not phys-
ically present to be negative’ (Precht 2000: 129). And Shaver et al. (1087:
1080–1) note differences in the construal of subjects’ self-reports of
emotional experience and reports of typical emotional experience. Non-
authorial affect is particularly important in conflict situations, preven-
tion, and the regulation of disappointment (Fiehler 1990: 132, who calls
this projektive Erlebensthematisierung – ‘a thematization of projected expe-
rience’). It may well be important for text analyses to be aware of these

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

160 Emotion Talk Across Corpora

distinctions. Occasionally, analysts might want to limit their analysis to

authorial or non-authorial affect, respectively. In the analyses of Chapter
6 both are examined, since the focus of this book is on emotion talk as
such. However, rather than using the current mapping of affect as just
introduced, a modified version will be used which is outlined below.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

5.3.2.2 Modification of affect types
While I do not want to attempt a completely new classification of emo-
tion lexis, nor establish what Martin & White (2005: 51) call a semantic
topology for this region of meaning,7 I want to suggest a modification
of the present description of affect types, relating to the categories of
in/security and dis/inclination.

In/security. If we consider un/happiness and dis/satisfaction, it becomes

apparent that the positive sub-categories correspond exactly to, or ‘mir-
ror’ the negative sub-categories (Table 5.7); but with in/security, this is
much less the case (Table 5.8). This concerns in particular the oppos-
ition of trust and surprise. I therefore suggest setting up the system
of in/security in analogy to the other systems, with the positive and
negative categories ‘mirroring’ each other (Table 5.9).

Table 5.7 Sub-categories of un/happiness and

dis/satisfaction

happiness unhappiness
cheer misery

affection antipathy

satisfaction dissatisfaction
interest ennui

pleasure displeasure

Table 5.8 Sub-categories of in/security

security insecurity
confidence disquiet

trust surprise

Table 5.9 Modified in/security system

security insecurity
quiet disquiet

trust distrust

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 161

Compared to the old system, confidence becomes subsumed under

the more general (technical) term of quiet, having to do with emotional
calm, as it were, as realized for example by lexical items such as com-
forted, reassured, confident, solace. Trust is now opposed to its opposite
emotional response, distrust, rather than surprise, which falls out of the

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

system. In other words, the proposal is that in/security feelings relate
to disquiet (for example uneasy,anxious,afraid,fear) or quiet (for exam-
ple reassured,comforted,at ease,blithely) as well as distrust (for example
reserve,emotional withdrawing, suspicious) or trust (for example confide,
trust, believe in) but do not include surprise. In fact, I propose to set up
surprise as a separate type of affect, and to treat it on its own terms, with
affect at this stage relating to four sets of affect or four different affect
types (Figure 5.4).
un/happiness

dis/satisfaction
affect

type in/security

surprise

Figure 5.4 Four sets of affect type

Apart from this new classification resulting in a more ‘logical’ struc-

tural organization of the in/security sub-category, it is based on the
belief that surprise is not culturally construed as negative – which would
be the implication if surprise were included as the ‘negative’ part of
in/security. The problematic status of surprise in terms of appraisal the-
ory also becomes apparent in that it is excluded in earlier frameworks, for
example Martin (1997: 22). It has also been noted by some researchers
that emotion terms labelling the emotion of surprise (for example aston-
ished, startled, surprised) seem to be relatively neutral rather than clearly
positive or negative (for example Nöth 1992: 73, Robinson 2005: 64).
Supporting evidence for this belief comes from corpus data, which will
now be discussed in more detail.
Firstly, I have investigated what kinds of bodily symptoms and
behaviour are associated with a number of emotion terms in the fic-
tion sub-corpus of the BRC (because this is where such associations are
most frequent; see Section 4.3). Instances that were counted as examples
for such associations include (see Sections 3.3.2 and 4.3):

(2) gave a delighted laugh…( BRC, B1X 276)

(3) He smiled, pleased with…(BRC, GUG 341)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

162 Emotion Talk Across Corpora

(4) a smile of affection (BRC, CEC 624)

(5) He shook his head, bitterly disappointed (BRC, GUG 1725)
(6) he rubbed his fingers through his hair and she could tell he was
anxious (BRC, CKD 600)
(7) the whites [of her eyes] showing like a frightened horse (BRC,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

BMR 822)
(8) the dark eyes, smouldering with hate (BRC, CEC 2498)
(9) Cleo’s jaw dropped open in surprise (BRC, GW2 3033)
(10) he sat frozen in his seat by surprise (BRC, AB9 2196)

Appendix A 4.2 online lists the associated bodily behaviour in detail, but
here is a summary of the most important tendencies:

• Delighted (A) is associated with kissing and hugging, squeals and

whoops, and, most importantly, with laughter and smiles.
• Pleased (A) is associated predominantly with smiling and grinning.
• Affection (N): is associated with eye contact, smiles, kissing and
hugging.
• Disappointed (A) is associated with biting one’s lips, sighing, shaking
one’s head, and a ‘turning away’ movement (turn away, jump to her
feet, turn) or a ‘downwards’ movement (sink back, sat on).
• Anxious (A) is associated with red-eyes, a white face, clenched hands,
shivering and rubbing one’s fingers through one’s hair.
• Frightened (A) is associated with wide eyes, crying, shouts and
screams, a pale face and several types of body movement: fleeing,
freezing, and clinging/squeezing as well as a high pulse.
• Hate (V) is associated with dark, blazing or smouldering eyes, staring,
and a twisted mouth and face.
• Surprise (N, V) and surprised (A) are associated with wide eyes, and
staring and looking, blinking/flickering eyes, smiling and laughing,
screams, shouts, grunts, squeaks and gasps, speechlessness, raised eye-
brows, an open mouth, shaking one’s head, sudden body movements
or freezing.

Table 5.10 Behavioural surges of surprise in fiction

Shared with ‘positive’ emotions Shared with ‘negative’ emotions

squeals
delight screams, shouts
fear
laughter, smiles
delight, pleasure, affection wide eyes
fear
freezing
fear

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 163

These results show that in fiction the emotion of surprise is associated

with characters’ behavioural surges that are also related to both ‘positive’
and ‘negative’ emotions (Table 5.10). In other words, surprise is associ-
ated both with behaviour normally related to ‘negative’ emotions (wide
eyes, screaming, shouting, freezing) and with behaviour normally related

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

to ‘positive’ emotions (laughter, smiles, squeals).
Secondly, we can look at emotions that are conjoined with surprised
(A), surprise (V), and surprise (N) with the help of the conjunction and,
and compare this with emotions that are conjoined with other emotion
terms. Tables A.18-A.20 in Appendix A 5.1 online show the findings for
all emotion term occurrences, looking at the structures emotion term and
as well as and emotion term. The function of such doublets is rhetorical –
similar to ‘triplets such as appalled, perplexed and repulsed’ (Martin 2004b:
342), indicating intensification (graduation). They also allow us to con-
strue events from different emotional perspectives (Bamberg 1997b: 11).
Summing up the results, there is a tendency for ‘positive’ emotion
terms (delighted, enthusiastic, impressed, pleased, affection, admire) to be
conjoined with other ‘positive’ emotion terms, for example relieved,
pleased, encouraged, optimistic, grateful, kindness, love, trust. Vice versa,
‘negative’ emotion terms (anxious, disappointed, frightened, hate (N), hate
(V)) are typically (but not exclusively) conjoined with ‘negative’ emo-
tion terms such as tense, unhappy, fearful, miserable, frightened, ashamed,
worried, cross, confused, despise, fear, anger.
Surprise, on the other hand is conjoined much more equally with both
‘positive’ and ‘negative’ emotion terms (though ‘negative’ ones seem a
bit more frequent), for example:

relief and surprise

embarrassment and surprise
fear and surprise
shock and surprise
surprise and admiration
surprise and pleasure
surprise and delight
pleased and surprised
happy and surprised
sad and surprised
shocked and surprised
touched and surprised
surprised and interested
surprised and relieved

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

164 Emotion Talk Across Corpora

surprised and flattered

surprised and irritated

This seems to support the assumption that surprise is not clearly

construed culturally as a negative (or positive) emotion.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

This is also borne out by some additional corpus findings. For instance,
to surprise often co-occurs with expressions of volition (for example hop-
ing to, wanted to, urge to, it would be nice to) in fiction, which clearly
conceptualize this meaning of surprise as a positive emotion:

(11) I was hoping to surprise you – a sort of unexpected wedding

present, but it was no good. (BRC, FPM 2334)
(12) “Well, I have three children of my own now and I thought it would
be nice to surprise them with the sugar mice on the tree, and also
the chocolate cat.” (BRC, AT7 409)

And as premodifying noun, surprise can occur with ‘negative’, (surprise

attack) ‘neutral’ (surprise move, surprise step, surprise decision) and ‘positive’
(surprise party, surprise gift) lexis.8
This does not mean that surprise cannot be construed by a
speaker/writer in a given context as a positive or negative emotion.
For instance, we can use a lexical item such as shock which seems to
inscribe (Martin & White 2005: 61) mostly negative (rather than neutral
or positive) surprise:

Inscribed construal: I was shocked [negative surprise].

On the other hand, the positivity or negativity of an emotion such as

surprise can be contextually implied with the help of other evaluations
preceding or following the emotion term:

Contextually implied construal:

Prospective/cataphoric:
What a lovely [positive appreciation] surprise [positive surprise]
Retrospective/anaphoric:
I was surprised [positive surprise] and delighted [happiness].

In the case of prospective construal, the positive appreciation (Section

1.4) conveyed via lovely turns the surprise cataphorically, as it were, into
positive surprise. In the case of retrospective construal, the conjoined
positive emotion term delighted anaphorically turns the surprise into

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 165

positive surprise. In other cases, however, the construal of the surprise

may remain unclear, ambiguous or rather neutral.9
Summing up the discussion so far, it appears that while a large num-
ber of emotions are culturally construed as positive or negative, some
emotions (such as surprise) have a more ambiguous status (but may,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

though they need not, be construed in a given context as a positive or
negative emotion). Other candidates for such an ambiguous status are
desire or non-desire (see discussion below) – compare also Section 2.4
and Appendix A 2.4 online for further examples.10

Dis/inclination. The second modification of appraisal theory’s affect

types concerns dis/inclination which I include as indicating a certain
kind of emotion in the same way as un/happiness, in/security and
dis/satisfaction. However, if we compare in/security and dis/inclination
there seems to be some possibility for overlap between disinclination:
fear and insecurity: disquiet, with both relating to emotions of anxiety
or fear (Table 5.11). I thus propose to re-define dis/inclination as sug-
gested in Table 5.12. Dis/inclination is here re-construed not in terms of
a positive (desire) or negative (fear) emotion, but rather with respect to
polarity, referring to desire (volition) and non-desire (non-volition).

Table 5.11 Fear vs. disquiet

inclination disinclination


desire fear
miss, long for, yearn for wary, fearful, terrorized
security insecurity
quiet disquiet

solace, comfort uneasy, anxious, freaked out

Table 5.12 Modified dis/inclination system

dis/inclination
desire non-desire
miss, long for, yearn for, want refuse, reluctant, disinclined

It must be emphasized that non-desire relates to lexicalized (for

example refuse) and morphological (for example disinclined) negation

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

166 Emotion Talk Across Corpora

only. This allows us to keep up the distinction that is made in appraisal

theory between negative emotions and negated positive emotions:

We also find it useful to distinguish negative feelings from positive

feelings that are grammatically negated, thus drawing a distinction

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

between sad and not happy; by notating grammatical negation as ‘neg’,
we can code not happy as ‘neg + happy’, opposed to sad as ‘−hap’. Mor-
phological negation (eg unhappy, insecure) on the other hand is not
arguable, since it is realised lexically …; so we will code it as negative
rather than negated attitude (ie -hap for unhappy, neg + hap for not
happy)
(Martin & White 2005: 73).

This permits us to code miss, willing as ‘inclination’, and not miss, not
willing as ‘neg + inclination’, in contrast to refuse, which is coded as ‘dis-
inclination’, and not refuse, which is coded as ‘neg + disinclination’. This
does not mean, however, that inclination is considered positive and dis-
inclination negative, since the organization of the relevant meanings is
made in terms of polarity. This classification of dis/inclination in terms
of polarity rather than positive/negative cultural construal takes into
account the fact that neither desire nor non-desire seem to be unequivo-
cally construed culturally as positive or negative. And the evaluation
implied by desire and non-desire terms may also depend on the con-
text: if I desire something that you do not think I deserve, or if I desire
something that you think is bad (for example to steal, to lie, to cheat),
your evaluation of me is presumably negative. But if I desire something
that you think I deserve, or that you evaluate positively, your evalua-
tion of me might be more positive. An example for negative desire is
also represented by envy, a term for a specific type of desire which is
culturally construed as negative.
The new set-up of dis/inclination avoids a double classification of
emotion terms such as fear and anxiety, and takes into account the
ambiguous status of desire in terms of cultural construal. It also takes
into consideration the fact that irrealis cuts across all emotions, and
can be realized differently, either grammatically (Martin & White 2005:
48) or lexically (desire, want, fear, afraid of); see Table 5.13. In other
words, all emotions can relate to an irrealis trigger (for example I’d be
happy if. . .) and dis/inclination is not the only type of affect that can
be irrealis, though desire and fear terms usually lexicalize irrealis (desire,
want, fear, afraid of). However, dis/inclination does seem different from
the other types of affect in not allowing a realis trigger (Table 5.14).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 167

We can set this up as an if-then relation (if dis/inclination then irrealis

trigger), shown in Figure 5.5.

Table 5.13 Affect types and irrealis trigger

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

trigger surprise dis/inclination un/happiness dis/satisfaction in/security
irrealis I’d be pleasantly I don’t want him your parents I’d be furious I fear that
surprised if. . . to do this. . . would be if. . . he might die
delighted if. . .
I’d like to. . . I’d hate it if. . . I’m afraid of
her leaving me

Table 5.14 Affect types and realis trigger

trigger surprise dis/inclination un/happiness dis/satisfaction in/security

realis I’m surprised – (?) I’m delighted I’m bored with This frightens
that to meet you. . . this. . . me. I’m scared.

un/happiness

in/security
affect
dis/satisfaction
type
surprise

dis/inclinationI

realis
trigger

irrealisT

Figure 5.5 Affect type and (ir)realis trigger

A fuzzy system of modified affect. To recapitulate, I propose that we clas-

sify emotion terms according to five rather than three categories, and
re-construe the systems of in/security and dis/satisfaction accordingly
(Figure 5.6). In this Figure the broken line signals that this system is
set up as a fuzzy system, with no clear boundaries between the affect
types, and possible blends (such as jealousy as disquiet (in/security) and
affection (un/happiness). It is also assumed (in line with much psycho-
logical research into emotion terms, for example Shaver et al. 1987,

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

168 Emotion Talk Across Corpora

un/happiness (old)

dis/satisfaction (old)
affect
in/security (new)
type

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

dis/inclination (new)

surprise (new)

Figure 5.6 Modified affect system

surprise

in/security

un/happiness
dis/inclination
dis/satisfaction

Figure 5.7 Core and marginal emotions

Russell 1991) that linguistic affect is organized as a prototype cate-

gory, with core, better and worse members, and family resemblances
(Wittgenstein) between the category members. Un/happiness is core –
‘probably the first to come to mind when we think about emotions’
(Martin & White 2005: 49) – whereas surprise and dis/inclination as well
as parts of dis/satisfaction (see above) are non-core members. On the one
hand, this reflects psychologists’ debates about whether both interest
(dis/satisfaction category) and surprise are cognitive rather than emo-
tional states (on interest, see Ekman 1999a: 8, Milton 2005: 33); on the
other hand, I assume that interest terms (for example involved, absorbed,
engrossed) as well as surprise terms are considered as marginal rather than
prototypical examples of emotion terms by speakers. And dis/inclination
arguably refers to volition rather than emotion (volition and emotion
are often differentiated, for example by Halliday & Matthiessen 2004:
208) and it has no realis trigger. Both surprise and dis/inclination are
also culturally not construed as positive/negative. Dis/inclination has
only two sub-types, and surprise has only one, because there is no
corresponding lexicalized emotion of ‘no surprise’ (even though we

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 169

can refer to this with the help of negation). We can visualize this as
Figure 5.7, which captures this fact better than a system network, show-
ing that un/happiness includes the most ‘core’ emotions (followed by
in/security) and that surprise, dis/inclination and dis/satisfaction include
more ‘marginal emotions’. This new classification is not neatly orga-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

nized, and incorporates fuzziness and gaps, but arguably this is offset
by the fact that it allows us to classify instances of emotion terms more
realistically. It de-prioritises elegance in order to account for authentic
linguistic data – in line with many current models of language, including
but not limited to SFL (Gonzálvez-García & Butler 2006). Summing up,
Table 5.15 compares the old and new categorization of affect types in
detail.

Table 5.15 Comparison of affect types

Old New
un/happiness un/happiness
unhappiness happiness unhappiness happiness
misery antipathy cheer affection misery antipathy cheer affection
in/security in/security
insecurity security insecurity security
disquiet surprise confidence trust disquiet distrust quiet trust
dis/satisfaction dis/satisfaction
dissatisfaction satisfaction dissatisfaction satisfaction
ennui displeasure interest pleasure ennui displeasure interest pleasure
dis/inclination dis/inclination
disinclination inclination disinclination inclination
fear desire non-desire desire
surprise

Affect types and basic emotions. A brief note on the relation between
this proposed re-classification of emotion terms, and suggestions to clas-
sify emotions: even though there is no reason to suspect that there is a
one-to-one relation between a classification of emotion terms, and a clas-
sification of emotions (and neither should be used to classify the other),
it might be interesting to take a look at the kinds of emotions that are
mentioned in psychological research.
Some researchers in fact treat all emotions as basically the same,
with differences only in intensity and pleasantness (Ekman 1999a: 1),

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

170 Emotion Talk Across Corpora

while others differentiate between a number of ‘basic’ (Ekman 1992)

emotions.11 Generally speaking, there is no agreement in emotion
research on how many emotions there are, how many are basic, how
they are to be distinguished, or which causes and consequences they have
(Jahr 2000: 24). But Ekman (1992: 170) points out that there is quite a

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

lot of overlap concerning basic emotions (for a helpful overview see also
Turner & Stets 2005: 13–15). And Jahr notes that most researchers seem
to recognize fear, happiness, sadness, anger, disgust, and contempt with
some also including interest, guilt and surprise (Jahr 2000: 23–4). For
example, Johnson-Laird & Oatley (1989) mention happiness, sadness,
anger, fear and disgust, and Ekman’s (1999a) list includes as possible
basic emotions amusement, anger, contempt, contentment, disgust,
embarrassment, excitement, fear, guilt, pride in achievement, relief,
sadness, satisfaction, (sensory) pleasure, and shame. Adding interest,
surprise, desire, love and sympathy which have also been listed by some
researchers (for example Ekman 1992, Anderson & Leaper 1998: 426,
Kövesces 2000: 4, Oatley et al. 2006: 93, Milton 2005: 33), the emotion
terms referring to these emotional responses can certainly all be clas-
sified with the help of the modified affect framework, as visualized in
Table 5.16.12 (In this table, bold face + underlining signals that these are
recognized by all, bold face alone signals that these are recognized by
many – according to Turner & Stets (2005: 13).
It is interesting that no basic emotion terms can be found in the secu-
rity sub-category. Presumably, the emotional states that are referred to

Table 5.16 Affect types and basic emotions

un/happiness happiness happiness, pride, contentment, relief (cheer)

love, sympathy (affection)
unhappiness sadness, guilt, shame (misery)
contempt (antipathy)
in/security security
insecurity embarrassment, fear (disquiet)
dis/satisfaction satisfaction interest, excitement (interest)
pleasure, satisfaction, amusement (pleasure)
dissatisfaction anger, disgust (displeasure)
dis/inclination desire desire
non-desire
surprise surprise surprise

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 171

with security terms (confident, together, assured, trusting, comfortable with)

are not considered as emotions by psychologists – perhaps indicating
another borderline area of affect. But it is promising that basic emo-
tions can be found in each of the five types of affect suggested here:
un/happiness, in/security, dis/satisfaction, dis/inclination and surprise.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

A brief summary. Before making some more detailed remarks on
methodology, let me briefly summarize what I have said above: I have
proposed a fuzzy system of affect, with five major sets of emotions, dis-
tinguished in their more ‘core’ or less ‘core’ membership: un/happiness,
in/security, dis/satisfaction, dis/inclination and surprise. I have also sug-
gested that emotions and terms referring to them can be classified as
positive, negative or neutral (compare also Storm & Storm 1987: 810,
Nöth 1992: 82, Schrauf & Sanchez 2004: 273), and that dis/inclination
and surprise are not culturally construed as positive or negative. Further,
I have also talked above about different ways of portraying emotion.
Let’s modify our definition of affect accordingly (in analogy to Martin
& White 2005: 42). Affect is concerned with registering feelings: do we
desire something or not, do we feel happy or sad, confident or anxious,
interested or bored, surprised or unsurprised? These feelings are usually
construed as positive (happy) or negative (sad) in Western culture. Some
emotions, however, seem relatively ‘neutral’ in terms of this cultural eval-
uation, for instance surprise and desire. The portrayal of emotion (affect)
can be more or less indirect: we can label emotions directly, refer to
para-/extralinguistic symptoms of emotions, eliciting conditions, caused
behaviour, and so on, because emotion terms evoke complex knowledge
structures (schemas) that speakers associate with particular emotions.
Table 5.17 sums up the modifications of the affect system established so
far. More research is needed on the relation between, and the interaction

Table 5.17 Modifications of affect system

Valence Description of emotional response Trigger Emotion type

positive
emoters’ emotions realis in/security
emoters’ psycho-physiological expressions irrealis dis/satisfaction
of emotion
neutral emoters’ emotional language un/happiness
emoters’ actions or behaviour dis/inclination
negative
causes of emoters’ emotions surprise

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

172 Emotion Talk Across Corpora

of, these factors. For instance, the question of whether the trigger is realis
or irrealis is only relevant if a trigger is present in the first place, and
dis/inclination is one type of emotion which allows only an irrealis trig-
ger. There is, then, a definite interaction between affect type, direction
and trigger, as shown in Figure 5.8. Figure 5.8 shows that dis/inclination

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

needs a trigger (Martin & White 2005: 48), and that this trigger is always
irrealis (I-T), and also describes the fact that realis/irrealis depends on the
presence of a trigger.

un/happiness

in/security
affect
dis/satisfaction
type
surprise
affect dis/inclinationI

realis

directed at trigger
direction
irrealisT

undirected

Figure 5.8 Interaction between affect type, direction and trigger

5.3.2.3 Analysis of affect types

The classificatory grid that was used to categorize affect types in the
BRC baby (Chapter 6) is the modified system of five affect types that
was just introduced. In order to simplify the analysis, no double cod-
ings (for emotion blends, see above, and Shaver et al. 1987: 1082) were
undertaken; in each problematic case, one emotion only was given pref-
erence. Table 5.18 lists the relevant affect categories on the left, with
their definitions (that is, typical emotions in a particular category), and
the emotion terms analyzed as belonging to a certain affect category to
the right. This is a complete list of all (covert and overt) emotion terms
analyzed in the BRC baby.13

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
Table 5.18 Emotion terms analyzed in the BRC baby
Affect type Typical emotions Emotion terms
dis/inclination
desire ‘wishes, willingness, volition’ appetite for, aspire to, care to, coveted, deliberate, deliberately,
desirable, the desire (to), desire (V), desired, desperate for, eager
to, eagerly, enviable, envious, fancy, happy (for) to, hopeful,
hopefully, hope that (V), the hope (that), jealousy, long to,
had a mind to, miss, nostalgic, nostalgic affection, prefer (to),
readily, seductive, felt tempted to, tempt, wanna/want,
will/would (V), the will to do sh, willing to, willingly, wish (N),
wish (V), would/’d like, ’d rather, yearning
non-desire ‘reluctance, unwillingness, non-volition’ disinclined, involuntary, refuse to (not refuse an offer),
refusal to, reluctant, reluctantly, reluctance to, unwilling to
in/security
security: trust ‘trust in someone or in a future happening’ confident about, optimism that, optimistic, trusted
insecurity: distrust ‘distrust, reserve, suspicion’ doubtfully, emotional withdrawing, reserve, suspicious of,
suspicion about
security: quiet ‘assurance, confidence, ease, safety, assured, blithely, confident, confidence, comfortable with,
relaxation’ comforting, the ease, at ease with, reassuring, reassurance,
relax, feel safe, solace, unashamed, untroubled

(Continued)

173
10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek
 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect -
174
Table 5.18 (Continued)
Affect type Typical emotions Emotion terms
insecurity: disquiet ‘fear, worry, anxiety, puzzlement, afraid of, agitation, alarm, anxiety, anxious, anxiously,
confusion, embarrassment’ concerned (that), concern (N), confusing, daunting, dauntingly,
feels destabilised, disconcerting, distress, embarrass, fear (V),
fear of/that, fretful, frighten (V), frightened, frightening, horror,
intimidated by, intimidating, intimidation, overawed, puzzled,
puzzling, perplexing, rattled, self-conscious, self-consciously,
scared of, terrified, troubled by, unnerve, unsettled, upset,
worried, worrying, worry about (V), worries about (N)
dis/satisfaction

2011-03-17
satisfaction: interest ‘interest in, fascination with, arresting, went berserk, care about (V), carried away, curious,
excitement, entertainment’ drawn to, engaging, entertaining, as entertainers, enthusiasm
for, exhilarating, excited, excitedly, exciting, fanatical about,
fascinated, fascination with, frenzy, interest (V), interested in,
interesting, interestingly, interest (N), magnetised by,
spellbound, stirring, thrilling
dissatisfaction: ennui ‘boredom’ boring
satisfaction: pleasure ‘admiration, appeal, contentment, gratitude, admire, admirable, admiring, the appeal of, appealing,
being impressed, pleasure, pride’ appreciate, appreciative, appreciation, content with,
(feelings of) contentment, enjoy,14 entrance, gratitude,
impressed, pleased, please, pleasing, proudly, proud,
satisfaction, satisfying, satisfied with, thankful

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17
dissatisfaction: ‘anger, frustration, dissatisfaction’ anger, anger (V), angry, annoyed at, bother (V), bothered,
displeasure discontent, fed up with, frustration, furious, grudge against,
impatience at, irate, irritated, irritably, maddeningly,
mind (V), outrage, resignedly, spite (V), held on to his temper,
unsatisfactory, vexing
un/happiness
happiness: affection ‘like, love, respect, pity’ affection, beloved, care for, cherish, devotion, fond of, be
into sth, keen on, like (to), loving, his love’s, love (N),
love (V), much-loved, pity for, pitiful, respectful, reverently,
taken with, warmth, warmly, a weakness for
unhappiness: antipathy ‘hate, dislike, scorn’ cannot bear, can’t stand, disdain, dislike, gone off it, hate,
horrified, loathe, sick-making, scornful, scorned, resent
happiness: cheer ‘amusement, cheer, happiness’ amusement, amused at, brightening, cheerfulness, cheerfully, cheered
by, cheery, glad of/that, happy, happiest, happily, jovial, merry
unhappiness: misery ‘sadness, guilt, disappointment, agony, anguish, conscience was troubled over (‘guilt’),
regret, grief’ dejection, depressed, depressing, disappointing, disappointed
with, disappointment, grief (that), grieved, haunting, let down,
malaise, melancholy, melancholic, miserable, moving, painful,
regret, regretfully, sad, sadly, sorrow, (I’m/was) sorry,15 soulfully,
sullen, tormented, unhappy that, worked up about, wretchedly
surprise ‘surprise’ amazing, amazingly, astonishment, astounded, shocked, shock,
shock of, shocking, staggered, start, surprised, surprising, surprisingly,
surprise

175
10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek
5.3.3 Covert vs. overt affect
The distinction that was made in Section 3.3.4 between overt and
covert affect was also applied to the manual text analysis. In addition
to the patterns noted in Section 3.3.4 with count usages of certain
emotion nouns (for example It BE a surprise/disappointment that,What a

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

surprise/disappointment, the surprise/disappointment/appeal of), and certain
emotion nouns used as pre-modifier (for example surprise decision, hate
figure), further noun usages as well as certain adjectives and adverbs are
included as covert affect, whenever they do not label an emoter’s emo-
tional response as such but rather refer to events, things, situations and so
on that (have the power to) trigger an emotional response. This concerns
the use of adjectives such as amazing, disappointing,16 adverbs such as
amazingly, disappointingly (as stance adverbial/modal comment adjunct
rather than circumstance adverbial/circumstance of quality), and nouns
such as appeal, concern, disappointment, frenzy in certain patterns (the
frenzy of, be of concern and so on). Thus:

covert affect: It was amazing; happily, he missed; he was a disappointment

overt affect: I was amazed; he smiled happily; to our great
disappointment. . .

Table 5.19 lists emotion terms that were coded as covert affect.

Table 5.19 Analysis of covert affect in the BRC baby

Coded as Adj admirable, amazing, anxious (moments), appealing,

‘covert affect’ arresting, boring, comforting, confusing, coveted, daunting,
depressing, desirable, disappointing, disconcerting, engaging,
entertaining, enviable, exhilarating, exciting, fretful,
frightening, happy (living spaces/days), haunting, interesting,
intimidating, jovial (retort), loving (family support), moving,
painful, perplexing, pleasing, puzzling, reassuring,
(a) respectful (distance), sad (events/day), satisfying,
seductive, shocking, sick-making, stirring, surprising,
thrilling, unashamed, unsatisfactory, vexing, worrying
Adv amazingly, dauntingly, disappointingly, happily, hopefully,
interestingly, maddeningly, pitifully, sadly, surprisingly
N the appeal of, X’s concern was/is, be of concern, (a big)
disappointment,the sense of ease conveyed by, as
entertainers, my greatest fear, the frenzy of, her
horror is, his love’s, the shock of, a surprise

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect -
Table 5.20 Analysis of valence in the BRC baby
‘negative’ alarm, agitation, agony, anguish, anger/angry, annoyed, anxiety/anxious/ly, boring, bother/bothered, can’t bear/stand,
concern/concerned, confusing, dauntingly/daunting, dejection, depressing/depressed, destabilised, disappointing/
disappointed/disappointment, discontent, disconcerting, disdain, dislike, distress, embarrass, envious, fed up, fretful,
frightening/frighten//frightened/fear/afraid, frustration, furious, gone off, grief, grudge, hate, horror/horrified, impatience,
intimidation/intimidating/intimidated, irate, irritably, irritated, jealousy, loathe, let down, maddeningly, malaise,
melancholy/melancholic, miserable, miss, outrage, painful, perplexing, rattled, regret/regretful, resent, sadly/sad,
scared, scornful/scorn, self-conscious/ly, shock/shocked, shocking, sick-making, sorrow, spite, sullen, suspicion/suspicious,
temper, terrified, terrorise, tormented, troubled, unhappy, unsatisfactory, unsettled, unnerve, upset, vexing, worked up,
worry/worrying, wretched
‘positive’ appealing/appeal, appreciative/appreciate/appreciation, affection, admirable/admiring/admire, amused, assured/
reassuring/reassurance, blithely, brightening, cheered/cheerfulness/cheery/cheerfully, cherish, comfortable/comforting,

2011-03-17
confidence/confident, content/contentment, coveted, desirable, devotion, ease, enjoy, engaging, entertaining/entertainers,
enthusiasm, entrancing, enviable, exciting, exhilarating, fond of, glad, gratitude, happy/happily, hopefully/hope, impressed,
be into, jovial, keen on (so), like, love)/loving/beloved/much-loved, merry, optimism/optimistic, pity, pleased/please/pleasing,
proud/ly, relax, respectful, reverently, feel safe, satisfaction/satisfying/satisfied, solace, soulfully, taken with, thankful, trust,
untroubled, warmth/warmly
‘neutral/ambiguous’ ‘desire’: desire, involuntary, deliberate/ly, yearning, want, fancy, happy to, seductive, wishes, aspire, wish, willing, will,
unwilling, desired, eagerly, readily, keen, refused/refusal, to, reluctant/reluctance, disinclined, appetite for, would like, long,
care for/about/to, desperate (to/for), prefer to, had a mind to, disinclined, tempted, I’d rather

‘interest’: interest in/interest/interesting, fascinated/fascination with, magnetised by, fanatical about, (being) drawn to,
carried away by, excited/ly, arresting, stirring, thrilling, frenzy, curious

‘surprise’: surprising, surprised/surprise, astonishment, amazing/amazingly, spellbound, astounded, start, staggered

Other: pitifully, overawe, haunting, unashamed, doubtfully, resignedly, sorry, berserk, weakness for, puzzled/puzzling,
nostalgic, reserve, emotional withdrawing, moving, mind

177
10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek
178 Emotion Talk Across Corpora

5.3.4 Valence
Valence concerns the coding of emotion terms as positive, negative
or neutral/ambiguous, as discussed in Section 5.3.2.1 above. For the
text analysis, a general classification scheme was adopted which regards
most emotion terms relating to ‘interest’, ‘surprise’ and ‘desire’ as neu-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

tral/ambiguous. Exceptions are for example hope (coded as positive) and
shock (coded as negative). Where the valence was neither clear nor strong,
a coding of the emotion term as neutral/ambiguous was also preferred.
Again, though there may be some debate about the coding of individual
emotion terms, the analysis was consistent and is outlined in Table 5.20
(on p. 177) to ensure transparency (compare also Appendix A 2.4 online).
While this chapter was mainly concerned with outlining the theoret-
ical and methodological background to the text analysis, Chapter 6 will
report the results of analyzing affect according to these and other factors.

Notes
1. See for example, Ortony et al. (1988), Robinson (2005: 8–27). Confusingly
for linguists perhaps, this process is called appraisal (for example Ellsworth &
Scherer 2003). The hypothesis is that ‘for events to prompt emotions, they
must be evaluated, or appraised, in relation to the individual’s goals’ (Oatley
et al. 2006: 167). This can be an unconscious or potentially conscious process
(primary vs. secondary appraisal) (Oatley et al. 2006: 167). As the authors put
it, ‘[i]f we know what appraisals (or evaluations) are made we can predict the
emotion; if we know what the emotion is we can infer the appraisals’ (Oatley
et al. 2006: 21).
2. Action readiness items are, for example, ‘I wanted to oppose, to assault, hurt,
or insult’ (for antagonism), ‘I felt inhibited, paralyzed, or frozen’ (for inhib-
ition), ‘I wanted to do something, but I did not know what’ (for helplessness)
(Oatley et al. 2006: 132; more examples in Fridja et al. 1995: 129). In the
following, I will include these as actions without differentiating between
action readiness and actions.
3. Schema theory suggests that our knowledge of the world is organized in
terms of mental knowledge structures which capture the typical features
of the world (for an overview see Bednarek 2005a). They are part of our
semantic memory, and usually shared by members of the same linguistic
community (they are more or less conventionalized but can vary cross-
culturally or among sub-cultures) and can refer to both more or less factual
knowledge (spiders usually have eight legs), and to scientifically wrong folk
beliefs (spiders are insects). Concerning the structure of schemas, they are
often assumed to consist of categories and the specific interrelations (for
example X has a Y, X is on Y, X is a part of Y) existing between them, the
categories providing default assignments (by supplying prototypes) and asso-
ciated expectations (Ungerer & Schmid 1996: 212–13). In Kövecses’s terms,

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 179

‘[f]olk understandings can be thought of as knowledge structures in our con-

ceptual system’ (Kövecses 2000: 114). This knowledge is presumably based
on our experience of actual emotional responses (for example increased
body heat when angry), and observing them in others, as well as exposure
to discourse on emotions and other socialization mechanisms. That is, I
want to follow Kövecses’s suggestion that emotion schemas are both motiv-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

ated by human physiology and produced by the socio-cultural environment
(Kövecses 2000: 14). Folk knowledge also has to do with feeling rules (Section
1.3). Related to this, emotional intelligence involves knowledge of more than
emotion schemas, and has been used to refer to differences in:

communicative skills …, including discerning others’ emotions, realizing

that expressions and experience do not necessarily match, knowing the
cultural rules for displaying emotion, using the emotional vocabulary,
feeling empathy, recognizing that your own expressions affect others,
and using all of these skills in relationships with others.
(Planalp 1999: 70)

4. To this we might perhaps need to add ways that show ‘the way the world
appears to a person in that emotional state’ (Robinson 2005: 275, italics in
original).
5. With respect to a proposed cline of implicitness or indirectness, we can say
that this depends on how strongly an action, behaviour and so on is asso-
ciated with an emotion, and how prototypical it is for an emotion. The
stronger the association, and the more prototypical the action, behaviour
and so on, the less inferencing is perhaps involved. For instance, if a strong
causal relation exists, inferences are more easily drawn (O’Halloran 2003:
141) so that ‘various aspects of a scenario [schema] may be differentially
accessible at various points in processing’ (O’Halloran 2003: 188, citing
Sanford & Garrod 1994: 704).
6. The remaining four variables are hypotheticality, negation, part of speech,
and speech act – a complete guide to the analysis with respect to all nine
variables is provided in Appendix A 6.2 online.
7. I am neither a semanticist nor a lexicographer by training, but presumably,
the analysis of such a semantic field should involve thorough corpus-based
lexicographic, semantic and discourse analytic studies of these terms, per-
haps complemented by a native speaker survey. Since more than 1000
emotion terms can be identified, this is well beyond the scope of this book
(for an attempt of a taxonomy of emotion vocabulary based on elicited
data, see Storm & Storm 1987). Nevertheless, I agree with Martin & White
that ‘there is a need to develop social semiotic principles for classifying
lexis’ (Martin & White 2005: 58). There are of course existing alternatives
to Martin & White’s classification of emotion terms, for example the clas-
sification of emotion and related terms adopted by the Encarta Thesaurus
(see Section 1.6.3) or Wierzbicka (1999: 49). In terms of folk classifications,
Ortony et al. report that people asked to sort emotion terms into categories
use love, joy, surprise, anger, sadness, and fear at the basic level (Oatley
et al. 2006: 182) and Fillenbaum and Rapoport (1971: 100–24) are inter-
ested in subjects’ similarity ratings of emotion terms. Mees (2006), looking at

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

180 Emotion Talk Across Corpora

German emotion terms, classifies them as denoting ereignisfundierte Emotio-

nen (‘event-based’), Attributions-Emotionen (‘attribution-based’), Beziehungs-
Emotionen (‘relationship-based’), and Verbindungs-Emotionen (‘connection-
based’). Johnson-Laird & Oatley (1989: 96), on the basis of a specific theory
of emotion, classify emotion terms as denoting generic emotions (feelings),
basic emotions (happiness), emotional relations (love), caused emotions (glad-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

ness), causatives (reassure), emotional goals (desire) and complex emotions
(embarrassment). Ortony et al. (1987) also offer a taxonomy of 500 emotion
terms. However, these are psychologists’ classifications which are mostly
intuition-based and do not use corpus or discourse data. As pointed out
earlier, I also believe that we need to draw a clear dividing line between
classifying emotions and classifying emotion terms. Different kinds of emo-
tions have been defined as event-triggered (interest, excitement, surprise,
fear), state-triggered (happiness, distress) or action-triggered (pride, shame,
guilt) (Izard 1977, in Daneš 2004; compare also Fiehler 1990: 47). Simi-
larly, Ortony et al. (1988: 14) differentiate between Event-based emotions
(related to goals), Attribution emotions (related to agents and standards of
behaviour), and Attraction emotions (related to attitudes). Emotions can also
be distinguished with reference to goal achievement (Ekman 1999a: 2). Such
classifications are problematic for linguistic purposes since expressions such
as I am angry can be used for describing emotions about actions, situations or
persons (Daneš 2004: 28 and Chapter 4 above). Compare also the overviews
given in Jahr (2000: 23), Daneš (2004: 30–1), and Turner & Stets (2005: 289)
on descriptions of different types of emotions.
8. The assumption that surprise is not construed as a negative emotion is also
confirmed by the classification of experimental subjects of surprise terms
as positive rather than negative (Wallace & Carson 1973: 16), as well as by
semantic change (amazing meaning ‘very surprising, especially in a way that
makes you feel pleasure or admiration’, OALD, emphasis mine). Note also
that there are both lexical items that conflate unexpectedness and positive
evaluation (miraculous) and items that conflate unexpectedness and negative
evaluation (alarming) (Lemke 1998).
9. This is perhaps the case because the mental state of surprise has a problem-
atic status in emotion research (Ungerer 1997: 326, Robinson 2005: 64). For
example, Daneš (2004: 27) lists surprise as a neutral emotion, Johnson-Laird
& Oakley (1989: 102) classify it as a reaction rather than a distinct emotion,
and others include it only as emotion if it involves an evaluative specification
(Jahr 2000: 26). Ekman suggests that surprise might be ‘perceived differently
than other emotions’ (Ekman 1992: 176). While most emotion theorists
seem to include surprise (Ortony et al. 1988: 32), this depends on the par-
ticular theory of emotion involved. For example, Ortony et al. (1988) claim
that emotion is only related to positive–negative appraisal, and therefore
excludes neutral surprise (since this is related only to an appraisal of unex-
pectedness). Reasons for including surprise as an emotion in psychological
research are that, like other emotions (see below):

• it is based on a cognitive evaluation of the environment;

• it has associated typical actions/behaviour;
• it has an associated physiological response;

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Mapping and Analyzing Affect 181

• it has an associated facial expression (for an illustration of this see

(http://www.woodstock.edu/myers6e/content/psychsim/, last accessed
12 October 2006).

For appraisal theory, whether or not psychologists consider surprise as an

emotion is only one of the considerations to be taken into account, another

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

might be whether speakers consider surprise terms as being emotion terms or
not. In experiments where subjects had to free-list emotion terms, surprise
terms were in fact not core members of the emotion category (Shaver et al.
1987), but arguably still members. If we assume that the field of emotion
terms has no clear boundaries, and that emotion terms ‘designate fuzzy sets’
(Shaver et al. 1987: 1063), feelings involving surprise, like dis/satisfaction
(see above), are on the borders of ‘true’ affect but can still be included in the
affect system.
10. Let me make a more general point: it seems necessary to make a distinction
between positivity/negativity in terms of cultural construal vs. positiv-
ity/negativity in terms of the evaluation that can be implied by using an emo-
tion term. For instance, I may use an expression denoting surprise to evaluate
someone negatively or positively (Bednarek 2006b), but this does not nec-
essarily mean that the surprise felt by the emoter is construed as positive or
negative. Furthermore, there is no automatic correlation between positive
emotions and positive evaluation: to refer to an emoter’s positive emotional
reaction (for example delight) to a trigger that is evaluated negatively by the
writer/speaker may result in a negative evaluation of the emoter, rather than
a positive one (Bednarek 2006a: 210). This might constitute a problem for
computational linguistics when a seed list is used with emotion terms that are
tagged as ‘positive’ or ‘negative’, in order to analyze automatically the type
of evaluation in texts (for computational linguistic approaches to appraisal
see Taboada & Grieve 2004, Whitelaw et al. 2005, Bloom et al. 2007).
11. There are at least three different ways in which the term basic is used in
emotion research (Ekman 1992: 170, Kövecses 2000: 3). Different criteria for
basic emotions are given by Johnson-Laird & Oatley (1989), Ekman (1992:
175–89), and Jahr (2000: 24), and include the presence of distinctive uni-
versal signals (for example, facial expression), the existence of the given
emotion in other primates, a quick onset of the emotion and a brief dur-
ation (Ekman 1992, 1999a). For some researchers a basic emotion is not a
distinct emotional state but rather ‘an emotion family [which] can be con-
sidered to constitute a theme and variations’ (Ekman 1999a: 8). Similarly,
Johnson-Laird & Oatley point out that ‘[a]round each mode [happiness, sad-
ness, anger, fear, disgust], there may cluster a family of related emotional
experiences’ (Johnson-Laird & Oatley 1989: 85).
12. It is of course problematic that emotion research uses English words to clas-
sify emotional states (compare Wierzbicka 1992a, b, 1998, 1999). Wierzbicka
instead proposes to use a language made up of universal semantic primitives
for their description. As Athanasiadou & Tabakowska explain: ‘If there are
some innate and universal cognitive scenarios which play a special role in
human lives all over the world, such scenarios would have to be identified
via such lexical universals, and not via culture-specific words such as sad-
ness or anger’ (Athanasiadou & Tabakowska 1998b: xiv). Wierzbicka (1999:

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

182 Emotion Talk Across Corpora

Chapter 2) gives an analysis of about fifty emotion concepts in English and

their underlying cognitive scenarios in terms of semantic primitives (but not
based on large-scale corpus linguistics).
13. Word combinations such as nostalgic affection, preferential love, compassionate
love were coded as one emotion (nostalgia, love) rather than two.
14. In contrast to Martin (2000a: 153) I classify enjoy as satisfaction: pleasure

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

rather than happiness: affection, since enjoying something means that
something gives you pleasure, and seems to be goal-related (which is what
dis/satisfaction is about, see Martin & White 2005: 49). Vice versa, I clas-
sify respect as happiness: affection rather than dis/satisfaction (Martin &
White 2005: 49), since it seems to me not related to the pursuit of goals
but rather to affection/liking. It must be noted that there is some overlap
between dis/satisfaction and un/happiness.
15. I coded I’m/I am sorry only as unhappiness: misery; the general, convention-
alized use of sorry was excluded from affect, since its emotional meaning is
bleached.
16. These are classified by appraisal theory as appreciation: reaction rather than
affect (White 2001: 3, White 2002: 16, Martin & White 2005: 57–8), but,
as I have argued elsewhere (Bednarek forthcoming 2008), they constitute a
‘bridge’ between appreciation/judgement and affect, and imply emotional
responses, which is why they were included as covert affect.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

6
Enacting Affect: Pragmatic Analysis

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

As proposed, the large-scale studies of Chapters 2 to 4 are complemented
by a brief foray into the pragmatics of emotion talk in the BRC baby using
a modified version of appraisal theory and the methodology described in
Chapter 5 (also in Appendix A 6.2 online). The 85,000 words of the BRC
baby were manually analyzed for all listed variables, but because of space
constraints only extracts are included in this chapter. Such a small-scale
analysis allows us to make safer guesses about the pragmatics of emotion
talk in the four registers than are possible using large-scale analysis, but
is less representative. (The small-scale analyses are necessary because it
takes time to analyze texts manually according to the methodology out-
lined in Chapter 5.) No detailed, ‘truly’ qualitative analyses of each of
the roughly 60 texts in the BRC baby were undertaken, which would take
into account textual development, or logogenesis (Halliday & Matthiessen
2004: 43) and social context, though an extended extract from fiction
will be commented upon. This is because there are already many such
qualitative analyses using the appraisal framework (see Section 5.1 for ref-
erences). The aim of this chapter is thus to open up research by pointing
to future fields of study, showing opportunities for new research projects,
and laying out a ‘road-map’ for further research into affect. More specif-
ically, this chapter discusses the notions of affective key and stance before
outlining some typical functions of emotion terms.

6.1 Affective key and stance

Much research has shown that affect clusters or patterns in text. This has
been discussed as ‘prosody’ in appraisal theory (Macken-Horarik 2003b,
Martin & White 2005: 19–23; on ‘emotive prosody’ from a non-SFL

183

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

184 Emotion Talk Across Corpora

perspective, see Bublitz 2002, 2003), and is explained as ‘the spread,

sprawl, smear or diffusion of interpersonal meanings that accumulate,
reinforce, or resonate with each other to construct an evaluative “key”
over an extended segment of text’ (Hood 2006: 38). Evaluative keys con-
cern the co-occurrence of evaluative options (Martin & White 2005: 164):

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

‘patterns in the use of evaluative resources within texts by which certain
types of evaluation and stance are favoured or foregrounded while others
occur infrequently, only in restricted settings, or not at all.’ (Martin &
White 2005: 161). Evaluative key is defined in terms of patterns for all
systems of appraisal (attitude, engagement and graduation), but since
the focus of this book is on emotion talk, I make use of the notion of the
more narrow concept of affective key instead. This is described in terms
of the co-occurrence of particular configurations of affect.
Whereas ‘key’ refers to patterns across texts, ‘stance’ refers to evalu-
ative patterns ‘within a given “key” associated with particular rhetorical
objectives and the construction of authorial personae’ (Martin & White
2005: 164). (This notion of stance is not to be confused with the defin-
ition of ‘stance’ adopted by Biber and his colleagues, where it is more
or less equivalent to appraisal: see Section 1.5.) I shall broaden the
notion of stance to describe the construction of personae in texts in
general – whether these personae are authorial or not. Again, the focus
is on affect. That is, I define affective stance as the co-selection or pat-
terning of affect types in stretches of text related to the construal of
certain authorial and non-authorial personae with respect to emotional
experience.
In the following sections, I shall comment on affective key and stance
in the four registers, based on analyses of the BRC baby. While the discus-
sion of conversation focuses on affective key and that of fiction focuses
on affective stance, comments on both key and stance shall be made
with respect to news reportage and academic discourse.

Conversation. Affective keys in conversation seem to depend on the

conversational topic, and be task-related. Only one conversation in my
data (KE2) exhibits a clear affective key, containing exclusively desire and
affection terms, ostensibly associated with a shopping trip. As Table 6.1
shows, it is predominantly the speaker and the listener (I, you) who are
the emoters. We can see here a co-occurrence of the affect types desire
and affection – an affective key that we might perhaps label ‘consumer
voice’ (‘I like’ → ‘I want’; ‘I don’t like’ → ‘I don’t want’). Within the con-
sumer voice, the text alternates between a desire and an affection stance.
None of the other conversations represent clear examples of additional

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 185

Table 6.1 Affect in KE2 (BRC)

Emoter Emotion term Affect type Trigger

you wanted inclination: desire some pot some [flowers]
you want inclination: desire them [flowers] now

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

you wanted inclination: desire them [flowers]
I like happiness: affection it [the weather]
I like happiness: affection that [the weather]
you want inclination: desire me to show you what they
look like
my wife tempted inclination: desire context: flowers
she wants inclination: desire your advice
I want inclination: desire a better white bigger than
that one
I liked happiness: affection the look of [the one]
you want inclination: desire it in the red one and not
the silver one
I like happiness: affection that [flower]
[does] he want inclination: desire to play
you want to inclination: desire change another note
I like happiness: affection it in there
Tim taken with happiness: affection that big card
you want inclination: desire whatever

affective keys, as Figure 6.1 (available in colour in Appendix A 6.3 online)

shows.
According to Figure 6.1 (on page 186), perhaps KE4 is most similar
to KE2 in terms of affect sub-types; however, it also has occurrences of
antipathy and pleasure, and there are differences regarding emoters. The
occurrence of affection and desire across all corpora also reflects the find-
ings in 2.2.1 above, that the verbs want and like are uncharacteristically
frequent in conversation. An important question to be tackled in future
research, then, is whether the notion of affective key has validity in con-
versation, and if so, which other affective or evaluative keys might exist
in different genres of conversation (a possible candidate is a negative
judgement key in gossip; compare Eggins & Slade 1997: 273–311).

Fiction. Affective keys in fiction can perhaps be associated with the

types of focalization identified by Rimmon-Kenan, who differentiates
between objective (uninvolved) vs. subjective (involved) emotive point
of view (Rimmon-Kenan 2002: 82). Differences in emotive point of
view, she says, also have to do with whether character emotions have

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

186 Emotion Talk Across Corpora

120
Misery
100 Non-desire
Displeasure
80 Interest

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Antipathy
60 Surprise
Cheer
40
Pleasure

20 Disquiet
Affection
0 Desire
KBC KBD KBP KD0 KD8 KE2 KE4

Figure 6.1 Distribution of affect sub-types (but not emoters) in the conversation
sub-corpus of BRC baby (Figures refer to percentages of affect sub-type with respect
to all emotion terms in given file; for instance, 44.4% of all emotion terms in KBC
realize Desire)

to be inferred from behaviour or whether the emotions are explicitly

revealed by interior monologue or the narrator (compare the distinctions
made in 5.3.1). However, ‘[t]he whole gamut of stylistic possibilities [of
indicating focalization] has not yet been established’ (Rimmon-Kenan
2002: 85). In SFL, it has been suggested that evaluation is crucial
in the establishment of narrative voice (Martin & White 2005: 72,
Macken-Horarik 2003a) but, to my knowledge, affective or evaluative
key has not yet been systematically related to focalization. This mat-
ter is, however, a very complex one (for overviews of focalization see
Simpson 1993, Toolan 2001), and a burning issue is to investigate how
different systems of appraisal and modality (epistemic/modalization or
deontic/modulation) construe a focalizer’s point of view. For instance,
Simpson (1993) makes a useful distinction between four facets of speaker
attitude: deontic (obligation, duty, commitment, related to modula-
tion and engagement), boulomaic (desire: related to attitude), epis-
temic (knowledge, belief and cognition, related to affect [quiet, trust],
engagement and modalization), and perception (related to attitude and
engagement).
On account of this complexity, the analysis below focuses on affec-
tive stance rather than key; that is, it is concerned with how characters
are construed as personae, rather than narrative voice (compare also

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 187

the emotional response inventory established in Appendix A 6.4 online).

Here is an extended extract from one of the analyzed samples:

(1) AB 9 (Death of a Partner, by Janet Neel, Constable Company Ltd,

London, 1991)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

After chatting to Davidson for a few minutes longer, he went back
to his office, looking for his secretary, a plumpish, infinitely com-
petent, middle-aged mother of three. She was on the phone and he
stood silently, filling the doorway, raising an eyebrow at her. She
motioned him to stay.
‘It’s Francesca, John.’
‘Thanks, Jenny.
I’ll take it back there,’ he said, brightening, and she watched, with
a little jealousy, as he hurried back to his desk.
‘Darling.
I’ve got a horrendous problem.
Can you possibly duck out and have coffee in the caff in ten
minutes?’
The voice at the other end was slightly husky but very clear.
‘It’s Tristram.
He’s been arrested in New York.
Better not talk on the phone, had we?’
‘Oh, Christ.
No.
I’ll meet you.’
He put down the phone, rattled. With his much-loved Francesca
came her four younger brothers for whom, as the eldest child of a
widow, she had always considered herself responsible. All four were
talented musicians and difficult people, in varying degrees; Tris-
tram, one of the twenty-four-year-old twins, had proved the most
difficult, perhaps because he was not the most talented. McLeish
gritted his teeth, and, trying not to consider the implications of
what he had been told made one quick phone call, then took the
lift down and walked across the road from New Scotland Yard to the
little café which was, as usual, full of workmen engaged in rebuild-
ing the offices in the area. Stopping to buy a paper, he caught sight
of Francesca through the window, perched on one of the bar stools,
totally unconscious of the table full of men next to her all eyeing her
long legs. He stood and watched her, putting off the moment when,
as he half knew, he would be asked to acquiesce in some lunatic

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

188 Emotion Talk Across Corpora

scheme for pulling Tristram out of trouble, and saw her for a moment
from a position of detachment: a tall young woman looking younger
than her twenty-nine years. dark [sic], with a long straight nose
and arched eyebrows. She was looking particularly uncompromis-
ing today, tired and pale, her dark, short hair spiking up at the back.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

She had been crying, McLeish observed resignedly, as he pushed
through the door.
‘All right, tell me,’ he said, as he sat down heavily beside her, and
got a careful, measuring, sidelong look.
‘I have to go to New York in a hurry and bail Tristram out.
He was arrested last night, with one of the backing group and three
of the band.
They’re all in the nick, charged with possession.
Cocaine, as I understand it.’
‘Jesus.’
‘I know, I know, just when he was beginning to be a success like
Perry.
It simply went to his head – you know what he was like when he left.’
‘Thought he could walk on water,’ McLeish said, in irritated
memory.
‘Frannie, why do you have to go?
He has a manager and a studio, doesn’t he?
What can you do that they can’t?’
She sighed so heavily that her whole ribcage moved.
‘I am to some extent on home ground there,’ she said, reluctantly
and not looking at him.
‘Mike – Michael O’Brien – will help, but I need to be on the spot.’
McLeish held on to his temper, reminding himself that he was
very tired. Francesca’s much publicized affair with Senator Michael
O’Brien was the reason that she had been sent home rather early
from a tour of duty in the Embassy in Washington, over a year
previously, just before he had met her in London. As one of
Francesca’s DTI colleagues had maliciously observed, it had been
felt that fraternization with the American colleagues could be car-
ried too far. Francesca herself had characteristically taken the line
that the Foreign Office ought to have been glad that someone
on the staff was that closely involved with the American political
Establishment.
‘What are you going to do with O’Brien?’

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 189

‘Well, I’m hardly going to get back into bed with him after more
than a whole year.
There’ll certainly be another incumbent by now.
But he is the senior Senator, we were close when I was in
Washington, and whichever way you slice it he won’t want my

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

brother being buggered or beaten up in a New York jail.’
Her classically good diction always became even clearer under stress,
so that this statement emerged with the slightly metallic clarity of
a dubbed film, and the clientèle of the small café was obviously
appreciating every moment. McLeish decided that since she was
unaware of her audience it did not become him to be selfconscious,
and asked what she expected even a Senator to achieve in these
circumstances?
‘Oh, darling.
The American legal system is so odd that I’ve been told I could get
Tristram deported in my custody.
It’s a disaster, of course it’s a disaster, he’ll be banned from the US
for ever presumably; but at least he won’t be being interfered with
in some unspeakable foreign nick.’
‘I thought he was off drugs?’
‘Well we all hoped, didn’t we?
But evidently he isn’t, and I must get him back.’
McLeish found himself on the verge of suggesting that a thoroughly
unpleasant time in a New York jail might succeed in curing Tristram
where all other methods, including exhortation, loving family sup-
port and a spell in a comfortable private hospital in Devon, had
failed. He looked at his love’s shuttered, miserable face and realized
he would get nowhere along those lines.
‘Why can’t your Mum or one of the boys go?’
‘Mum is in bed with bronchitis, as you would know if you had
managed to get out of that place since last Friday.’
Francesca, a true eldest child, knew how to score her points.
‘Charlie’s baby is due tomorrow, Perry is in Japan on tour, just like
all the papers say.
Jeremy is coming with me but he is too young to do this alone, and
in any case I’m the only one who can deploy O’Brien.’
‘How is the DTI receiving all this?
You’ve got four rescue cases.’
Francesca sighed.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

190 Emotion Talk Across Corpora

‘They are as fed up with me as you are, but they won’t stop me
taking leave.’
She stopped sharply, and blushed scarlet.
‘Wait a minute.’
John McLeish felt his blood-pressure going up.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

‘What about skiing, if you’re using leave?’
She looked at him, wretchedly.
‘I have to go.
I really expect to get back in three or four days and I will try and
hang on to the holiday.
I know we need it, it seems ages since we went to bed.’
McLeish scowled round the fascinated audience, returning the
customers to their egg and chips.
‘Anyway, it’s been you who have been too busy for months.’
McLeish was too honest not to acknowledge the point.
‘The more important, then, for us to have a holiday together.’
‘I know.
Darling, I am sorry, I really expect to be back inside the week, and
I wouldn’t go if anyone else could.
I am trying – I mean I know I let the boys lean too much.’
He looked at her, defeated, and she saw that she had carried the
point, but at a heavy cost.
‘Do you still love me?’ she asked anxiously.
‘Not at the moment.’
An indrawn breath from the spectators unsettled both of them for
a moment.
Francesca nearly laughed, but McLeish’s expression sobered her.
‘Let’s get out of here.’
She slid obediently off the bar stool, bidding a civil good-morning
to the café owner who looked, McLeish observed, as if he would
willingly have swept her into his plump Italian embrace, and they
walked together to the gates of New Scotland Yard.
‘I’ll ring you when I get to New York.’
She looked, worried, at his profile.
‘You’re furious with me.
I’m sorry.
I love you.’
‘I don’t think we can go on like this.’
McLeish surprised himself as well as her.
‘John.’

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 191

It was an appeal and he was not proof against it.

He bent and kissed her.
‘I am furious, and I want you back quickly.
But good luck with it — ring me if you need help.’
The look she flicked him reminded him that this was one area where

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

she would not appeal for his help; she had been more than careful
to protect him from any involvement with a drug-taking brother. It
was one of her many advantages that as a professional civil servant
herself she understood the constraints of his career. She turned to
go, head down. McLeish saw that she was crying again, but decided
coldly there was nothing he was going to do and trudged wearily
back to the lift, the morning’s cheerfulness totally evaporated.
(BRC, AB 9 26–139)

Table 6.2 Affect types of characters

Affect type Emotion terms

John happiness: cheer brightening, cheerfulness
McLeish happiness: affection much-loved, love, love
security: disquiet rattled, self-conscious
satisfaction: displeasure resignedly, irritated, held on to his temper,
fed up with, furious, furious
inclination: desire want

Francesca inclination: non-desire reluctantly

happiness: misery miserable, wretchedly, sorry, sorry
security: disquiet anxiously, worried
happiness: affection love

Disregarding the audience at the café, whose affective stance is one of

satisfaction: pleasure (appreciating) and satisfaction: interest (fascinated),
and other minor characters (O’Brien, the café owner, McLeish’s secre-
tary and so on), Table 6.2 shows which affect types are associated with
McLeish and Francesca in this sample. McLeish has mostly emotions
of cheer (later negated), affection (towards Francesca), and satisfaction:
displeasure, whereas Francesca has mostly emotions of misery and dis-
quiet. There might perhaps be a reason for arguing that Francesca’s
emotions (anxiety, worry) are more ‘female’ than McLeish’s emotions
(anger), though it is worth pointing out that McLeish also has a consid-
erable amount of affection and some disquiet. However, the references to
McLeish’s love (affection) for Francesca occur only in descriptive passages
rather than character dialogue, with one passivized reference and one of

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

192 Emotion Talk Across Corpora

covert rather than overt affect:

• With his much-loved Francesca came her four younger brothers for
whom, as the eldest child of a widow, she had always considered
herself responsible.
• He looked at his love’s [covert affect] shuttered, miserable face and

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

realized he would get nowhere along those lines.

One reference to McLeish’s love does occur in character dialogue, but

Francesca is the speaker, and the existence of the emotion is actually
questioned by her, conforming to a wide-spread stereotype about female
insecurity and behaviour:

• ‘Do you still love me?’ she asked anxiously.

In fact, it is interesting to look more closely at the dialogue between

McLeish and Francesca and their use of emotion terms there.
Francesca:

• ‘Well we all hoped, didn’t we?

• ‘They are as fed up with me as you are, but they won’t stop me taking
leave.’
• ‘I know. Darling, I am sorry, I really expect to be back inside the week,
and I wouldn’t go if anyone else could.
• ‘Do you still love me?.’
• ‘You’re furious with me.
• I’m sorry.
• I love you.’

McLeish

• ‘Not at the moment.’ [ellipted: I don’t love you at the moment]

• ‘I am furious, and I want you back quickly.

There are clearly more emotion terms in Francesca’s contribution to

the dialogue, namely authorial misery and affection, and non-authorial
affection and displeasure with McLeish as emoter. In contrast, McLeish’s
dialogue has only one ellipted reference to negated affection, one to dis-
pleasure and one to desire – all authorial statements, with no questions
to Francesca about her emotions (construing him as more confident
than her).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 193

Summing up, a number of factors contribute to the relatively stereo-

typical construal of the characters in this extract:

• the attribution of ‘male’ emotions to McLeish and ‘female’ emotions

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

to Francesca
• Francesca’s questioning of McLeish’s love and her overt statement of
her own affection (I love you)
• the use of more emotion terms by Francesca than by McLeish in
character dialogue.

It can of course be argued that this construal does not so much reflect
male–female differences and a stereotypical construal of gender, but that
it in fact relates to the social action at hand. In fact, Francesca is the
one jeopardizing the relationship, and therefore has to make up for it,
which puts her in a perhaps only temporary position of lesser power
or confidence. Politeness strategies and relationship management rules
then constrain her affective behaviour. Thus, one could even argue that
Francesca’s behaviour is stereotypically ‘masculine’ or ‘male’ (cancelling
their holiday, taking care of things without help). In fact, at least some of
Francesca’s usage of emotion terms is the result of her trying to negotiate
the relationship, with the problem being in a way ‘her fault’: her misery
and disquiet are the result of McLeish’s displeasure at her decision. How-
ever, she arguably uses stereotypically ‘female’ strategies to do so (saying
how sorry she is, that she loves him), and if we look at other aspects
of the text such as descriptions of non-verbal behaviour (for example
Francesca being perched on one of the bar stools, sighing, blushing scar-
let, crying; McLeish gritting his teeth, scowling), and the characteristics
of their language (for example Francesca using ‘female’ terms such as
darling, horrendous, caff and tags (had we); McLeish preferring the more
‘male’ swear words Christ, Jesus and Frannie rather than darling/sweetheart)
a gender-related construal of affective stereotypes does strongly suggest
itself. We can also look at behavioural surge expressions and agency,
such as He bent and kissed her (rather than she kissed him), which reflects
a general tendency in the culture for men to be represented as kissing
women rather than the other way round (Hunston 2006). For a compari-
son with present research on affect and gender refer to Besnier (1990),
Lutz (1990), Gallois (1994: 306–7), Anderson & Leaper (1998), Planalp
(1999: 36), Goldshmidt & Weller (2000), Galasiński (2004), Oatley et al.
(2006: 246–8).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

194 Emotion Talk Across Corpora

Table 6.3 Affect sub-types in news reportage categories in the BRC baby

Affect sub-type Commerce Report Science Social1 Sports Arts All

inclination: desire 7 4 4 3 4 7 28
inclination: non-desire – 1 1 1 1 – 4

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

satisfaction: interest 2 2 3 – 5 4 16
satisfaction: pleasure – 2 2 – 5 2 11
satisfaction: ennui – – – – – – –
satisfaction: displeasure 2 – 1 – – 1 4
security: quiet – – – 3 – 3 6
security: disquiet 6 2 1 2 2 3 16
security: trust – – – 2 – –
security: distrust – – – – – 1 1
happiness: affection – 2 1 4 – – 7
happiness: antipathy – 1 – – – – 1
happiness: cheer – – 1 – 2 – 3
happiness: misery 1 2 1 1 4 4 13
surprise – – 1 – – 1 2
All 18 16 16 16 23 26 112

1 See Note 1 to the chapter.

News reportage. Affective key in news reportage can be investigated with

regard to at least four aspects of affect: (1) sub-types of affect, (2) covert
vs. overt affect, (3) authorial vs. non-authorial affect, (4) mediated vs.
unmediated affect (having to do with different types of reporting affect,
discussed below).
Starting with (1), Table 6.3 lists affect sub-types and their frequency
in the six BRC categories of news reportage. What Table 6.3 shows
is that inclination: desire, security: disquiet and happiness: misery
occur across all categories, and are frequent overall, with some varia-
tion between the categories where disquiet and misery are concerned.
It is not surprising that two of these three categories are negative, as
I have shown elsewhere (Bednarek 2006a: 179) that negative emotions
clearly outweigh positive emotions in news stories – this is related to
the news value of negativity. Fowler notes that references to terms that
denote negative emotions such as fear and confusion can create stri-
dency and contribute to ‘hysterical discourse’ (Fowler 1991: 164) in
the press. Satisfaction: interest and satisfaction: pleasure are also rela-
tively frequent overall, but occur only in four/five of the six categories,

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 195

Table 6.4 Affect types of 50 most frequent emotion terms in the BRC news
reportage sub-corpus

Affect type Emotion terms

disquiet worried, horror, worry, concerned, anxious, worry, fear,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

embarrassment, concern
desire keen, desperate, willing, care, desire, hope, prepared, ambition
pleasure pleasure, pleased, proud, delighted, admire, delight, enjoy
displeasure hate, anger, fury, furious, angry
misery disappointed, sad, unhappy, disappointment
surprise shock, surprised, shocked, surprise
affection love, sympathy, love, respect
interest enthusiasm, excitement, passion
cheer joy, happy
antipathy tension
other feeling, expectation, bitter

respectively, and there is a considerable amount of variation between

the categories. The importance of desire, disquiet and pleasure (and to
a certain extent that of misery) for news reportage is in fact confirmed
by a classification of the 50 most frequent emotion terms in the 2.6
million BRC news reportage sub-corpus – as summarized in Table 6.4.
Beyond this, it is difficult to make generalizations because of the small
size of the BRC baby, though there is perhaps a particular preference
in Sports and Arts reporting for satisfaction: interest, and in Sports
reporting for satisfaction: pleasure as well as in Commerce reporting
for security: disquiet (however, these tendencies may just be present in
my data).
Turning now the spotlight on authorial/non-authorial, overt/covert
and mediated/unmediated affect, definitions for the first two pairs of
concepts have already been given in Chapter 5. The difference between
mediated and unmediated affect goes back to a comment by Mar-
tin and White (2005: 177), where mediated affect relates to emotion
terms in reported speech and thought, and unmediated affect directly
asserts a non-authorial emoter’s emotion. Consequently, both direct
and indirect speech are included as mediated affect. I also code cases
of reported affect as mediated where speakers report on someone else’s
emotion (for example ‘Goldsmith decided that if you wanted to be
a capitalist the UK and France were not the places to be’). Examples for

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

196 Emotion Talk Across Corpora

unmediated and mediated covert and overt non-authorial affect from my

data are:2

Non-authorial affect

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Covert

unmediated: I remember once seeing a small girl remove a tin of soup

from halfway down a display stack in a supermarket. The
predictable, and very satisfying, result attracted the
attention of a nearby assistant . . . 3 (A1M 62–63)
mediated: Calcavecchia said, yawning in mid-sentence. . . .
‘It seems that every time I come back from a big
disappointment I win’ (A1N 112–115)

Overt

unmediated: Once people overcome their fears about

computers. . . (A1M 54)
mediated: ‘I really enjoyed it’, Calcavecchia said (A1N 112)
Mr Rich said many dentists were concerned that . . .
(AL5 11)
Hard, they say, to be too overawed by a character
like Cage . . . (A7S 11)

Table 6.5 shows the distribution of these different aspects of affect in

the news data from the BRC baby.
Two provisory sets of texts emerge from this comparison: in the first
set, most texts exhibit a preference for either unmediated or mediated
non-authorial overt affect, mostly preferring unmediated affect over
mediated affect. If authorial affect occurs (which it does only rarely), it
is covert. Most texts in Commerce, Report, Science, and Social reporting
fall into this group. The second set – Sports and Arts reporting – is charac-
terized by the presence of authorial affect, though it also has a lot of non-
authorial mediated and unmediated overt affect. Sports has only covert
authorial affect, whereas Arts includes both covert and overt authorial
affect (though the latter is generalized: you have to hope, making one long
to). In terms of affective key, the first set probably falls into Martin &
White’s (2005: 178) reporter voice (− authorial affect + observed/non-
authorial affect), and the second set is perhaps more akin to their
commentator voice (+ authorial affect + observed/non-authorial affect),
although I have not looked at judgement/appreciation patterns that

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 197

Table 6.5 Affect and mediation in the BRC baby news reportage sub-corpus

Commerce Report Science Social Sports Arts

Authorial affect
covert 1 (1)4 2 – 6 8

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

overt – (2) – – – 2
Non-authorial
affect
covert
unmediated 2 – 2 – 2 –
mediated – – – – 2 1
overt
unmediated 9 8 9 4 8 12
mediated 6 4 3 12 4 3

4 See Note 4 to the chapter.

distinguish reporter, commentator, and correspondent voice (for details

see Iedema et al. 1994, White 1998, Martin & White 2005: 164–84).
Finally, let’s look very briefly at affective stance to consider the con-
strual of news actors’ personae in news texts. News actors can use
affective patterns to construe themselves as particular personae in their
statements to reporters (and hope that this construal is taken up), as in
this example:

(2) She [Health Secretary Virginia Bottomley] recalled a promise made

by Mr Major when he became Prime Minister: that he would work
for a nation at ease with itself. ‘I want to see a health service at ease
with itself — optimistic and confident about its essential work.
(BRC, Al5 25–26)

Here co-selections of desire (want) and security (at ease, optimistic,

confident) position the speaker as someone whose goal is the well-being
of the health service, appealing to a common human desire for security.
It is an example of a non-human emoter (the NHS) being construed
as a conscious being capable of experiencing emotion (Halliday &
Matthiessen 2004: 202–3) – compare also the discussion of emoters in
3.3.2 above.
Another example, from Social reporting, is:

(3) One [woman in the study] insisted on cleanliness but didn’t care
about tidiness, while her husband didn’t care about cleanliness but

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

198 Emotion Talk Across Corpora

was fanatical about tidiness. Sabine begins the day by making her
husband look at a pile of dirty washing by the bed. Anne-Sophie,
on the other hand, lives a life wracked by doubt because she can
never decide whether a garment is dirty or not. Each time she tries
to judge she suffers ‘a feeling of agitation and indeed anguish’. Her

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

chief horror is to have clothes lying around that are neither one
thing nor the other. She inspects the scene, she smells the garment.
Eventually she puts it away, deciding not to wash it because she
hates ironing. But the malaise persists and all day she feels ‘desta-
bilised’. [· · · ] Leon and Madelaine, meanwhile, have chairs on either
side of their bed on which to put their respective clothes. They pile
up. Madelaine refuses to put Leon’s away. This bothers him deep
down. He believes it is her job. Out loud, he complains about the
pile on her chair. She resents this. (BRC, AL5 89–113)

In the extract, the affective stance of the first couple (woman and
husband) is construed in terms of satisfaction: interest (didn’t care about,
didn’t care about, fanatical about – insist on implies desire). (This shows
how emotion terms can be used to construe people’s values rather than
momentary emotional states – see also fiction below. As has been noted,
‘some emotional words are systematically ambiguous because they can
be used to refer either to an immediate subjective feeling or to a general
predisposition.’ (Johnson-Laird & Oatley 1989: 90; similarly Ortony et al.
1987: 350, Edwards 1999: 282). For example, an emoter can be described
as a generally angry person (disposition) or as being angry at a particular
thing at a particular time (momentary). In contrast, Anne-Sophie’s affect-
ive stance is predominantly one of disquiet (agitation, destabilised) and
misery (anguish, horror, malaise). In the description of the third couple,
a causal chain is presented where Madelaine’s non-desire causes Leon
displeasure (shown externally by complains about), which in turn causes
Madelaine’s antipathy – an example of a vicious circle that cannot easily
be broken. In order to break it, the participants ‘have to step outside the
circle’ (Watzlawick et al. 1967: 96).

As a final example, consider:

(4) The government in Addis Ababa will [here interpreted as desire]

say nothing about the situation, and the war is not reported by
the state-run media. However, recent visitors to Dese describe the
areas as chaotic, with whole units leaderless and unwilling to fight.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 199

[. . .] The Russians have said they believe in a negotiated settlement

to the civil war, but they are continuing to supply arms to Presi-
dent Mengistu, apparently to allow him to negotiate from a position
of strength. A large consignment of weapons was unloaded from a
Soviet ship at Assab on 21 September. Weapons will not, however,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

save President Mengistu if no one is willing to use them. (BRC,
A1G 11–18)

Here, the affective stance is one of non-desire: non-desire on the part

of the government to comment, and non-desire on the part of the war
parties to fight/use weapons. Thus both personae (government and war
parties) are similarly construed in terms of affect, though the functions
of the references to affect differ: both The government in Addis Ababa will
say nothing about the situation, and the war is not reported by the state-run
media comment on the media’s accessed voices (cf. Bednarek 2006a: 92,
177) as well as evaluate the government in power, whereas the other ref-
erences seem to be used to describe and predict behaviour. From a critical
linguistic position, it might perhaps be possible to argue that such and
similar construals can contribute to Western perceptions about Africa,
if they accumulate over time and occur in connection with judgements
such as chaotic, whole units leaderless.

Academic discourse. Five different affective keys can be identified

for academic discourse from an analysis of (non-) authorial affect in my
data.

Descriptor: no affect or rare non-authorial affect (EWW, FPG)

Illustrator: no or extremely rare authorial affect; low probability of
non-authorial affect (FC1, ACJ)
Arguer: no authorial affect; median probability of non-authorial
affect (ALP)
Observer: low probability of authorial (overt or covert) affect;
highest probability of non-authorial affect (A6U, AS6, EA7)
Mentor: highest probability of authorial (covert) affect; median
probability of non-authorial affect (FEF)

This can be visualized with respect to authorial vs. non-authorial affect

as in Figure 6.2.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

200 Emotion Talk Across Corpora

More Less Less More

Mentor Observer Illustrator Mentor Observer

Descriptor Arguer

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Authorial affect Non-authorial affect

Figure 6.2 Affective key in academic discourse

The following texts exemplify these five groupings.

(5) Descriptor
If a matrix A has m rows and n columns, i.e. it is of order (m × n),
its typical element [formula] lies at the intersection of the ith row
and jth column ([formula]).
If we form the matrix [formula] by writing columns for rows in order,
the new matrix, of order (n × m) is called the transpose of A and is
denoted by [formula].
For example, [formula].
The process is clearly one of reflection in the diagonal containing
the terms [formula], which is described as the principal diagonal.
(BRC, EWW 84–90)

(6) Illustrator
In In re A Debtor (No. 1 of 1987) the Court of Appeal refused
to follow that approach and held that a statutory demand, which
on the face of it contained a number of puzzling and perplex-
ing statements, was nonetheless valid because the debtor knew
precisely what he owed and there was no injustice in holding that
his failure to pay the debt gave rise to a presumption that he was
unable to pay. (BRC, FC1 63)

(7) Arguer
In the sections to follow the stark question is therefore put ‘Is the
Probation Service in the business of inflicting pain?’ and the answer
which unfolds may be summarized ‘No, because it neither aspires
nor wishes to!’ [. . .]
In 1985 the management of Somerset Probation Service was faced
with the fact that nearly one in three of the offenders supervised in
their area were either known to be or suspected of misusing alcohol

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 201

(Singer, 1985). Aware of practitioners’ frustration in not being able

to work effectively with these problematic clients, Probation man-
agement, in partnership with the voluntary sector, determined to
develop an Alcohol Education Course (AEC). (BRC, ALP 32–45)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(8) Observer
The recourse to books, documents, or sources, underscores a percep-
tion which is somehow already in the air. The effort to recapture art’s
‘history’ is always entangled with the desire to remake its identity
and meaning as part of contemporary struggles. This article comes
out of the familiar experience of being drawn to a particular image,
or set of images, without at first knowing why, and the attempt to
account for this feeling. Looking intermittently at the so-called colo-
nial art of Latin America in churches, museums, private collections
and books, I became magnetised by the figures of angels. (BRC, A6U
7–10)

(9) Mentor
In order to have a unified view of the subject we have started with
Maxwell’s equations. It means a new approach but not a radical
departure. The subject is still the same. You will be able to see that
the laws you love and cherish (Coulomb’s, Biot-Savart’s, Snell’s,
etc.) all follow from our eqns (1.1)–(1.7). The order of discussion
will follow the traditional one: electrostatics first, followed by steady
currents, then we shall move on to slowly varying phenomena, and
reach finally the most interesting part, fast-varying phenomena,
exhibiting the full beauty of Maxwell’s wonderful equations. (BRC,
FEF 47–51)

In the descriptor voice (one text aimed at undergraduate students of

engineering and one from a text on the design of computer data files),
descriptions abound, both of the contents of the book (Chapter 8 deals
with . . .) and of the subject-matter (matrices, computer files). No refer-
ences to affect are present in the extract from engineering; the voice is
very non-affective. In the text on computer files, the rare non-authorial
affect references are used only to explain the concept of a directly organ-
ized file and to introduce a problem and show why these files are
necessary:

(10) It is possible to process the records in a directly organized file

either directly, using the keys in any desired order, or serially,

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

202 Emotion Talk Across Corpora

taking the records in the order they are stored and ignoring their
keys. . . .
A bank cannot predict which of its customers wish to with-
draw or deposit money on any given day; a travel firm cannot
control which of its package tours will be booked at a particular

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

time. Customers arrive as a result of their own decisions, and not
in an order dictated by the travel agent or bank. As these transac-
tions are not predictable they present a processing problem. (BRC,
FPG 6–12)

In the illustrator voice (texts from The Weekly Law Reports, Principles of
Criminal Law) there is some non-authorial affect (but also rare authorial
affect, one instance each), which is used to illustrate the case at hand.
This becomes apparent in example (6) above, which reports on a Court
of Appeal decision as well as in the following extract:

(11) A fairly typical set of facts is provided by Nedrick (1986), where D

had a grudge against a woman and had threatened to ‘burn her
out’. One night he went to her house, poured paraffin through
the letter-box and on to the front door, and set it alight. One
of the woman’s children died in the ensuing fire. When asked
why he did it, D replied: ‘Just to wake her up and frighten her.’
A defence of this kind, a claim that the purpose was only to
frighten and not to cause harm, requires the full definition to be
put to the jury. The question is: granted that D’s purpose was to
frighten, did he nonetheless realize that it was practically certain
that his act would cause death or grievous bodily harm to someone?
The jury should answer this by drawing inferences from the evi-
dence in the case and from the surrounding circumstances. (BRC,
ACJ 89–95)

Here, we have a clustering of in/security: disquiet terms simply because

a case is reported which deals with the question of purpose/intent.
This points to the fact that many emotion terms in criminal discourse
may derive simply from describing the motives and emotions of peo-
ple that have committed crimes – since this is of some relevance in
criminal law.
In the arguer voice, there is no authorial affect, but a slightly higher
probability of non-authorial affect than in the illustrator voice. In the

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 203

given text, from the British Journal of Social Work, the references to
overt non-authorial affect are mostly part of cause–effect relations or
arguments:

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(12) In the sections to follow the stark question is therefore put ‘Is the
Probation Service in the business of inflicting pain?’ and the answer
which unfolds may be summarized ‘No, because it neither aspires
nor wishes to!’ (ALP 32) → explaining answer . . .
THE PROJECT: MORE DEMANDING THAN PUNISHMENT
Government concern about the problem of drink-related offend-
ing is as old as the Probation Service. (ALP 33–34) → arguing for
relevance of problem . . .
In 1985 the management of Somerset Probation Service was faced
with the fact that nearly one in three of the offenders supervised in
their area were either known to be or suspected of misusing alco-
hol (Singer, 1985). Aware of practitioners’ frustration in not being
able to work effectively with these problematic clients, Probation
management, in partnership with the voluntary sector, determined
to develop an Alcohol Education Course (AEC). (ALP 44–45) →
reasoning and relevance

In the observer voice (texts from Oxford Art Journal, Tackling the Inner
Cities, France in the Making) humans and their emotions are the con-
cerns of the research itself, resulting in a fairly high proportion of
non-authorial affect to describe the research topic. For example, the first
text deals with ‘the familiar experience of being drawn to a particular
image, or set of images, without at first knowing why, and the attempt
to account for this feeling.’ The second text deals with urban poverty and
its effect on social life (mainly in terms of negative feelings of in/security,
discussed below). And the third text is about aristocratic life from 98 to
1108. Disregarding intra-textuality, sentences containing emotion terms
in this text are:

(13) The famous letter from the archbishop of Rheims to Baldwin V of

Flanders, praising him for his activities in draining and ditching the
coastal areas (presumably after the Dunkirk inundation of 1014–42
had at last subsided) is explicit evidence of an interest in agricultural
expansion. (EA7 11)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

204 Emotion Talk Across Corpora

Deliberate assistance to economic growth went hand in hand with

administrative reform aimed at channelling its profits firmly into
princely treasuries. (EA7 24)
Ademar of Chabannes told of the murder of the duke of Aquitaine’s
prévôt at St Jean d’Angély in 1026, and the destruction of his house

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

by the angry inhabitants of the town. (EA7 34)
At this point, one of Garsinde’s noble followers, unwilling to accept
the verdict, arose and demanded that the decision be made not by
arbitration but by battle. In the tumult which ensued, the terrified
monks made their escape, only to find their way blocked by the same
young lord with fifty knights, who threatened them with death.
(EA7 57–58)
At this juncture, St Foy sent a storm, the aggressive aristocrat was
struck by lightning, his terrified armed men ran away, and when
the news of divine intervention reached her, the Lady Garsinde
abandoned her case. (EA7 59)
But out of pity for Bouchard’s poverty, Geoffrey prevailed on the
abbot to allow him possession of the mill for the rest of his life. (EA7
65)
So, though Geoffrey expressed his anger after the knight Walter had
killed one of his kinsmen, he accepted two mills from him as the
price of his peace. (EA7 68)
This may have been the simple truth; it may, alternatively, have
been an instance of the use of delay by an intelligent but reluctant
pleader. (EA7 77)

Since the text deals with the ruling class in early medieval soci-
ety, it involves statements about potential emotions both of the
nobles/aristocrats as well as their subjects (armed men, monks, inhab-
itants and so on). These are important in explaining aspects of social
life, providing illustrations of the writer’s statements on wealth and just-
ice, and explanations for the behaviour of people in these times. Hence,
in the observer voice, the references to affect relate predominantly to
observations on the subject matter at hand.
Finally, the mentor voice has the highest probability of authorial (covert)
affect, and a median probability of non-authorial affect. This text (from
Lectures on Electromagnetic Theory) is in fact very different from all the
other texts in terms of its overall style. Compare:

(14) As far as the interrelationship of electromagnetic quantities is

concerned Maxwell knew as much as we do today. He did not

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 205

actually suggest communication between continents with the aid

of geostationary satellites, but if he was taken now to a satellite
ground-station he would not be numbed with astonishment. If
we would give him half an hour to get over the shock of his resur-
rection he would quietly sit down with a piece of paper (the back

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

of a bigger envelope, I suppose) and would work out the relevant
design formulae.
[. . . ]
I could go on for a long time in praise of Maxwell. Unfortunately we
have little time for digressions however entertaining they might
be.
[. . . ]
The rest of the course will be concerned with the various solutions
of eqns (1.1)–(1.7). Isn’t this boring for an engineer?
Shouldn’t this be done by mathematicians or by computer pro-
grammers?
Not for the time being.
[. . .]
In order to have a unified view of the subject we have started
with Maxwell’s equations. It means a new approach but not a rad-
ical departure. The subject is still the same. You will be able to
see that the laws you love and cherish (Coulomb’s, Biot-Savart’s,
Snell’s, etc.) all follow from our eqns (1.1)–(1.7). The order of dis-
cussion will follow the traditional one: electrostatics first, followed
by steady currents, then we shall move on to slowly varying phe-
nomena, and reach finally the most interesting part, fast-varying
phenomena, exhibiting the full beauty of Maxwell’s wonderful
equations.
[. . .]
The choice of [formula] in the form of eqn (2.5) immediately
ensures that eqn (2.1) is satisfied: [formula], so we have less to
worry about.
[. . .]
Interestingly, the electric field does not depend on the actual
positions of the charges. (BRC, FEF 11–99)

In this text, there are many first and second person pronouns (I, we,
our, you) and question–answer sequences (underlined), which make the
text very interactive. Apart from the emotion terms relating to Maxwell,
which are used to illustrate a point, the other emotion terms are crucial
in construing the writer–reader relationship, referring either to the writer

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

206 Emotion Talk Across Corpora

(for example interestingly), the reader (for example boring for an engineer)
or both at the same time (we have less to worry about). Presumably, the
peculiar style of this text results from the transformation of what was
originally spoken academic discourse into written academic discourse.
I call this the mentor voice, because the writer seems to construe him-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

self very much as a guide, or mentor into the subject matter, someone
who is willing to share his ample experience with the students, and who
construes himself very much as part of a bigger research community to
which the students (will) belong.
In terms of the data analyzed here, five (or potentially six5 ) affective
keys have suggested themselves. However, affective keys are only reliably
identifiable by looking at a relatively large number of texts (Martin &
White 2005: 163), and can depend on the subject matter, discipline, and
on the type and genre of academic discourse. Clearly, more research into
different types of academic discourse, with the help of a bigger corpus,
is necessary (a) to confirm and refine the identified affective keys, and
(b) to add further affective keys.
To finish, a brief glance at affective stance in academic discourse. As
in the other registers, we can find clusters of affect types in different
stretches of text that construct a certain stance. Because of space con-
straints, I can only give one example here – text AS6, which concerns
urban poverty, and its effect on society. This is conceptualized in terms
of mainly negative emotions, and there is a preponderance of negative
security: disquiet terms relating to poverty and attendant consequences.
For example:

(15) The biggest single difficulty in drawing attention to urban poverty

is that it is not new, but simply – in some of its most worrying
manifestations – getting worse. Like the poor themselves, the inner
city has long been with us. Ever since the Industrial Revolution
created a mass urban society, the conditions of the poorest city
dwellers have given rise to anxiety among the better off. (BRC,
AS6 5–7)

The feelings in this file are also mostly general, relating to the public
or society as such. Table 6.6 gives a summary of emoters, affect types and
triggers. In this example the construal of personae in terms of affective
stance (here mostly the population, society or a general emoter), in fact
functions rhetorically to construct urban poverty in terms of its effect on
people. The emphasis is not so much on the emoters, but rather on the

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

Table 6.6 Affect types in AS6 (BRC)

Emoter Emotion Affect type Trigger

term
author (+ general) worrying security: disquiet poverty (in some of
its manifestations)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

the better off anxiety security: disquiet the conditions of the
poorest city dwellers
general outrage satisfaction: The inner city
displeasure
population impatience satisfaction: at the slackening
displeasure pace of social reform
population optimism security: trust that little more than
economic growth,
fuelled by technological
change, was needed to
remove the main causes
of urban deprivation
the poor in London, frustration satisfaction: plight of the inner cities,
Bristol, Birmingham, displeasure collapse of manufacturing
Liverpool industry, outlook for
unskilled looked bleak
the residents of the fear security: disquiet of attack on themselves
poorest areas or their property
general frightening security: disquiet an underclass (football
hooligans, muggers,
inner city rioters)
general alarm security: disquiet The spiral in drug abuse
and trafficking, with
direct consequences for
the AIDS epidemic
general fears security: disquiet that this particularly
dangerous substance
[crack] might become
widespread
public anxiety security: disquiet about urban squalor
political will inclination: desire to do something
general (people who happy happiness: cheer living-spaces
live there)
social discontent satisfaction: none [mood]
displeasure
government approach hopefully inclination: desire reducing the potential
for conflict among
the urban poor

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

208 Emotion Talk Across Corpora

trigger in terms of its emotional impact – displeasure and, predom-

inantly, disquiet. At the same time as characterizing the research object,
it also underlines the importance of the problem.

6.2 Investigating and problematizing affective key

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Instantiation and individuation. One possible research project into affec-
tive key would be to investigate how far the keys that were identified here
on the basis of a small corpus are relevant across a larger group of texts,
addressing the question of how we can generalize from instances to keys.
In terms of SFL theory, this problem can be discussed with respect to the
cline of instantiation (see Martin & White 2005: 23–5). Ultimately, future
research will show whether the identified keys will stand the test of time,
and the extent to which these affective co-selections recur across texts.
Another research project could address the relation between signatures
and keys. Signatures concern ‘syndromes of evaluation which charac-
terise an individual’ (Martin & White 2005: 203) rather than groups of
texts written by different individuals. In SFL, this involves the cline of
individuation (see Martin in press). Examining only affective patterns that
are present in one text, it is difficult to say whether they constitute signa-
tures or keys. Researchers investigating instantiation and individuation
would therefore need to look at a corpus of similar texts by different
authors, and a corpus of different texts by the same author to explore
how a particular text is alike or unlike other texts of the same register,
and how it is alike or unlike other texts by the same author.6

Key, patterns, and frequency. Another controversial issue is the narrow

definition of key adopted here. Martin & White (2005) speak of evalu-
ative keys in terms of all systems of appraisal (attitude, engagement,
graduation), meaning that the concept relates to patterning across the
taxonomy of appraisal. Affective key, on the other hand, relates only to
one of the sub-systems of attitude (affect), meaning that the concept
relates to patterning within a taxonomy. Thus, incorporating configur-
ations of patterns outside affect might suggest different sets of keys. This
issue relates to how we know which aspects of appraisal establish key. For
instance, when we consider purely frequency of emotion terms in news
reporting we find that Sports and Arts have more emotion terms per 1000
words than the texts in the other affective key set (with the exception
of Social reporting): see Table 6.7. This distribution is partly confirmed
when we look at 15 frequent emotion terms in the six categories of
reporting in the 2.6 million word BRC news reportage corpus (anxious,

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 209

Table 6.7 Number of emotion terms in news

reportage categories of the BRC baby

Category Number of emotion terms

raw frequency per 1000 words

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Arts 26 8.6
Social 16 8.2
Sports 23 7.5
Science 16 5.8
Commerce 18 5.6
Report 16 5.1

surprised, delighted, grateful, embarrassed, anxiety, optimism, humiliation,

happiness, regret (N), admire, anger (V), surprise (V), baffle, frighten), which
is as follows:

Social (573.9 ppm) > Sports (339.2 ppm) > Arts (338.2 ppm) > Report
(257.8 ppm) > Science (229.7 ppm) > Commerce (181.2 ppm).

That is, in terms of frequency different sets of texts suggest themselves:

Social – Sports/Arts – Report/Science – Commerce.
Comparing this with the establishment of keys in journalistic dis-
course, Martin & White (2005: 176–7) note that their reporter voice has
rates of authorial appreciation of between 0.9 and 6.3 per 500 words, and
their writer voice has rates of between 1.6 and 11.3 per 500 words. But if
some reporter voice texts have a rate as high as 6.3 and some writer voice
texts have a rate as low as 1.6, this means that some reporter voice texts
may contain more appreciation than writer voice texts. In other words,
if frequency of appreciation were taken as the distinguishing criterion,
different evaluative keys would be identified, as the texts would fall into
different sets.
The same is true to some extent for academic discourse. Looking at
the distribution of 15 frequent emotion terms (willing, anxious, worried,
surprised, disappointed, enthusiasm, reluctance, willingness, optimism, temp-
tation, worry (V), anger (V), dislike (V), admire, resent) in the different
disciplines represented in the larger BRC academic discourse sub-corpus,
emotion terms are distributed as follows:

Humanities (327.3 ppm) > Politics/law/education (277.9 ppm) > Social

and behavioural sciences (240 ppm) > Medicine (99 ppm) > Technol-
ogy/computing/engineering (58.3 ppm) > Natural sciences (31 ppm).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

210 Emotion Talk Across Corpora

These findings can be compared to Hyland’s (1999) study which

showed that attitude markers vary in frequency according to subject.
They were found to be least frequent in biology and physics, followed
by electrical engineering, mechanical engineering, and most frequent
in sociology, philosophy, marketing and applied linguistics. Hyland

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

concludes that ‘the distribution of stance [overall] shows a clear correl-
ation with the traditional distinction between hard and soft disciplines’
(Hyland 1999: 109). Hyland (2004) also finds that the soft sciences
express ‘far more explicitly personal stance’ (Hyland 2004: 18) compared
to the hard sciences. Even though Hyland’s attitude markers include
more than emotion terms, and the analysis of emotion terms includes
more than just the expression of (authorial) stance (cf. Section 1.5),
meaning that the results are not directly comparable, the similarities
are quite striking. The frequency distribution of emotion terms, like the
distribution of attitude markers, seems to correlate with the distinction
between hard and soft sciences, being most frequent in the humanities
and least frequent in the natural sciences.
In terms of frequency, then, six affective keys suggest themselves (cor-
responding to different disciplines), which only partly overlap with the
affective keys suggested above: see Table 6.8. Thus, the observer voice con-
tains both texts from the social sciences and the humanities, rather than
being associated with one discipline only. This is not surprising since
both have a common interest in social life, and its products, and analy-
ses of art will make reference to aspects of social life, as will historical
analyses.

Table 6.8 Affective keys and disciplines

Descriptor Illustrator Argumentor Observer Mentor ?

EWW, FPG: FC1, ACJ: ALP: A6U, AS6, EA7: FEF: HWV:
technology/ politics/law/ social social natural medicine
engineering education sciences sciences, sciences7
humanities

These findings for frequency of affect in journalistic and academic dis-

course throw up a question that future research will need to address:
should patterning or frequency of appraisal be taken as a basis for estab-
lishing key? A related question is how we know just which aspects of
appraisal actually establish key. Perhaps we need more sophisticated
techniques for more precise measurements, such as the factor analysis
applied by Biber and his colleagues (Section 1.5). It is also necessary to

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 211

find out about the minimum corpus size needed to reliably identify keys –
the BRC baby is certainly too small, but are the corpora used by Martin
& White (2005) and Coffin (2006) big enough? In other words, corpus-
based research could tackle these issues by compiling representative cor-
pora, analyzing the data according to all aspects of appraisal that could

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

be potentially relevant and using multivariate statistical techniques
to determine relevant configurations of appraisal. Precht (2000) does
research in conversation along these lines using the stance paradigm.

6.3 Functions of emotion terms

In this section I look more closely at the pragmatic functions of emotion

talk in the four registers of the BRC baby. This is not intended to be
exhaustive of possibilities in the four registers. All that is intended is an
illustration of some typical functions of emotion terms as evidenced in
the BRC baby.

Conversation. As is well-known, casual conversation can involve

moments of story-telling which are generically structured as narra-
tive (Eggins 2004: 74). In the context of story-telling, emotion terms
are employed by narrators to describe the emotive reaction of charac-
ters (including the speaker) in a narrative or by hearers to provide an
evaluative comment on it:

(16) [Narrative about a near-collision with a car]

AUDREY [288] A woman said, my God! [289] And it [the car]
came through on red Gordon! [290] You know where
the er, the er MacDonalds is on the corner?
GORDON [291] Yeah.
AUDREY [292] There! [293] I was crossing there. [294] . . . I
couldn’t believe it.[295] Everybody sort of, you know,
they were, they were just absolutely staggered! [296]
It frightened me to death! (BRC, KBC)

(17) [Narrative about evacuees]

GORDON [141] And then she was going out making money on
munitions.
AUDREY [142] Yes.
GORDON [143] So there’s a big income coming and no kids
to feed.
AUDREY [144] Yes. [145] Yeah.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

212 Emotion Talk Across Corpora

GORDON [146] And then of course, this was all blown

when . . . th the raids stopped for a quite a long
time, all these bloody kids came back!
AUDREY [147] Yes.
GORDON [148] And they didn’t really want them back at

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

that time.
AUDREY [149] Oh I know. [150] I know
GORDON [151] They were quite happy where they were
you see.
AUDREY [152] Mm. [153] . . . Yeah, it’s amazing really.
(BRC, KBC)

These two functions have attracted the attention of many linguists in

the past: the classical reference is Labov (1972) on evaluation in nar-
rative – Cortazzi & Jin (2000: 108) give an overview of the application
of his model in linguistics, and Bamberg (1997c) includes many rele-
vant contributions. Emotion terms belong to the evaluative resources
that story tellers use to show the point of a story (Labov 1972: 366) or
that story recipients employ to ‘show appreciation of story content or
narrative performance’ (Cortazzi & Jin 2000: 110). Even though evalu-
ation is concentrated in one section, it pervades the whole narrative
in ‘waves of evaluation’ (Labov 1972: 369). Similarly, Bamberg (1991,
1997a) discusses how references to emotion connect and frame narrative
episodes, providing a certain perspective on the narrated events. The
structuring function of emotion terms has also been noted by Precht,
who emphasizes that ‘the evaluation seems to be made just as often
by the listener as by the story teller’ (Precht 2003: 248). On evalua-
tion and narrative/textual structure compare also Fiehler (1990: 231),
Martin & Plum (1997), Martin (2001: 316, 2004a, 2004c), Macken-
Horarik (2003a), Reilly & Seibert (2003: 548), Galasiński (2004: 71), and
Hunston & Thompson (2000: 11, citing Sinclair 1987).
When conversation is more fluid, when ‘more open-ended struc-
tures . . . take over as the talk develops dynamically, with no clear end
point to be achieved, and few discrete steps or stage boundaries along the
way’ (Eggins 2004: 75), emotion terms can provide evaluations of trig-
gers that are offered in order to keep the conversation going (evaluations
other than affect are underlined in the example):

(18) ALAN [1916] Have you ever tried octopus?

NONE [1917] Erm . . . no.
BARRY [1918] Er . . . no I haven’t.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 213

NONE [1919] No.

BARRY [1920] I mean, squid I’ve eaten a lot of.
ALAN [1921] Bit like that innit?
[1922] Squid.
BARRY [1923] Yeah.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

ALAN [1924] I’ve had octopus soup.
[1925] That was quite . . . well . . . a weird taste.
BARRY [1926] I like shark.
[1927] Have you eaten shark?
ALAN [1928] No.
BARRY [1929] Oh I like shark, it’s really nice cos it . . . it’s
really meaty, I mean, it’s like eating meat (BRC, KBD)

This text shifts from Alan’s and Barry’s talk about and evaluation of octo-
pus and squid to Barry’s evaluation of shark, with both affective and
other evaluations ‘propelling’ the conversation (compare Martin 2000b:
31–4). The conversation continues with further talk about shark, a fish
shop, fish markets, cooking, octopus and related matters for another 128
utterances, with speakers offering various experiences and evaluations.
This demonstrates that, as Martin has put it, ‘appraisal telos . . . antici-
pates expansion, with respect to the attitudes available for negotiation (as
speakers align and individuate by way of negotiating solidarity relations’
(Martin 2000b: 38). It is not just authorial affect, but also non-authorial
affect that propels conversation, for instance questions about evaluations
by the listener or a third party (such as why does sea fishing interest you?).
Additionally, to enquire about someone’s emotions/evaluations natur-
ally means taking an active interest in that person, and can be regarded
as a positive politeness strategy. On the other hand, in example (18), the
inquiry is used to monitor the child’s progress at school, and is employed
very differently – showing the importance of social power, rank and sta-
tus with respect to emotion talk (see Poynton 1985: 78, Abu-Lughod &
Lutz 1990: 14, Irvine 1990: 128, Gallois 1994, Planalp 1999: 147, Precht
2003: 241):

(19) VALERIE [438] Did you get it [homework] all right?

[439] Was she pleased with you today? (BRC, KE4)

Some non-authorial affect is used jokingly or for teasing, contributing

to bonding. Note the laughter in the examples below:

(20) MARTINE [7251] The Tottenham game was really good

wasn’t it?

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

214 Emotion Talk Across Corpora

NONE 2 [7252] You let them down

MERIELLE [7253] Yeah
NONE 2 [7254] you, you were meant to come on and score
two goals and
MERIELLE [7255] Yeah

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

MARTINE [7256] You got the score
MARTINE [laugh] (BRC, KD8)

(21) MARTINE [7397] Oh you don’t want to have a Robin Hood

do you?
MIKE [7398] just shave that bit there and
MARTINE [7399] no, oh no,
MARTINE [7400] no, please don’t
MIKE [7401] Or just shave me tache off
MARTINE [7402] Oh
MIKE [laugh]
MARTINE [7403] I hate beards like that, I dislike that
MIKE [7404] No not to a point
MARTINE [7405] I know, but sort of none there and just all
sort of down there, like sort of Brian
MARTINE [7406] That’s no good
MIKE [7407] sticking out of his head
MARTINE [7408] [laugh] . . . oh . . . (BRC, KD8)

Other interpersonal resources such as mood (negation) and apprecia-

tion/judgement clearly contribute to the teasing at play here. Research
into appraisal and humour is still in its infancy, but see Eggins &
Slade (1997) for exploratory research, and Knight (2007) for an ongoing
research project into appraisal, humour and bonding.
On a more local level emotion terms are used in association with
cause–effect relations, often providing explanations for behaviour or
justifications of decisions (on the importance of causal explanation for
comprehension see O’Halloran 2003: 208–11):

(22) GORDON [67] She said we always sleep together

AUDREY [laugh]
GORDON [68] sister and me.
AUDREY [69] Yeah.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 215

GORDON [70] She said, why do you do that? [71] Says, well . . .
if I wake up in the night and my sister’s not there
I’ll think someone’s taken her or she’s run away,
I’ll be frightened. [72] And the same applies
the other way so we always stick together. (BRC, KBC)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Investigating just what kind of causal connections are made with emo-
tion terms is an interesting and fruitful area of research (see Bednarek
forthcoming). Such references show exactly what kind of behaviour can
be justified by making reference to emotional experience, and also reflect
the kind of folk construal of emotional behaviour that was referred
to in terms of emotion schema in Chapter 5 (with respect to causal
antecedents). The importance for affect in cause–effect relations also con-
firms findings by Painter regarding mother–child talk that ‘apparently
impersonal areas such as causal relations and generalizations arise initially
from the impetus to share “attitude”’ (Painter 2003: 183, italics in original)
as well as findings by Edwards (1999) on emotion discourse and rational
accountability in relationship counselling sessions. It also supports the
claims made by the philosopher Errol Bedford in the 1950s that emotion
terms can be used to explain and justify behaviour or actions (Bedford
1956/57: 303).8

News reportage. I have already discussed the functions of emotion terms

in news reportage to some extent elsewhere, and shall not comment on
them in detail here. In Bednarek (2006a: 156–60, 170–8) I looked at the
discourse functions of lexical items referring to news actors’ emotion
and volition in British news stories, examining what is essentially non-
authorial affect in the appraisal framework (though appraisal theory was
not applied). Generally speaking, such references can be used to evaluate,
to emotionalize, and to dramatize. The main functions identified for
references to EMOTION and VOLITION in Bednarek (2006a) were:

• evaluating the content of attributed propositions as positive/negative

• evoking the reader’s emotion (see also Ungerer 1997)
• evoking the reader’s interest, appealing to their emotional voyeurism
(for example in headlines and other prominent positions)
• shifting blame (with VOLITION)
• triggering positive/negative evaluation
• enhancing news value (in particular negativity, colour, unexpected-
ness and facticity)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

216 Emotion Talk Across Corpora

In terms of authorial affect, White notes that

AFFECT strongly foregrounds authorial inter-subjectivity when it is the

author’s own emotional responses which are being presented. For
authors to describe their own anger, fear, sadness, boredom etc is

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

clearly to inject their own subjectivity into the text. (White 1998: 272)

Authorial affect, when it occurs in the news data, seems mostly to

express a personalized opinion on the part of the news writer, hence
introducing an unwanted subjectivity in a purportedly objective type of
discourse. This explains why such references are rare in news discourse
in general (Martin & White 2005: 177) – in particular, overt rather than
covert authorial affect, as shown above (Table 6.5).

Fiction. As suggested in Section 4.3, emotion terms in fiction provide

characterizations, create imagery, cause empathy, and establish point
of view. The construal of characters can lead us to like or dislike them,
empathize with them or not, identify with them or not. This has been
commented on in terms of reader involvement, engagement and affin-
ity (Watson 1999), identification, empathy (Dijkstra et al. 1994: 142,
Macken-Horarik 2003a: 286), and the author’s ‘play of sympathy and
antipathy’ (Martin & White (2005: 73).9
Books of fiction can be considered as ‘macro-genres’ (Martin & Rose
2003: 209) that comprise different types of genres. One possibility for
linguistic analysis is hence to investigate the function of emotion terms
with respect to logogenesis in the generic structure. However, space does
not allow such an analysis here. The perspective that is adopted in this
section is therefore local rather than global, considering the immediate
functions of emotion terms in text passages without taking into account
the larger co-text of the chapter or book.
Firstly, emotion terms in dialogue and in descriptive passages can
construe the emotional relationship between characters:10

(23) ‘I’ve missed you so much, this time,’ she told him breathlessly.
‘Can’t think why.
Really forgotten what you look like.
Let’s have a look at you.
Hum, not bad, considering your age and everything I suppose. . . .’
‘It s great to be home, darling.
I’ve missed you too,’ he whispered in her ear.
‘Don’t know why I feel this way.
Must be Spring!’ (BRC, AC2 35–43)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 217

As readers, we are invited to incorporate this affective relationship into

our ongoing mental construction of the narrative, and to ‘care’ about the
characters – whether through positive or negative emotions.
Emotion terms can also provide readers with a character’s affective
perspective on an entity or a proposition. This affective perspective may

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

either be long-term (shading into judgement as a construal of an emoter’s
character values) or momentary (compare news reportage above):

(24) Mark always experienced the same feelings of contentment when

entering the home straight. Today he felt particularly pleased to
be back. (BRC, AC2 21–22)

How such construals of temporary or long-term affective states cre-

ate emotional responses such as empathy or identification or evaluative
reactions in readers may depend on the personality of the reader, but is
also associated with cultural norms. For instance, it would be hard for
any reader to adopt a positive perspective on O’Reilly in the following
example:

(25) O’Reilly was impossible to please and suspicious of everybody

and everything. He trusted nobody, not even fellow Catholics on
the Board. (BRC, AC2 85–86)

Depending on affective key (see Section 6.1 above), emotion terms can
also be part of descriptions, with an emphasis on the emotional impact of
the described entity on a given character or in general, and also creating
a certain atmosphere or ambience:

(26) And now, for the first time, he thought that he could smell the
North Sea, that potent but half-illusory tang evoking nostalgic
memories of childhood holidays, of solitary adolescent walks as
he struggled with his first poems, of his aunt’s tall figure at his
side, binoculars round her neck, striding towards the haunts of her
beloved birds. And here, barring the road, was the familiar old farm
gate still in place. Its continued presence always surprised him
since it served no purpose that he could see except symbolically
to cut off the headland and to give travellers pause to consider
whether they really wanted to continue. (BRC, C8T 4–6)

This can also occur in descriptions of characters. For instance, in the

following passage, a female character is described through the eyes of

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

218 Emotion Talk Across Corpora

a male character (the focalizer; compare the underlined verbs that refer
to cognition and perception), providing an evaluation of that character
and contributing to imagery, giving the reader a mental picture from the
focalizer’s point of view:

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(26) It was a distinguished face with the deep-set, widely spaced eyes
beneath straight brows, a well-shaped, rather secretive mouth and
strong greying hair swept upwards and curled into a chignon. In
her publicity photographs she could, he recalled, look beautiful in
a somewhat intimidating, intellectual and very English mould.
But seen face to face, even in the informality of her own house,
the absence of a spark of sexuality and, he sensed, a deep-seated
reserve, made her seem less feminine and more formidable than
he had expected, and she held herself stiffly as if repelling invaders
of her personal space. The handshake with which she had greeted
him had been cool and firm and her brief smile was surprisingly
attractive. He knew that he was oversensitive to the timbre of
the human voice and hers, although not jarring or unpleasant,
sounded a little forced as if she were deliberately speaking at an
unnatural pitch. (BRC, C8T 34–38)

Describing the inner life of characters, then, helps to engage us emo-

tionally, to make us care. Where this is not the case, a character can
remain ‘a shadowy figure’ or ‘less than fully human’ (Robinson 2005:
217). Descriptions of characters’ emotions have moral connotations of
virtue and vice (van Meel 1994: 162), and conceptualize characters as
particular types of persons (Dijkstra et al. 1994: 142).

Academic discourse. I am only going to list some important functions of

emotion terms in academic discourse together with an illustrative exam-
ple, since they have already been commented on above in the analysis
of affective key:

• Providing explanations: It is possible to process the records in a directly

organized file either directly, using the keys in any desired order;
• Introducing a problem: A bank cannot predict which of its customers wish
to withdraw or deposit money on any given day . . .. As these transactions
are not predictable they present a processing problem;
• Illustrating a point/argument: A fairly typical set of facts is provided by
Nedrick (1986), where D had a grudge against a woman;

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 219

• Offering reasoning (cause-effect relations): Aware of practitioners’ frus-

tration in not being able to work effectively with these problematic clients,
Probation management [. . . ] determined to develop an Alcohol Education
Course;
• Observing aspects of human life: This article comes out of the familiar

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

experience of being drawn to a particular image, or set of images, without
at first knowing why;
• Construing the writer/reader relationship and writer/reader personae:
Isn’t this boring for an engineer? Shouldn’t this be done by mathematicians
or by computer programmers? Not for the time being;
• Construing an argument: In the sections to follow the stark question is
therefore put ‘Is the Probation Service in the business of inflicting pain?’
and the answer which unfolds maybe summarized ‘No, because it neither
aspires nor wishes to!’;
• Construing a research topic, often in terms of its impor-
tance/relevance: For this reason, the variable and often unsatisfactory
seroconversion rates in developing countries after routine administration of
OPV are of concern.

It is not affect alone that is associated with these functions; for

instance, question–answer sequences are just one of a number of lin-
guistic features that interact with affect to produce these effects. Since
much research into evaluation in English for Academic Purposes exists
(compare Section 5.1), no more will be said about this here. There is
no doubt, however, that research into affect needs to take into account
different genres and registers (Swales 2004, Biber 2006) of academic
discourse.

6.4 A road-map for research on affect

The proposed road-map for research on affect mentioned at the begin-

ning of this chapter aims at a three-pronged analysis for each genre,
text type or register, such that the study of, say, affect in news reviews
would involve a large-scale quantitative corpus analysis, a small-scale
corpus analysis, and a qualitative analysis of one or a few texts, with
each analysis having certain foci, advantages and disadvantages. Com-
pare Table 6.9. This extends previous studies in linguistics that involve
both corpus and discourse analysis, for instance corpus-based dis-
course analyses or Matthiessen’s ‘two-pronged approach’ (Matthiessen
2006: 110).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

220 Emotion Talk Across Corpora

Table 6.9 A three-pronged approach

Large-scale corpus analysis Small-scale corpus analysis Analysis of one text

frequency, distribution key, stance, functions of logogenesis, socio-
patterns, hypothesized emotion terms, patterns with cultural analysis, co-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

general functions respect to co-articulated articulated meanings
meanings
most representative less representative least representative
less detailed description, more detailed description, most detailed
more descriptive more interpretive description, most
interpretive

It is evident that the focus of this book has been predominantly on

large- and small-scale corpus analysis rather than qualitative analysis,
and that only certain aspects of affect could be investigated. Summing
up, we need bigger corpora to investigate affect in different genres,
text types, and registers (also taking into account intra-register/intra-
genre/intra-text type variation), to explore instantiation and individu-
ation (keys vs. signatures), and to examine the polyfunctionality and
‘flexibility’ (Edwards 1999: 281) of emotion talk. We also need studies
that involve more variables of affect (and other appraisal sub-systems)
and examine the co-selection and patterning of these variables, as well as
qualitative studies that take into account logogenesis and social context.
The studies reported on in this chapter are therefore intended to serve
as a stepping stone to future research which elaborates on, confirms or
contradicts these findings.

6.5 Envoi

Chapter 6 has taken a brief foray into the study of the pragmatics of
emotion terms, outlining projects and challenges for future research. At
a more general level, and echoing what Martin & White (2005: 260) note
for the study of evaluation, research into affect needs to find the right
balance between corpus-linguistic (quantitative) and discourse analytical
(qualitative) research, as well as research that is situated somewhere in-
between (such as small-scale corpus analysis). Another challenge is the
combination of both social and cognitive approaches to the language of
emotion, as we need to reach beyond the limits of any single analytical
perspective. If emotion is at the same time rooted in biology (in the
mind) as well as socio-culturally construed, we arguably need to look
for an approach that can bring out both – providing a kaleidoscopic

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 221

lens which fractures perception, and simultaneously shows us different

aspects of emotion talk.
This book is intended to make a contribution to the combination
of corpus and discourse-analytic as well as functional and cognitive
research. As such, the analyses might be useful for:11

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

• the application of appraisal theory;
• the modelling of probabilistic register variation;
• lexicography and language teaching;
• natural language processing (automated register recognition, parsing
of affect).

The analyses have also offered ways of seeing what we get up to

when we use emotion terms in language, whether in conversation, news
reportage, fiction or academic discourse. From this examination, it does
not seem to be strictly true that ‘how we talk about emotion . . . is
a question analogous to how we talk about nature, about the body,
about teaching’ (Weigand 2004b: 11). Rather, emotion talk tells us much
more about human interaction, about power relations, about social and
cultural norms – about human life as such.

Notes
1. In order not to skew the analysis in Table 6.3 one article was excluded from
Social reporting for these frequency calculations, as it reports French emo-
tions/attitudes towards clothes/laundry, and therefore has as many as 26
emotion terms (more than all the other articles in this category combined).
Nevertheless, I have analyzed this article, and comment on it below.
2. I make no distinction within authorial affect between mediated/unmediated
affect because news writers rarely quote themselves. This may be different in
other text types, for example academic discourse.
3. The small girl in the first sentence seems to be the emoter here with the writer
adopting her perspective; hence this is classified as covert unmediated affect.
4. In Table 6.5, the terms in brackets occur in one text only, A1G. A1G clearly
differs from all other texts in the Report category in including both covert
and overt authorial affect (as well as first person references to the writer) – it
is perhaps misclassified. The text, given below in example (i) deals with the
Siberian crane, and concludes with a clear evaluation on the part of the writer,
making it more of a commentator voice text (affect in bold; other evaluations
underlined):
(i) Given the awesome sum of human misery in this part of the world,
I suppose I shouldn’t get too worked up about the fate of a flock of
birds — particularly as the much larger eastern race of the Siberian crane,
which migrates from north-eastern Yakutia to the lower reaches of the

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

222 Emotion Talk Across Corpora

Yangtse river in China, is apparently surviving well. But if the snow-

wreaths of the Ob river don’t reach Bharatpur this year, I shall feel
inescapably sad. (A1G 128–129)

5. One of the files, HWV, does not easily fit in any of the keys (though it is

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

somewhat similar to the descriptor voice in its predominant absence of affect).
It contains only two instances of covert authorial affect, but no non-authorial
affect, and constitutes perhaps an affective key in itself. In this key, author-
ial covert affect is used only in the discussion section, here outlining the
significance of the problem:
(ii) . . . For this reason, the variable and often unsatisfactory seroconver-
sion rates in developing countries after routine administration of OPV are
of concern, especially in tropical areas where wild poliovirus infection
remains endemic. (HWV 82)

As Hood points out, negative emotions can be used to construe research topics
as problematic, and worthy of research (Hood 2006: 41).
6. There is not yet much research in SFL on instantiation and individuation
(Martin in press), and the relation between these clines cannot be discussed
here. Generally speaking, ‘instantiation interprets the relation of system
to instance’ (Martin in press) whereas ‘[i]ndividuation interprets the rela-
tion of system to individual’ (Martin in press). Watson (1999) has shown
that there may be variation in affect even between novels by the same
author.
7. This is not to imply that the mentor voice is the only voice present in the
natural sciences; rather, it is the voice that is adopted in the random sample
from natural sciences investigated here. I suspect that most natural sciences
texts in fact make use of the descriptor voice, and that the mentor voice has
more to do with lecturing than written academic discourse.
8. Additionally, emotion terms relating to dis/inclination – what has been
described as boulomaic modality (for example Simpson 1993) – are frequently
associated in conversation with making plans and offers (see also Precht 2003:
250). Desire terms are clearly important with reference to goals, which are vital
for the regulation of human life and well-being, as social cognitive psychol-
ogists have shown (Pishwa 2005). Desire terms are also often used to express
offers (Do you want some help with it; KD0), orders (plus you want the archiving
doing, the A three drawer doing [= reported order]; KD8), and demands (Have
you got two tens you want to change for a twenty Paul?; KD0). In many such
examples want is simply used as part of a conversational formula. Compare
Werth (1998) and Precht (2000: 137, 139) on emotion terms and convention-
alized formulae (examples are I’m afraid, I’m sorry, to my regret, and so on).
(In SFL terms, some of them can be described as interpersonal grammatical
metaphor; see Martin 2000b.) Terms of volition can also be used to evaluate
a trigger (clausally realized in the following example) positively or negatively
on the part of the emoter:
(iii) DAVID [472] Don’t want to go.
VALERIE [473] Well you’ll have to get ready for going. (KE4)
→ ‘I go’ = bad

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Enacting Affect: Pragmatic Analysis 223

This is because saying that you desire an event to happen implies that,
from your perspective, it is good if this event actually happens, and vice
versa. Desire terms can also provide more indirect evaluations of entities
in terms of the fulfilment or non-fulfilment of the emoter’s goals (Bednarek
forthcoming).
9. Emotion terms are also associated with suspense (cf. Dijkstra et al. 1994: 142)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

and provide clues to the reader (in crime fiction). Emotion terms that are part
of dialogue in fiction arguably mimic conversation and have similar func-
tions, which I will not discuss here again. It would be interesting, however,
to compare fictive dialogue in detail with real-life conversation.
10. Compare also Martin & White (2005: 89). Emotion terms can also construe
emotional relationship of characters towards animals, as in file AC2.
11. These applications are discussed more extensively in Appendix A 6.4 online.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

References

Abu-Lughod, L. & C.A. Lutz 1990. ‘Introduction: emotion, discourse, and the
politics of everyday life’. In Lutz & Abu-Lughod (eds): 1–23.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Adendorff, R. & V. de Klerk 2005. ‘The role of APPRAISAL resources in constructing
a community response to AIDS’. Linguistics and the Human Sciences 1/3: 489–513.
Agresti, A. 1990. Categorical Data Analysis. New York: John Wiley & Sons.
Anderson, K.J. & C. Leaper 1998. ‘Emotion talk between same- and mixed-gender
friends. Form and function’. Journal of Language and Social Psychology 17: 419–48.
Arndt, H. & R.W. Janney 1987. InterGrammar. Towards an Integrative Model of Verbal,
Prosodic and Kinesic Choices in Speech. Berlin: Mouton de Gruyter.
Athanasiadou, A. & E. Tabakowska (eds) 1998a. Speaking of Emotions. Conceptuali-
sation and Expression. Berlin: Mouton de Gruyter.
Athanasiadou, A. & E. Tabakowska 1998b. ‘Introduction’. In Athanasiadou &
Tabakowska (eds): xi–xxii.
Bamberg, M. 1991. ‘Narrative activity as perspective taking: the role of emotionals,
negations and voice in the construction of the story realm’. Journal of Cognitive
Psychotherapy 2: 275–90.
Bamberg, M. 1997a. ‘Emotion talk(s): the role of perspective in the construction
of emotions’. In Niemeier & Dirven (eds): 209–26.
Bamberg, M. 1997b. ‘Language, concepts, and emotions. The role of
language in the construction of emotions’. Available to download at
http://www.massey.ac.nz/∼alock/virtual/bamberg.htm (last accessed 16 August
2006).
Bamberg, M. (ed.) 1997c. Oral Versions of Personal Experience: Three Decades of
Narrative Analysis. Special issue of Journal of Narrative and Life History 7.
Barnbrook, G. & J. Sinclair 1995. ‘Parsing COBUILD entries’. In Sinclair, J.,
Hoelter, M. & C. Peters (eds) The Languages of Definition: The Formalization of
Dictionary Definitions for Natural Language Processing. Studies in Machine Trans-
lation and Natural Language Processing, Vol. 7. Luxembourg: Office for Official
Publications of the European Communities: 13–58.
Bedford, E. 1956–1957. ‘Emotions and statements about them’. Proceedings of the
Aristotelian Society LVII: 77–89.
Bednarek, M. 2005a. ‘Frames revisited: the coherence-inducing function of
frames’. Journal of Pragmatics 37: 685–705.
Bednarek, M. 2005b. ‘Construing the world: conceptual metaphors and event
construals in news stories’. Metaphorik.de 09/2005. Available to download at
http://www.metaphorik.de/09/bednarek.pdf (last accessed 15 January 2007).
Bednarek, M. 2006a. Evaluation in Media Discourse. Analysis of a Newspaper Corpus.
London/New York: Continuum (Research in Corpus and Discourse).
Bednarek, M. 2006b. ‘Subjectivity and cognition. Inscribing, evoking and pro-
voking opinion’. In Pishwa, H. (ed.) Language and Memory. Berlin/New York:
Mouton de Gruyter: 187–221.
Bednarek, M. forthcoming. ‘Emotion-talk and emotional-talk: cognitive and
discursive perspectives’. In Pishwa. H. (ed.) Language and Social Cognition.
Berlin/New York: Mouton de Gruyter.

224

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 References 225

Bednarek, M. forthcoming 2008. ‘Language Patterns and ATTITUDE’. Functions of

Language.
Belfrage, B. 1986. ‘Berkeley’s theory of emotive meaning (1708)’. History of
European Ideas 7: 643–9.
Bell, A. 1991. The Language of News Media. Oxford: Blackwell.
Bellelli, G. 1995. ‘Knowing and labelling emotions’. In Russell et al. (eds):

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

491–504.
Besnier, N. 1990. ‘Language and affect’. Annual Review of Anthropology 19:
59–88.
Besnier, N. 1994. ‘Involvement’. In Asher, R.E. (ed.) The Encyclopedia of Language
and Linguistics. Oxford: Pergamon: 1770–2.
Biber, D. 1988. Variation across Speech and Writing. Cambridge: CUP.
Biber, D. 2006. University Language. A Corpus-based Study of Spoken and Written
Registers. Amsterdam/Philadelphia: John Benjamins.
Biber, D. & E. Finegan 1989. ‘Styles of stance in English: lexical and grammatical
marking of evidentiality and affect’. Text 9: 93–124.
Biber, D., Conrad, S. & R. Reppen 1998. Corpus Linguistics: Investigating Language
Structure and Use. Cambridge/New York: CUP.
Biber, D., Johansson, S., Leech, G., Conrad, S. & E. Finegan 1999. Longman
Grammar of Spoken and Written English. London: Longman.
Black, M. 1948. ‘Some questions about emotive meaning’. The Philosophical Review
57: 111–26.
Bloom, K., Garg, N. & S. Argamon 2007. ‘Extracting appraisal expressions’. Pro-
ceedings of HLT/NAACL 2007, April 2007, Rochester, NY. Available to download
at http://lingcog.iit.edu/doc/bloom_naacl2007.pdf.
Bondi, M. & A. Mauranen (eds) 2003. Special issue of Journal of English for Academic
Purposes 2 on ‘Evaluation in Academic Discourse’.
Braun, C. 1992. ‘A note on the effect of semantic anomaly on the intensity of
emotional impact of metaphors’. Metaphor and Symbolic Activity 7: 1–10.
Bublitz, W. 2002. ‘Anmerkungen zur emotiven Prosodie des Textes’. In Bublitz,
W., von Roncador, M. & H. Vater (eds) Philologie, Typologie und Sprachstruktur.
Festschrift für Winfried Boeder zum 65. Geburtstag. Frankfurt: Peter Lang: 25–40.
Bublitz, W. 2003. ‘Emotive prosody: how attitudinal frames help construct con-
text’. In Mengel, E., Schmid, H.-J. & M. Steppat (eds) Anglistentag 2002,
Bayreuth – Proceedings. Trier: Wissenschaftlicher Verlag Trier: 381–92.
Bühler, K. 1934. Sprachtheorie. Stuttgart: Gustav Fischer. [2 1965]
Burger, L. & P.J. Miller 1999. ‘Early talk about the past revisited: affect in
working-class and middle-class children’s co-narrations’. Journal of Child Lan-
guage 26: 133–62.
Busse, W. (ed.) 1991. Anglistentag 1991. Proceedings of the Conference of the German
Association of University Professors of English, Vol. XIII. Tübingen: Max Niemeyer
Verlag.
Butler, C. 2004. ‘Corpus studies and functional linguistic theories’. Functions of
Language 11: 147–86.
Caffi, C. & R.W. Janney 1994. ‘Toward a pragmatics of emotive communication’.
Journal of Pragmatics 22: 325–73.
Caple, H. forthcoming. ‘Intermodal relations in image nuclear news stories’.
In Unsworth, L. (ed.) Multimodal Semiosis: Functional Analysis in Contexts of
Education. London: Continuum.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

226 References

Chafe, W. 1982. ‘Integration and involvement in speaking, writing, and oral lit-
erature’. In Tannen, D. (ed.) Spoken and Written Language. Norwood, NJ: Ablex:
35–53.
Charleston, B. M. 1960. Studies on the Emotional and Affective Means of Expression
in Modern English. Bern: Francke.
Christie, F. & J.R. Martin (eds) 1997. Genres and Institutions: Social Processes in the

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Workplace and School. London: Cassell.
Čmejrková, S. 2004. ‘Emotions in language and communication’. In Weigand
(ed.): 33–53.
Coffin, C. 2006. Historical Discourse: the Language of Time, Cause and Evaluation.
London: Continuum.
Coffin, C. & K. O’Halloran 2005. ‘Finding the global groove: theorising and
analysing dynamic reader positioning using APPRAISAL, corpus, and a concor-
dancer’. Critical Discourse Studies 2/2: 143–63.
Coffin, C. & K. O’Halloran 2006. ‘The role of Appraisal and corpora in detecting
covert evaluation’. Functions of Language 13/1: 77–110.
Conrad, S. & D. Biber 2000. ‘Adverbial marking of stance in speech and writing’.
In Hunston & Thompson (eds): 56–73.
Cortazzi, M. & L. Jin 2000. ‘Evaluating evaluation in narrative’. In Hunston &
Thompson (eds): 102–20.
Dalgleish, T. & M. Power (eds) 1999. Handbook of Cognition and Emotion. Sussex:
John Wiley & Sons.
Daneš, F. 1987. ‘Cognition and emotion in discourse interaction’. Preprints of the
Plenary Session Papers, XIV International Linguistic Congress Berlin (East), 10–15
August 1987: 272–91.
Daneš, F. 1994. ‘Involvement with language and in language’. Journal of Pragmatics
22: 251–64.
Daneš, F. 2004. ‘Universality vs. culture-specificity of emotion’. In Weigand (ed.):
23–32.
Davidson, R.J., Scherer, K.R. & H. Hill Goldsmith (eds) 2003. Handbook of Affective
Sciences. Oxford: OUP.
De Beaugrande, R. 1992. ‘Topicality and emotion in the economy and agenda of
discourse’. Linguistics 30: 243–65.
Dem’jankov, V., Sergeev, A., Sergeeva, D. & L. Voronin 2004. ‘Joy, astonishment
and fear in English, German and Russian. A corpus-based contrastive-semantic
analysis’. In Weigand (ed.): 163–78.
Dijkstra, K., Zwaan, R.A., Graesser, A.C. & J.P. Magliano 1995. ‘Character and
reader emotions in literary texts’. Poetics 23: 139–57.
Dirven, R. 1997. ‘Emotions as cause and the cause of emotions’. In Niemeier &
Dirven (eds): 55–86.
Dorfmüller-Karpusa, K. 1990. ‘Intensity markers’. Journal of Pragmatics 14:
476–83.
Downes, W. 2000. ‘The language of felt experience: emotional, evaluative and
intuitive’. Language and Literature 9: 99–121.
Dry, H. 1992. ‘Evidentiality and affect in the prose of Hanna Arendt’. Paper pre-
sented at International Association of Literary Semantics. Canterbury, England.
Dunning, T. 1993. ‘Accurate methods for the statistics of surprise and coinci-
dence’. Computational Linguistics 19/1: 61–74.
Edwards, D. 1999. ‘Emotion discourse’. Culture & Psychology 5/3: 271–91.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 References 227

Eggins, S. 2 2004. An Introduction to Systemic Functional Linguistics. London/New

York: Continuum.
Eggins, S. & D. Slade 1997. Analysing Casual Conversation. London/Washington:
Cassell.
Ekman, P. 1992. ‘An argument for basic emotions’. Cognition and Emotion 6:
169–200.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Ekman, P. 1997. ‘Should we call it expression or communication?’. Available to
download at http://www.paulekman.com/pdfs/should_we_call_it_exp.pdf (last
accessed 14 July 2006). [Published in Innovations in Social Science Research 10/4:
333–44.]
Ekman, P. 1999a. ‘Basic emotions’. Available to download at http://www.
paulekman.com/pdfs/basic_emotions.pdf (last accessed 13 July 2006).
[Published in Dalgleish & Power (eds).]
Ekman, P. 1999b. ‘Facial expressions’. Available to download at http://www.
paulekman.com/pdfs/facial_expression.pdf (last accessed 13 July 2006).
[Published in Dalgleish & Power (eds).]
Ellsworth, P. C. & K. Scherer 2003. ‘Appraisal processes in emotion’. In Davidson
et al. (eds): 572–95.
Enfield, N. J. & A. Wierzbicka (eds) 2002. Special issue of Pragmatics and Cognition
10 on ‘The Body in Description of Emotion’.
Fiehler, R. 1990. Kommunikation und Emotion. Berlin: Mouton de Gruyter.
Fiehler, R. 2002. ‘How to do emotions with words: emotionality in conversation’.
In Fussell (ed.): 79–106.
Fillenbaum, S. & A. Rapoport 1971. Structure in Subjective Lexicon. New
York/London: Academic Press.
Fillmore, C.J. 1997. ‘Idiomaticity’. Available to download at http://www.icsi.
berkeley.edu/∼kay/bcg/lec02.html (last accessed 14 September 2006).
Finegan, E. 1995. ‘Subjectivity and subjectivisation: an introduction’. In
Stein, D. & S. Wright (eds) Subjectivity and Subjectivisation. Linguistic Perspectives.
Cambridge: CUP: 1–15.
Foley, G. & S. Hood 2006. ‘Exploring the dynamic flow of emotions in customer
call centre communication’. Paper presented at ALLE (Applied Linguistics and
Language in Education) conference, 21 September 2006, Macquarie University,
Sydney, Australia.
Foolen, A. 1997. ‘The expressive function of language: towards a cognitive
semantic approach’. In Niemeier & Dirven (eds): 15–23.
Fowler, R.G. 1991. Language in the News: Discourse and Ideology in the Press. London:
Routledge.
Francis, G., Hunston, S. & E. Manning (eds) 1996. Collins Cobuild Grammar Patterns
1: Verbs. London: HarperCollins.
Francis, G., Hunston, S. & E. Manning (eds) 1998. Collins Cobuild Grammar Patterns
2: Nouns and Adjectives. London: HarperCollins.
Fries, N. 1995. ‘Emotionen in der semantischen Form und in der konzeptuellen
Repräsentation’. In Kertész, A. (ed.) Sprache als Kognition – Sprache als Interaktion:
Studien zum Grammatik-Pragmatik-Verhältnis. Frankfurt: Peter Lang: 139–81.
Frijda, N. H., Markam, S., Sato, K. & R. Wiers 1995. ‘Emotions and emotion words’.
In Russell et al. (eds): 121–43.
Fussell, S.R. (ed.) 2002. The Verbal Communication of Emotions: Interdisciplinary
Perspectives. Mahwah N.J.: L. Erlbaum.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

228 References

Galasiński, D. 2004. Men and the Language of Emotions. Houndmills: Palgrave

Macmillan.
Gallois, C. 1994. ‘Group membership, social rules, and power: a socio-
psychological perspective on emotional communication’. Journal of Pragmatics
22: 301–24.
Galtung, J. & M.H. Ruge 1965. ‘The structure of foreign news’. Journal of Peace

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Research 2/1: 64–91.
Garside, R. 1987. ‘The CLAWS word-tagging system’. In Garside, R., Leech, G. &
G. Sampson (eds) The Computational Analysis of English. Harlow: Longman
30–41.
Goldberg, B. 1971. ‘The linguistic expression of feeling’. American Philosophical
Quarterly 8: 86–92.
Goldshmidt, O.T. & L. Weller 2000. ‘Talking emotions: gender differences in a
variety of conversational contexts’. Symbolic Interaction 23: 117–34.
Gonzálvez-García, F. & C. S. Butler 2006. ‘Mapping functional-cognitive space’.
Annual Review of Cognitive Linguistics 4: 39–96.
Goodwin M.H. & C. Goodwin 2000. ‘Emotion within situated activity’. In
Duranti, A. (ed.) Linguistic Anthropology. A Reader. Oxford: Blackwell: 239–57.
Gordon, S.L. 1990. ‘Social structural effects on emotions’. In Kemper, T.D. (ed.)
Research Agendas in the Sociology of Emotions. Albany: State University of New
York Press: 145–79.
Gottman, J., Fainsilber Katz, L. & C. Hooven 1996. ‘Parental meta-emotion phi-
losophy and the emotional life of families: theoretical models and preliminary
data’. Journal of Family Psychology 10/3: 243–68.
Grondelaers, S. & D. Geeraerts 1998. ‘Vagueness as a euphemistic strategy’. In
Athanasiadou & Tabakowska (eds): 357–74.
Gross, M. 1993. ‘Local grammars and their representation by finite automata’.
In Hoey, M. (ed.) Data, Description, Discourse: Papers on the English Language in
Honour of John McH. Sinclair. London: HarperCollins: 26–38.
Günther, S. 1997. ‘The contextualization of affect in reported dialogues’. In
Niemeier & Dirven (eds): 247–76.
Györi, G. 1998. ‘Cultural variation in the conceptualisation of emotions: A
historical study’. In Athanasiadou & Tabakowska (eds): 99–124.
Halliday, M.A.K. & C.M.I.M. Matthiessen 1999. Construing Experience through
Meaning: a Language-based Approach to Cognition. London: Cassell.
Halliday, M.A.K. & C.M.I.M. Matthiessen 2004. An Introduction to Functional
Grammar. London: Arnold [3rd edition].
Harkins, J. & A. Wierzbicka (eds) 2001. Emotions in Cross-linguistic Perspective.
Berlin: Mouton de Gruyter.
Harré, R. (ed.) 1986 . The Social Construction of Emotions. Oxford: Basil Blackwell.
Harré, R. & W. Parrott (eds) 1996. The Emotions. Social, Cultural and Biological
Dimensions. London: Sage.
Haviland, J.B. 1991. ‘ “Sure, sure”: evidence and affect’. Text 9: 27–68.
Heelas, P. 1986. ‘Emotion talk across cultures’. In Harré (ed.): 234–66.
Hoey, M. 2006. ‘Language as choice: what is chosen?’. In Thompson & Hunston
(eds): 37–54.
Hood, S. 2005. ‘What is evaluated and how in academic research writing?: the
co-patterning of attitude and field’. Australian Review of Applied Linguistics (Series
S) 19: 23–40.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 References 229

Hood, S. 2006. ‘The persuasive power of prosodies: Radiating values in academic

writing’. Journal of English for Academic Purposes 5: 37–49.
Hübler, A. 1998. The Expressivity of Grammar: Grammatical Devices Expressing
Emotion Across Time. Berlin: Mouton de Gruyter.
Humphrey, S. 2006. ‘ “Getting the reader on side”: exploring adolescent online
political discourse’. E-Learning 3/2: 143–57.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Hunston, S. 1994. ‘Evaluation and organization in a sample of written academic
discourse’. In Coulthard, M. (ed.) Advances in Written Text Analysis. London:
Routledge: 191–218.
Hunston, S. 2002 ‘Pattern grammar, language teaching, and linguistic variation’.
In Reppen et al. (eds): 167–83.
Hunston, S. 2003. ‘Frame, phrase or function: a comparison of frame semantics
and local grammars’. In Archer, D., Rayson, P. Wilson, A. & T. McEnery (eds)
Proceedings of the Corpus Linguistics 2003 conference. UCREL Technical Papers 16.
Lancaster: UCREL: 342–58.
Hunston, S. 2006. ‘Phraseology and system: a contribution to the debate’. In
Thompson & Hunston (eds): 55–80.
Hunston, S. & G. Francis 2000. Pattern Grammar. A Corpus-driven Approach to the
Lexical Grammar of English. Amsterdam/Philadelphia: Benjamins.
Hunston, S. & J. Sinclair 2000. ‘A local grammar of evaluation’. In Hunston &
Thompson (eds): 74–101.
Hunston, S. & G. Thompson (eds) 2000. Evaluation in Text: Authorial Stance and
the Construction of Discourse. Oxford: OUP.
Hyland, K. 1999. ‘Disciplinary discourses: writer stance in research articles’. In
Candlin, C. & K. Hyland (eds) Writing: Text, Processes and Practices. London:
Longman: 122–42.
Hyland, K. 2000. Disciplinary Discourses: Social Interaction in Academic Genres.
London/New York: Longman.
Hyland, K. 2004. ‘Engagement and disciplinarity: the other side of evaluation’. In
Del Lungo Camiciotti, G. & E. Tognini Bonelli (eds) Academic Discourse – New
Insights into Evaluation. Bern: Peter Lang: 13–30.
Iedema, R., Feez, S. & P.R.R. White 1994. Media Literacy. Sydney: Disadvantaged
Schools Program, NSW Department of School Education. Extracts available to
download from http://www.grammatics.com/appraisal.
Irvine, J. 1990. ‘Registering affect: heteroglossia in the linguistic expression of
emotion’. In Lutz & Abu-Lughod (eds):126–63.
Izard, C.E. 1977. Human Emotions. New York: Plenum.
Jahr, S. 2000. Emotionen und Emotionsstrukturen in Sachtexten. Berlin/New York:
Walter de Gruyter.
Jakobson, R. 1960. ‘Closing statement: linguistics and poetics’. In Sebeok, Th.
(ed.) Style in Language. Cambridge, Massachusetts: MIT Press: 350–77.
Janney, R. 1996. Speech and Affect: Emotive Uses of English. Munich. [no publishing
house]
Jefferson, G. 1988. ‘On the sequential organization of troubles talk in ordinary
conversation’. Social Problems 35/4: 418–42.
Johansson, S. 1981. ‘Word frequencies in different types of English text’. ICAME
News 5: 1–13.
Johnson-Laird, P.N. & K. Oatley 1989. ‘The language of emotions: An analysis of
a semantic field’. Cognition and Emotion 3: 81–123.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

230 References

Kaltenbacher, M. 2006. ‘Culture related linguistic differences in tourist web-

sites: the emotive and the factual. A corpus analysis within the framework of
Appraisal’. In Thompson & Hunston (eds): 269–92.
Kaltenbacher, M. 2007. ‘Systemic functional linguistics and corpus analy-
sis: the language of exaggeration in web-sites of tourism’. In Gruber, H.,
Kaltenbacher, M. & P. Muntigl (eds) Empirical Approaches to Discourse. –

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Empirieorientierte Ansätze in der Diskursanalyse. Frankfurt: Peter Lang: 89–117.
Kidron, Y. & R. Kuzar 2002. ‘My face is paling against my will: emotion and control
in English and Hebrew’. Pragmatics & Cognition 10: 129–58.
Kilgarriff, A. 1997a. ‘Putting frequencies in the dictionary’. International Journal of
Lexicography 10: 135–55.
Kilgarriff, A. 1997b. ‘ “I don’t believe in word senses” ’. Computers and the
Humanities 31: 91–113.
Kilgarriff, A. 2001. ‘Comparing corpora’. International Journal of Corpus Linguistics
6: 97–133.
Knight, N.K. 2007. ‘Laughing off the wrinkles: affiliation and identity negotiation
of cooperative conversational humour among friends.’ PhD Thesis in progress,
University of Sydney, Australia.
Körner, H. & C. Treloar 2003. ‘Needle and syringe programmes in the local media:
“needle anger” versus “effective education in the community” ’. International
Journal of Drug Policy 15: 46–55.
Kövecses, Z. 2000. Metaphor and Emotion. Language, Culture, and Body in Human
Feeling. Cambridge: CUP.
Konstantinidou, M. 1997. Sprache und Gefühl. Hamburg: Helmut Buske Verlag.
Koven, M. 2004. ‘Getting “emotional” in two languages: bilinguals’ verbal
performance of affect in narratives of personal experience’. Text 24: 471–515.
Labov, W. 1972. Language in the Inner City. Philadelphia: University of
Pennsylvania Press.
Labov, W. 1984. ‘Intensity’. In Schiffrin, D. (ed.) Meaning, Form, and Use in Context:
Linguistic Applications. Washington: Georgetown University Press: 43–70.
Labov, W. & D. Fanshel 1977. Therapeutic Discourse. New York: Academic Press.
Lassen, I. 2006. ‘De-naturalizing ideology: presupposition and Appraisal in
biotechnology press releases’. In Lassen, I., Strunck, J. & T. Vestergaard (eds)
Mediating Ideology in Text and Image: Ten Critical Studies. Amsterdam: Benjamins:
97–118.
Lee, D. 2001. ‘Genres, registers, text types, domains, and styles: clarifying the
concepts and navigating a path through the BNC jungle’. Language Learning &
Technology 5: 37–72. Available to download at http://llt.msu.edu/vol5num3/lee/
(last accessed 20 October 2005).
Lee, D. 2002. ‘Notes to accompany the BNC WORLD edition (bibliographical)
index’. Unpublished manuscript, last edited 30 Oct 2002. Available to download
at http://clix.to/davidlee00 (last accessed 20 October 2005).
Leech, G. 1974. Semantics. Harmondsworth: Penguin Books Ltd.
Leech, G., Rayson, P. & A. Wilson 2001. Word Frequencies in Written and Spoken
English. London: Longman.
Lemke, J.L. 1998. ‘Resources for attitudinal meaning: evaluative orientations in
text semantics’. Functions of Language 5/1: 33–56.
Lutz, C. 1990. ‘Engendered emotion: gender, power, and the rhetoric of emotional
control in American discourse’. In Lutz & Abu-Lughod (eds): 69–84.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 References 231

Lutz, C. & L. Abu-Lughod (eds) 1990. Language and the Politics of Emotion.
Cambridge: CUP.
Lyons, J. 1982. ‘Deixis and subjectivity: loquor, ergo sum?’. In Jarvella, R. &
W. Klein (eds) Speech, Place, and Action. Studies in Deixis and Related Topics.
Chichester New York: Wiley: 101–24.
Maclean, P.D. 1993. ‘Cerebral evolution of emotion’. In Lewis, M. & J.M. Haviland

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(eds) Handbook of Emotions. New York: Guilford: 67–83.
Macken-Horarik, M. 2003a. ‘APPRAISAL and the special instructiveness of narra-
tive’. Text 23: 285–312.
Macken-Horarik, M. 2003b. ‘Envoi: intractable issues in appraisal analysis?’. Text
23: 313–19.
Macken-Horarik, M. & J.R. Martin (eds) 2003. Special issue of Text 23/2 on
‘Appraisal’.
Martin, J.R. 1993. ‘Genre and literacy – modelling context in educational
linguistics’. Annual Review of Applied Linguistics 13: 141–72.
Martin, J.R. 1995. ‘Reading positions/positioning readers: judgement in English’.
Prospect: a Journal of Australian TESOL 10: 27–37.
Martin, J.R. 1997. ‘Analysing genre: functional parameters’. In Christie & Martin
(eds): 3–39.
Martin, J.R. 2000a. ‘Beyond exchange: APPRAISAL systems in English’. In Hunston
& Thompson (eds): 142–75.
Martin, J.R. 2000b. ‘Factoring out exchange: types of structure’. In Coulthard, M.,
Cotterill, J. & F. Rock (eds) Working with Dialogue. Tübingen: Niemeyer:
19–40.
Martin, J.R. 2001. ‘Fair trade: negotiating meaning in multimodal texts’. In
Coppock, P. (ed.) The Semiotics of Writing: Transdisciplinary Perspectives on the
Technology of Writing. Belgium: Brepols: 311–38.
Martin, J.R. 2002. ‘Blessed are the peacemakers: reconciliation and evaluation’. In
Candlin, C. (ed.) Research and Practice in Professional Discourse. Hong Kong: City
University of Hong Kong Press: 187–227.
Martin, J.R. 2004a. ‘Sense and sensibility: texturing evaluation’. In Foley,
J. (ed.) New Perspectives on Education and Discourse. London: Continuum:
270–304.
Martin, J.R. 2004b. ‘Mourning: how we get aligned’. Discourse & Society 15/2–3:
321–44.
Martin, J.R. 2004c. ‘Positive discourse analysis: solidarity and change’. Revista
Canaria de Estudios Ingleses 49: 179–200.
Martin, J.R. in press. ‘Genre, ideology and intertextuality: a systemic functional
perspective’. Human Communication [Special issue on ‘Genre’ edited by John
Bateman].
Martin, J.R. & G. Plum 1997. ‘Construing experience: some story genres’. Journal
of Narrative and Life History 7/1–4: 299–308 [Special issue on ‘Narrative analysis’
edited by M. Bamberg].
Martin, J. R. & D. Rose 2003. Working with Discourse: Meaning beyond the Clause.
London: Continuum.
Martin, J.R. & D. Rose 2007. Genre Relations. Mapping Culture. London: Equinox.
Martin, J.R. & P.R.R. White 2005 (paperback edition 2007). The Lan-
guage of Evaluation: Appraisal in English. Basingstoke/New York: Palgrave
Macmillan.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

232 References

Martin, J.R., Painter, C. & L. Unsworth 2006. ‘ “Ambience” in children’s narra-

tive picture book images’. Paper presented at the Systemic Functional Linguistic
Research Seminar, 18 August 2006, University of Sydney, Australia.
Martinet, A. 1991. ‘Expressivité’. Linguistique 27/1: 3–14.
Marty, A. 1908/1976. Untersuchungen zur Grundlegung der allgemeinen Grammatik
und Sprachphilosophie. Tübingen: Niemeyer. [Published originally in 1908, Halle:

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Niemeyer.]
Matthiessen, C.M.I.M. 2006. ‘Frequency profiles of some basic grammatical
systems: an interim report. In Thompson & Hunston (eds): 103–42.
Meštrović, S. 1997. Postemotional Society. London: Sage.
Mees, U. 2006. ‘Zum Forschungsstand der Emotionspsychologie – eine Skizze’.
Available to download at http://www.psychologie.uni-oldenburg.de/mub/
Zum%20Forschungsstand%20der%20Emotions-psychologie.pdf (last accessed
13 July 2006). [Published in Schützreichel, R. (ed.) Emotionen und Sozialtheorie.
Disziplinäre Ansätze. Frankfurt: Campus: 104–24.]
Miller, D.R. 1999. ‘Meaning up for grabs: value-orientation patterns in British
parliamentary debate on Europe’. In Verschueren J. (ed.) Language and Ideology:
Selected papers from the 6th International Pragmatics Conference, Vol. I, Antwerp:
International Pragmatics Association: 386–404.
Miller, D.R. 2002. ‘Ways of meaning “yea” and “nay” in parliamentary debate as
register: a cost-benefit analysis’. In Bignami, M., Iamartino, G. & C. Pagetti (eds)
The Economy Principle in English: Linguistic, Literary, and Cultural Perspectives. Pro-
ceedings of the 19th Conference of the Associazione Italiana di Anglistica. Milano:
Edizioni Unicopli: 220–33.
Miller, D.R. 2006. ‘From concordance to text: appraising “giving” in Alma Mater
donation requests’. In Thompson & Hunston (eds): 248–68.
Milton, K. & M. Svašek (eds) 2005. Mixed Emotions. Anthropological Studies of
Feeling. Oxford/New York: Berg.
Milton, K. 2005. ‘Meanings, feelings and human ecology’. In Milton & Svašek
(eds): 25–41.
Moon, R. 1998. Fixed Expressions and Idioms in English. Oxford: Clarendon.
Moore, C.C., Romney, A.K., Hsia, T.-L. & C.D. Rusch 1999. ‘The universality of
the semantic structure of emotion terms: methods for the study of inter- and
intra-cultural variability’. American Anthropologist 101: 529–46.
Moore, R.C. 2004. ‘On log-likelihood-ratios and the significance of rare events’.
Proceedings of the 2004 Conference on Empirical Methods in Natural Language
Processing. Barcelona, Spain: 333–40.
Neale, A. 2006. ‘ “Matching” corpus data and system networks: using corpora
to modify and extend the system networks for TRANSITIVITY in English’. In
Thompson & Hunston (eds): 143–63.
Niemeier, S. & R. Dirven (eds) 1997. The Language of Emotions. Amsterdam/
Philadelphia: Benjamins.
Nissenbaum, H. F. 1985. Emotion and Focus. Stanford: CSLI.
Nöth, W. 1992. ‘Symmetries & asymmetries between positive and negative
emotion words.’ In Busse (ed.): 72–88.
Oakes, M.P. 1998. Statistics for Corpus Linguistics. Edinburgh: Edinburgh University
Press.
Oakey, D. 2002. ‘Formulaic language in English academic writing’. In Reppen
et al. (eds): 111–29.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 References 233

Oatley, K. 1994. ‘A taxonomy of the emotions of literary response and a theory

of identification in fictional narrative’. Poetics 23: 53–74.
Oatley, K. 2003. ‘Creative expression and communication of emotions in the
visual and narrative arts’. In Davidson et al. (eds): 481–502.
Oatley, K., Keltner, D. & J. Jenkins 2 2006. Understanding Emotions. Oxford:
Blackwell.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Ochs, E. & B. Schieffelin 1989. ‘Language has a heart’. Text 9: 7–25.
O’Halloran, K. 2003. Critical Discourse Analysis and Language Cognition. Edinburgh:
Edinburgh University Press.
Ortony, A., Clore, G.L. & M.A. Foss 1987. ‘The referential structure of the affective
lexicon’. Cognitive Science 11: 341–64.
Ortony, A., Clore, G.L. & A. Collins 1988. The Cognitive Structure of Emotions.
Cambridge: CUP.
Osmond, M. 1997. ‘The prepositions we use in the construal of emotion: why
do we say fed up with but sick and tired of?’. In Niemeier & Dirven (eds):
111–33.
Painter, C. 2003. ‘Developing attitude: an ontogenetic perspective on APPRAISAL’.
Text 23/2: 183–209.
Palmer, G.B. & D.J. Occhi 1999. ‘Introduction: linguistic anthropology and
emotional experience’. In Palmer, G.B. & D.J. Occhi (eds) Languages of Sen-
timent: Cultural Constructions of Emotional Substrates. Amsterdam: Benjamins:
1–22.
Parkinson, B. & A. Manstead 1993. ‘Making sense of emotion in stories and social
life’. Cognition and Emotion 7: 295–323.
Parrott, W. G. 1995. ‘The heart and the head. Everyday conceptions of being
emotional’. In Russell et al. (eds): 73–84.
Parrott, W. G. & R. Harré 1996. ‘Introduction: some complexities in the study of
emotions’. In Harré & Parrott (eds): 1–20.
Partington, A. 2004. ‘Corpora and discourse, a most congruous beast’. In
Partington, A., Morley, J. & L. Haarman (eds) Corpora and Discourse. Bern:
Peter Lang: 11–20.
Peirce, C. 1978. ‘Logic as semiotic: the theory of signs’. In Buchler, J. (ed.) The Phi-
losophy of Peirce. Selected Writings. New York: AMS: 98–119. [Reprint of 1940
edition, International Library of Psychology, Philosophy, and Scientific Method.
London: Kegan Paul Trench, Trubner & Co.]
Péter, M. 1984. ‘Das Problem des sprachlichen Gefühlsausdrucks in beson-
derem Hinblick auf das Bühlersche Organon-Modell’. In Eschbach, A. (ed.)
Bühler-Studien. Vol. 1. Frankfurt/Main: Suhrkamp: 239–60.
Pishwa, H. 2005. ‘Goal markers and their evaluative force’. Paper presented at
the International Conference on Evaluation and Text Types, 22–23 July 2005,
University of Augsburg, Germany.
Pittam, J. & C. Gallois 2002. ‘The language of fear: the communication of inter-
group attitudes in conversations about HIV and AIDS’. In Fussell (ed.): 209–29.
Planalp, S. 1999. Communicating Emotion. Social, Moral and Cultural Processes.
Cambridge: CUP.
Poynton, C. 1985. Language and Gender: Making the Difference. Geelong, Victoria:
Deaking University Press.
Precht, K. 2000. ‘Patterns of Stance in English’. Unpublished PhD thesis, Northern
Arizona University, Flagstaff, AZ.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

234 References

Precht, K. 2003. ‘Stance moods in spoken English: evidentiality and affect in

British and American conversation’. Text 23: 239–57.
Radden, G. 1998. ‘The conceptualisation of emotional causality by means of
prepositional phrases’. In Athanasiadou & Tabakowska (eds): 273–94.
Reilly, J. & L. Seibert 2003. ‘Language and emotion’. In Davidson et al. (eds):
535–59.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Reppen, R., Fitzmaurice, S.M. & D. Biber 2002a. ‘Introduction’. In Reppen et al.
(eds): vii–xi.
Reppen, R., Fitzmaurice, S.M. & D. Biber (eds) 2002b. Using Corpora to Explore
Linguistic Variation. Amsterdam/Philadelphia: Benjamins.
Richards, I.A. 1948. ‘Emotive meaning again’. The Philosophical Review 57:
145–57.
Rimé, B., Finkenauer, C., Luminet, O., Zech, E. & P. Philippot 1998. ‘Social shar-
ing of emotion: new evidence and new questions’. European Review of Social
Psychology 9: 145–89.
Rimmon-Kenan, S. 2 2002. Narrative Fiction. Contemporary Poetics. London/New
York: Routledge.
Robinson, J. 2005. Deeper than Reason: Emotion and its Role in Literature, Music, and
Art. Oxford: OUP.
Rosenberg, D. 1990. ‘Language in the discourse of the emotions’. In Lutz &
Abu-Lughod (eds): 162–85.
Rothery, J. & M. Stenglin 2000. ‘Interpreting literature: the role of APPRAISAL’.
In Unsworth, L. (ed.) Researching Language in Schools and Functional Linguistic
Perspectives. London: Cassell: 222–44.
Russell, J. 1991. ‘In defense of a prototype approach to emotion concepts’. Journal
of Personality and Social Psychology 60: 37–47.
Russell, J.A., Fernández-Dols, J.-M., Manstead, A.S. & J.C. Wellenkamp (eds) 1995.
Everyday Conceptions of Emotion. An Introduction to the Psychology, Anthropology
and Linguistics of Emotion. Dordrecht: Kluwer Academic Publishers.
Šabršula, J. 1982. ‘ “L’expressivité” et l’affectivité en tant que phénomènes
linguistiques’. Philologica Pragensia 25: 150–3.
Sanford, A.J. & S.C. Garrod 1994. ‘Selective processing in text understanding’.
In Gernsbacher, M.A. (ed.) Handbook of Psycholinguistics. San Diego: Academic
Press: 699–719.
Sandlund, E. 2004. Feeling by Doing: The Social Organization of Everyday Emotions
in Academic Talk-in-Interaction. Karlstad: Karlstad University Studies.
Sarangi, S. 2003. ‘Editorial: evaluating evaluative language’. Text 23/2: 165–70.
Scheibman, J. 2002. Point of View and Grammar. Amsterdam/Philadelphia:
Benjamins.
Schneider, K.P. 1991. ‘Affektive Lexik’. In Klein, E., Duteil, F.P. &
K.H. Wagner (eds) Betriebslinguistik und Linguistikbetrieb. Tübingen: Niemeyer:
233–41.
Schrauf, R.W. & J. Sanchez 2004. ‘The preponderance of negative emotion words
across generations and across cultures’. Journal of Multilingual and Multicultural
Development 25/2–3: 266–84.
Scott, M. 1997. ‘PC analysis of key words, and key key words’. System 25:
233–45.
Scott, M. & C. Tribble 2006. Textual Patterns. Key Words and Corpus Analysis in
Language Education. Amsterdam/Philadelphia: Benjamins.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 References 235

Selting, M. 1994. ‘Emphatic speech style – with special focus on the prosodic
signalling of heightened emotive involvement in conversation’. Journal of
Pragmatics 22: 375–408.
Shaver, P., Schwartz, J., Kirson, D. & C. O’Connor 1987. ‘Emotion knowledge:
further exploration of a prototype approach’. Journal of Personality and Social
Psychology 52: 1061–86.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Shimanoff, S.B. 1985. ‘Expressing emotions in words: verbal patterns of interac-
tion’. Journal of Communication 35: 16–31.
Shriver, L. 2006. We Need to Talk About Kevin. Melbourne: The Text Publishing
Company.
Simpson, P. 1993. Language, Ideology and Point of View. London/New York:
Routledge.
Sinclair, J.M. 1987. ‘Mirror for a text’. Unpublished manuscript. University of
Birmingham.
Sinclair, J.M. 2004a. ‘The search for units of meaning’. In Sinclair (2004e): 24–48.
[Published originally in Textus IX, 1996.]
Sinclair, J.M. 2004b. ‘The lexical item’. In Sinclair (2004e): 131–48. [Published
originally in Weigand, E. (ed.) 1998. Contrastive Lexical Semantics. Amsterdam:
Benjamins.]
Sinclair, J.M. 2004c. ‘Corpus and text: basic principles’. In Wynne (ed.): 1–16.
Sinclair, J.M. 2004d. ‘Appendix: how to build a corpus’. In: Wynne (ed.): 79–83.
Sinclair, J.M. 2004e. Trust the Text. Language, Corpus and Discourse. London/New
York: Routledge.
Stankiewicz, E. 1964. ‘Problems of emotive language’. In Sebeok, Th. (ed.)
Approaches to Semiotics. The Hague: Mouton: 239–64.
Stankiewicz, E. 1989. ‘The emotive function of language: an overview’. In
Koch, W. (ed.) For a Semiotics of Emotion. Bochum: Bochumer Universitätsverlag:
73–85.
Stearns, P.N. 1994. American Cool: Developing a Twentieth-Century Emotional Style.
New York: University Press.
Stearns, P.N. 1995. ‘Emotion’. In Harré, R. & P. Stearns (eds) Discursive Psychology
in Practice. London: Sage: 37–54.
Storm, C. & T. Storm 1987. ‘A taxonomic study of the vocabulary of emotions’.
Journal of Personality and Social Psychology 53: 805–16.
Stubbs, M. 1996. Text and Corpus Analysis. Oxford: Blackwell.
Stubbs, M. 2001. ‘On inference theories and code theories: corpus evidence for
semantic schemas’. Text 21/3: 437–65.
Stubbs, M. & I. Barth 2003. ‘Using recurrent phrases as text-type discrimina-
tors: A quantitative method and some findings’. Functions of Language 10/1:
61–104.
Swales, J. M. 2004. Research Genres. Explorations and Applications. Cambridge: CUP.
Taboada, M. & J. Grieve 2004. ‘Analyzing appraisal automatically’. Exploring
Attitude and Affect in Text: Theories and Applications. Papers from 2004 AAAI
Spring Symposium. Stanford: Stanford AAAI (American Association for Artificial
Intelligence) Technical Reports.
Teubert, W. 2004a. ‘When did we start feeling guilty?’. In Weigand (ed.) : 121–62.
Teubert, W. 2004b. ‘Did Lou inspire guilt, as well?’ JLS 33: 169–89.
Thompson, G. & S. Hunston (eds) 2006. System and Corpus. Exploring Connections.
London: Equinox.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

236 References

Tognini-Bonelli, E. 2001. Corpus Linguistics at Work. Amsterdam: John Benjamins.

Toolan, M. 2 2001. Narrative. A Critical Linguistic Introduction. London/New York:
Routledge.
Turner, J. & J.E. Stets 2005. The Sociology of Emotions. Cambridge: CUP.
Ungerer, F. 1997. ‘Emotions and emotional language in English and German news
stories’. In Niemeier & Dirven (eds): 307–28.

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

Ungerer, F. & H.-J. Schmid 1996. An Introduction to Cognitive Linguistics. London:
Longman.
van Dijk, T. 2001. ‘Bibliography discourse, media and emotions.’ Unpub-
lished manuscript. Available to download at http://www.discourse-in-
society.org/biblio-disc-media-emotions.html (last accessed 31 August 2005).
van Meel, J. M. 1994. ‘Representing emotions in literature and paintings: a
comparative analysis’. Poetics 23: 159–76.
Volek, B. 1977. ‘Die Kategorie der Emotionalität in der Sprache’. Papiere zur
Linguistik 17/18: 123–48.
Volek, B. 1987. Emotive Signs in Language and Semantic Functioning of Derived Nouns
in Russian. Amsterdam: Benjamins.
Wallace, A. & M.T. Carson 1973. ‘Sharing and diversity in emotion terminology’.
Ethos 1: 1–29.
Watters, P. 2002. ‘Discriminating English word senses using cluster analysis’.
Journal of Quantitative Linguistics 9: 77–86.
Watson, G. 1999. ‘Evidentiality and affect’. Language and Literature 8/3: 217–40.
Watzlawick, P, Beavin, J. H. & D. D. Jackson 1967. Pragmatics of Human Com-
munication. A Study of Interactional Patterns, Pathologies, and Paradoxes. New
York/London: W. W. Norton & Company.
Weigand, E. (ed.) 2004a. Emotion in Dialogic Interaction. Advances in the Complex.
Amsterdam: Benjamins (Current Issues in Linguistic Theory 248).
Weigand, E. 2004b. ‘Emotions: The simple and the complex’. In Weigand (ed.):
3–22.
Werth, P. 1998. ‘On the semantics and pragmatics of emotion verb complemen-
tation’. In Athanasiadou & Tabakowska (eds): 409–40.
White, G. 1990. ‘Moral discourse and the rhetoric of emotions’. In Lutz &
Abu-Lughod (eds): 46–68.
White, P.R.R. 1997. ‘Death, disruption and the moral order: the narrative impulse
in mass-media “hard news” reporting’. In Christie & Martin (eds): 101–33.
White, P.R.R. 1998. ‘Telling Media Tales: the News Story as Rhetoric’. Unpublished
Ph.D. thesis, University of Sydney, Sydney.
White, P.R.R. 2000. ‘Dialogue and inter-subjectivity: reinterpreting the semantics
of modality and hedging’. In Coulthard, M., Cotterill, J. & F. Rock (eds). Dialogue
Analysis VII: Working with Dialogue. Tübingen: Niemeyer: 67–80.
White, P.R.R. 2001. ‘Attitude/Appreciation’. Unpublished manuscript (word
processor version). Available to download at http://www.grammatics.com/
appraisal (last accessed 3 January 2005).
White, P.R.R. 2002. ‘Appraisal’. In Verschueren et al. (eds) Handbook of Pragmatics.
Amsterdam: Benjamins: 1–27.
White, P.R.R. 2003a. ‘Appraisal and the resources of intersubjective stance’.
Unpublished manuscript (word processor version). Available to down-
load at http://www.grammatics.com/appraisal (last accessed 28 September
2004).

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 References 237

White, P.R.R. 2003b. ‘Beyond modality and hedging: a dialogic view of the
language of intersubjective stance’. Text 23: 259–84.
White, P.R.R. 2004a. ‘Attitudinal meaning and inference – semantic prosodies in
text and inter-text’. Paper presented at the English Language Research Seminar,
17 February 2004, University of Birmingham (UK).
White, P.R.R. 2004b. ‘Alignment, rapport and the construed reader’. Paper pre-

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

sented at the Evaluation and Stance Workshop, 5–6 June 2004, University of
Birmingham (UK).
White, P.R.R. 2004c. ‘Subjectivity, evaluation and point of view in media dis-
course’. In Coffin, C., Hewings, A. & K. O’Halloran (eds) Applying English
Grammar: Functional and Corpus Approaches. London: Arnold: 229–46.
White, P.R.R. 2006. ‘Evaluative semantics and ideological positioning in jour-
nalistic discourse: A new framework for analysis’. In Lassen, I., Strunck, J. &
T. Vestergaard (eds) Mediating Ideology in Text and Image. Amsterdam: Benjamins:
37–67.
Whitelaw, C., Garg, N. & S. Argamon 2005. ‘Using appraisal groups for sentiment
analysis’. Proceedings from CIKM ’05, 31 October–5 November 2005, Bremen,
Germany.
Wierzbicka, A. 1992a. ‘Talking about emotions: semantics, culture and cognition’.
Cognition and Emotion 6: 284–319.
Wierzbicka, A. 1992b. ‘Defining emotion concepts’. Cognitive Science 16: 539–81.
Wierzbicka, A. 1998. ‘ “Sadness” and “anger” in Russian: the non-universality
of the so-called “basic human emotions” ’. In Athanasiadou & Tabakowska
(eds): 3–28.
Wierzbicka, A. 1999. Emotions Across Languages and Cultures. Cambridge: CUP.
Wilk, N.M. 2005. ‘Semiotik der Gefühle. Versuch einer Integration der Affekte in
eine linguistisch-semiotische Theorie des Verstehens’. Zeitschrift für Literaturwis-
senschaft und Linguistik 138: 129–41.
Wittgenstein, L. 1967. Philosophische Untersuchungen. Frankfurt: Suhrkamp.
Wollheim, R. 2001. Emotionen. Eine Philosophie der Gefühle. München: C. H. Beck.
Wowk, M.T. 1989. ‘Emotion talk’. In Torode, B. (ed.) Text and Talk as Social Practice.
Dordrecht/Providence, RI: Foris Publications: 51–71.
Wynne, M. (ed.) 2005. Developing Linguistic Corpora. A Guide to Good Practice.
Oxford: Oxbow Books/Arts and Humanities Data Service.

Corpora and Dictionaries

BNC = The British National Corpus, version 2 (BNC World). 2001. Distributed by
Oxford University Computing Services on behalf of the BNC Consortium. URL:
http://www.natcorp.ox.ac.uk/
BNC baby = The BNC Baby, version 2. 2005. Distributed by Oxford Uni-
versity Computing Services on behalf of the BNC Consortium. URL:
http://www.natcorp.ox.ac.uk/
BRC = The British Register Corpus. 2006. Compiled by Monika Bednarek from the
BNC, with the help of Lee (2001, 2002).
BRC baby = 85,000 word subset from the BRC (∼ 20,000 words from conversa-
tion, news reportage, fiction, academic discourse). 2006. Compiled by Monika
Bednarek.

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

238 References

ET = Rooney, K. 2001. Encarta Thesaurus. London: Bloomsbury Publishing.

OALD = Oxford Advanced Learner’s Dictionary. 7th edition. Oxford: OUP. (used on
CD-ROM: Oxford Advanced Learner’s Compass)

Websites (last accessed 10 January 2007)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

http://clix.to/davidlee00 (David Lee’s excel spreadsheet)
http://escorp.unizh.ch/ (BNCweb website)
http://framenet.icsi.berkeley.edu (FrameNet website)
http://www.altova.com (XML spy website)
http://www.ekl.oulu.fi/stance/index.html (Stance project website)
http://www.grammatics.com/appraisal (Appraisal website)
http://www.MonikaBednarek.com (author’s website)
http://www.natcorp.ox.ac.uk (BNC website)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

Index

academic discourse, 19–20, 21, 28, 32, affect types, 160–71, 172–5, 184, 185,

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

33, 35, 36, 37, 38, 40–2, 43, 44, 186, 191, 194, 195, 207
45–7, 48, 49, 50, 51, 52, 55, 57, dis/inclination, 154, 156–7, 165–6,
58, 59, 61, 63 (note 5), 126–31, 168, 172, 173, 222 (note 8)
132, 133, 134, 136–7, 144, 145, dis/satisfaction, 14, 154, 155, 156,
199–208, 209, 218–19, 222 168, 169, 170, 174–5, 181 (note
(note 7) 9), 182 (note 14)
action, 5, 69, 71, 76, 77, 88, 122, 123, in/security, 14, 154, 155, 156,
136, 148, 149, 150, 171, 178 160–1, 165, 169, 170–1, 173–4
(note 2) un/happiness, 14, 154, 155, 168,
affect, 7–10, 11–12, 13, 14–15, 16, 17, 169, 170, 175, 182 (notes 14, 15)
15 (note 1), 34, 35, 38, 69, 70, 74, surprise, 155, 160–5, 168, 169, 170,
136–7, 151, 152, 154–72, 186, 171, 175, 180 (note 8), 180–1
219, see also emotion (note 9), 181 (note 10)
authorial and non-authorial, 14, 33, affective key, 183–6, 194–7, 199–206,
58–9, 63 (note 11), 63-4 (note 12), 208–11, see also evaluative key
73, 85, 101, 102, 103, 104–5, 106, affective stance, 184, 186–94, 197–9,
107, 109, 110, 115, 116, 126, 129, 206–8
136, 146, 158–9, 194, 196–7, 200, ambience, 146, 217
213, 216 appraisal, 10, 12–16, 70, 73, 142–4,
covert vs. overt, 90–4, 95, 101, 102, 145–7, 154–7, 178 (note 1), 182
103, 106, 107, 109, 110, 111, 112, (note 16), 186
114, 116, 118, 122, 124, 126, 127, modification of, 147–52, 160–9,
128, 136–7, 137 (note 3), 139–40 171–2
(note 10), 154, 176–7, 182 (note appreciation, 13, 15, 35, 74, 97 (note
16), 192, 196–7 1), 164, 182 (note 16), 196, 209,
directed and undirected, 14, 73–7, 214
77–90, 95, 101, 102, 105, 106, attitude, 13, 14–15, 184, 186, 208,
107, 109, 111, 114, 117, 118, 122, 215, see also affect, judgement,
124, 126, 131–5, 136, 139–40 appreciation
(note 10), 154, 157, 172
emoted and unemoted, 72–3, 95,
111, 140 behavioural surge, 14, 26 (note 7),
intensity of, 9, 14, 102, 104, 108, 124–6, 146–52, 162–3, 193, see
112, 116, 154, 163 also emotion: bodily symptoms,
mediated and unmediated, 194, physiological expressions
195–6, 197, 221 (notes 2, 3) Bamberg, M., 4, 7, 9, 10, 149, 150,
positive and negative, see emotion 159, 163, 212
terms, valence Bednarek, M., 25 (note 4), 35, 43, 102,
realis and irrealis, 14, 86, 154, 156, 104, 108, 116, 143, 148, 150, 159,
166–7, 172 178 (note 3), 181 (note 10), 182
affect patterns, 70–99, 101–31, 136–7 (note 16), 194, 199, 215, 223
affect style, 35, 37, 38–40 (note 8)

239

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

240 Index

Biber, D., 16, 18, 19, 20, 21, 27, 28, 33, causes of, 67, 69, 70, 90, 92, 93, 98
39, 40, 55, 56, 57, 58, 59, 61, 62 (note 9), 148, 149, 150, 151–2,
(note 1), 64 (note 12), 72, 88, 94, 171, 214–15
97 (notes 3, 6), 110, 184, 210, 219 portraying vs. creating, 146
British National Corpus (BNC), 19, 22, socialization of, 5, 6, 33–4, 179
23, 38, 49, 55, 57, 70, 140–1 (note 3)

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

(note 14) emotion ideologies, 4
BNC baby, 144 emotion metaphors, 3, 96 (note 1)
British Register Corpus (BRC), 19–20, emotion profiles/profiling, 17–18, 27,
21, 23, 28, 29, 38, 40, 70, 209 see also variation
BRC baby, 144–5, 173, 176, 177, 183, emotion research, 4–10, 143–4
211 emotion/feeling rules, 6, 33, 60, 159,
179 (note 3)
cognitive research, 7, 33, 97 (note 4), emotion schemata, 7, 33, 98 (note 9),
147–9, 158, 180–1 (note 9), 181–2 148–51, 158, 171, 178–9 (note 3),
(note 12) 179 (note 5), 215
conversation, 19, 20, 21, 30, 31, 32, emotion talk, 2, 4, 6, 10–12, 16, 17,
33, 35, 36, 37, 38–9, 40–2, 43, 24, 25 (note 1), 149, 221, see also
44–7, 48, 49, 50, 51, 54, 55, 57, emotion terms
59, 63 (note 5), 101–6, 132, functions of, 2, 72, 95–6, 100,
133–4, 136, 143, 144–5, 149, 136–7, 163, 211–19
184–5, 186, 211–15, 222 (note 8) emotion terms, 11, 17, 20–4, 25 (note
corpus linguistics 16, 17, 18, 22, 62 1), 25–6 (note 6), 26 (note 8), 32,
(note 1), 208, 211 47–53, 61, 148, 152–4, 173–5,
large-scale and small-scale, 183, 179–80 (note 7), 181 (note 10),
219–20 222–3 (note 8), see also emotion
talk
display rules, 6, 179 meanings, 8, 22–3, 26 (note 8),
31–2, 32, 55–6, 94, 95–6, 96– 7
Ekman, P., 2, 3, 4, 4, 5, 25 (note 1), (note 1), 159
125, 148, 168, 169, 170, 180 positive and negative, 14, 16, 43,
(notes 7, 9),181 (note 11) 47, 49–51, 63 (note 6), 65, 154,
emoter, 14–15, 59, 61, 70, 74–5, 92, 157, 161, 163–5, 166, 168, 171,
158–9 177, 178, 180 (note 8), 181 (note
emotion, 2–6, 25 (notes 1, 2), 34–5, 10), 194, see also valence
60–1, 62 (note 3), 147–8, 157, salience of, 48, 50
159, 179–80 (note 7), 180 (note emotional culture, 4, 6, 34–5
9), 220–1, see also affect emotional response inventory, 187
and gender, 193 emotional style, 4
and language, 10–12, 34–5, 50, emotional talk, 10–12, 16, 35, 150
61–2, 69–71, 95–6, 146–52, empathy target, 71, 75, 96, 117,
154–69, 171–2, 176, 179–81 (note 119
7), 181–2 (note 12), 184 engagement, 13–14, 184, 186, 208
attitudes towards, 2–4 evaluation, 2, 5, 11, 16, 25 (note 4),
basic emotions, 5, 49, 169–71, 181 25–6 (note 6), 66, 73, 94, 101,
(note 11) 102, 103, 106, 127, 129, 147, 148,
bodily symptoms of, 124–5, 147, 157, 158–9, 164, 166, 181 (note
161–3, 171 see also behavioural 10), 186, 212, 213, 215, 221–2
surge, physiological expressions (note 4)

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

 Index 241

evaluative key, 184, 208–11 Lee, D., 19, 25 (note 5)

expressor, 71, 74, 75, 76, 77, 123, 124, lexico-grammatical patterns, 17, 18,
136, 140 (note 10) 27, 65, 70, 100
local grammar, 17, 24, 66–9, 71, 72,
feeling rules, see emotion rules 77, 95
fiction, 19, 20, 21, 30, 31, 32, 33–4, logogenesis, 183, 216, 220

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

35, 36, 37, 38, 39, 40, 41, 42, 43,
44, 45, 46, 47, 48, 49, 50, 51, 54, Martin, J.R., 13, 24, 25, 73, 142, 142,
55, 57, 58, 59, 63 (note 5), 64 146, 147, 151, 156, 157, 161, 163,
(note 12), 116–26, 132, 133, 134, 182 (note 14), 208, 212, 213, 216,
136, 144, 145, 162, 164, 185–93, 222 (notes 6, 8)
216–18, 223 (note 9) Martin, J.R. & P.R.R. White, 10, 13, 15,
Fillmore, C., 65, 67 32, 90, 94, 97 (notes 1, 5), 102,
focalization, 126, 144, 185–6, 218 104, 125, 142, 144, 145–52,
FrameNet, 17, 65, 67–9, 96–7 (note 1) 154–6, 160, 164, 166, 168, 171,
fuzziness, 26 (note 6), 167–9, 171, 181 172, 179 (note 7), 182 (notes 14,
(note 9) 16), 183, 184, 195, 196, 197, 206,
208, 209, 211, 216, 220, 223
graduation, see affect: intensity of (note 10)
grammatical metaphor, 119–20, 139 Matthiessen, C.M.I.M., 13, 25 (note
(note 8), 222 (note 8) 5), 56, 61, 74, 81, 97 (note 5), 135,
139 (note 8), 168, 183, 197, 219
Halliday, 13, 25, 56, 61, 74, 81, 97, media discourse, 143, see also news
135, 139 (note 8), 168, 183, 197 reportage
happiness, see affect types mental disposition, 14, 26 (note 7),
Hunston, S., 18, 24, 65, 66, 67, 68, 69, 146, 147, 149, 150, 152,
70, 71, 72, 77, 87, 96, 144, 193, mental processes, 11, 31, 56, 77, 81,
212 117, 121, 123, 127, 135, 143
Hunston, S. & G. Francis, 66, 77, 95, meta-emotion philosophy, 4
97 (note 3), 99 (note 10) metaphor, see emotion metaphors,
Hyland, K., 25 (note 4), 27, 144, 210 grammatical metaphor
modality, 8, 11, 25 (note 4), 26 (note
inclination, see affect types 7), 31, 57, 58, 111, 113, 114, 119,
individuation, 208, 220, 222 (note 6) 126, 129, 130, 136, 137, 139
interjection, 11, 12 (note 8), 186, 222 (note 8)
interpersonal function/meaning, 10, modalization, see modality
13, 111, 143, 146, 184, 214, 222 mode, see multimodality
(note 8) modulation, see modality
instantiation, 208, 220, 222 (note 6) multimodality, 144
intensification, see affect: intensity of,
graduation news reportage, 18, 19, 20, 21, 28–9,
involvement, 9, 13 31, 32, 33, 35, 36, 37, 38, 39, 40,
41, 42, 43, 44, 45, 46, 47, 48, 49,
journalistic discourse, see media 50, 51, 54, 55, 57, 58, 59, 63 (note
discourse, news reportage 5), 94, 106–16, 131, 133, 134,
judgement, 13, 14, 15, 143, 145, 182 136, 143–4, 145, 151, 194–9,
(note 16), 185, 196, 199, 214, 217 208–9, 215–16
news values, 35, 43, 108, 115, 116,
key words, 28–31, 40 137, 138 (note 6), 194, 215

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek

242 Index

pattern grammar, see local grammar Stubbs, M., 9, 27, 28, 31, 40, 62–3
physiological expression, 148, 149, (note 5), 67, 95
150, see also bodily symptoms, systemic functional linguistics, 10, 12,
behavioural surge 13, 24, 25 (note 5)
point of view, see focalization system networks, xii, 169, 161, 167,
Precht, K., 16, 17, 20, 26 (notes 6, 8), 168, 172

Copyright material from www.palgraveconnect.com - licensed to Universitetsbiblioteket i Tromso - PalgraveConnect - 2011-03-17

27, 32, 39, 63 (note 5), 143, 159,
211, 212, 213, 222 (note 8) Thompson, G., 18, 212
prosody, 9, 151–2, 183–4 trigger, 14–15, 69–71, 77, 79–81, 83,
prototype, 26 (note 6), 98 (note 9), 85, 86–7, 89–90, 95, 98 (note 9),
148, 168, 178 (note 3), 179 131–5, 156, 158, 166–7, 172, 212
(note 5)
psychological/psychocultural valence, 51, 154, 157, 171, 177, 178,
salience, 48, 50 see also emotion terms: positive
and negative
qualitative, 142, 145, 183, 219–20, see variation, 5, 8, 16–17, 19, 24, 27, 62
also corpus linguistics (note 1), 63 (note 5), 135, 194,
quantitative, 17, 32, 145, 219–20 see 195, 220, 221, 222 (note 6)
also corpus linguistics lexical, 17, 31–8, 47–52
part of speech, 18, 37, 38–52
register, 16–17, 18–20, 21, 25 (note 5), syntactic, 18, 53–8, 61–2, 63
27, 28, 95, 220, 221 (note 7)
lexico-grammatical, 18, 27, 100–41
satisfaction, see affect types volition, see affect types:
Scott, M., 28 dis/inclination
security, see affect types
semantic preference, 114, 123, 133, White, P.R.R., 13, 15, 111, 116, 142,
135 143, 159, 182 (note 16), 197, 216
signature, 208, 220 Wierzbicka, A., 7, 11, 149, 151, 179
Sinclair, J., 17, 20, 23, 66, 67, 212 (note 7), 181 (note 12)
surprise, see affect types
stance, 16–17, 25 (note 4), 26 (note 6),
33, 39, 63 (note 5), 88, 143,
183–4, 210, 211, 220, see also
affective stance
attitudinal, 16, 17, 26 (note 6)
epistemic, 16
style, 16

10.1057/9780230285712 - Emotion Talk Across Corpora, Monika Bednarek