Antaki

ANALYSING
EVERYDAY
EXPLANATIO
N
A Casebook of
Methods
Edited by
CHARLES ANTAKI
SAGEExplanations are given and received in all areas of

social life: in the home, at school, at work, in the
courtroom. They are exchanged between friends and argued
over by enemies.
The analysis of these ordinary explanations is regarded as a
notoriously difficult area of study by social scientists. This
book, for the first time, offers a clear and comprehensive
guide to the most fruitful and interesting techniques for
collecting, analysing and interpreting explanation.
The authors have been chosen to represent the most
important work being done in a variety of disciplines: social
psychology, linguistics, pragmatics, artificial intelligence,
ethogenics, narratology, conversation analysis and discourse
analysis.
Each chapter follows a uniform format. The author
introduces the general theoretical outlines of the technique
and describes his or her own theoretical position. The heart
of the chapter is then devoted to an extended description of
the analysis of a particular piece of 'data', which might be a
conversation, a collection of documentary accounts or a
corpus of explanatory phrases. Finally, the advantages and
disadvantages of each particular analytical method are
assessed.
Usefully organized into four parts, the book deals with the
nature of explanation in general; methods for analysing the
structure and content of accounts; the social context in
which accounts are exchanged; and the use of rhetorical and
ideological approaches to explanation.
Analysing Everyday Explanation is a unique casebook of methods
which will prove invaluable to all social scientists.
Edit
o r 08039 8139 2Charles Antaki is a Lecturer in
ISBN
Psychology at the University of Lancaster. Among other
publications he has edited The Psychology of Ordinary Explanation of
Social Behaviour and Mental Mirrors (with A. Lewis).
Contributors
GinaAgostinelli Charles Antaki Michael Billig Michael].
Cody Stephen W. Draper MaryM. Gergen RomHarre John
H. Harvey John Heritage IvanLeudar Margaret L.
McLaughlin Ian Parker Jonathan Potter Peter Suedfeld
Philip E. Tetlock DawnC.Tumquist James F. Voss Margaret
Wetherell
Jacket design by The Pinpoint Design Company
SAGE Publications 28 Banner Street London

EC1Y 8QE
Analysing Everyday
ExplanationAnalysing
Everyday
Explanation
A Casebook of Methods
Edited by Charles Antaki

(^)SAGE Publications
London Newbury Park Beverly Hills New Delhi

Preface, Chapter 1, Chapter 5 and editorial matter Charles Antaki 1988 Chapter 2 Stephen W. Draper 1988
Chapter 3 John H. Harvey, Dawn C. Turnquist and Gina Agostinelli 1988 Chapter 4 Philip E. Tetlock and Peter
Suedfeld 1988 Chapter 6 James F. Voss 1988 Chapter 7 Mary M. Gergen 1988
Chapter 8 Michael J. Cody and Margaret L. McLaughlin 1988 Chapter 9 John Heritage 1988 Chapter 10 Ivan Leudar
and Charles Antaki 1988 Chapter 11 Rom Harre 1988
Chapter 12 Margaret Wetherell and Jonathan Potter 1988 Chapter 13 Ian Parker 1988 Chapter 14 Michael Billig 1988
First published 1988
All rights reserved. No part of this book may be reproduced or utilized in any form or by any means, electronic or mechanical,
including photocopying, recording, or by any information storage and retrieval system, without permission in writing from the
Publishers.
British Library Cataloguing in Publication Data
SAGE Publications Ltd
28 Banner Street London
EC1Y 8QE
SAGE Publications Inc 2111 West

Hillcrest Street Newbury Park,
California 91320
SAGE Publications India Pvt Ltd C236 Defence Colony New Delhi 110
024
SAGE Publications Inc 275 South

Beverly Drive Beverly Hills,
California 90212
Analysing everyday explanation: a casebook of methods.

1. Explanation. Sociological perspectives
I. Antaki, Charles
302.2
ISBN 0-8039-8139-2 ISBN 0-8039-8140-6 Pbk
Library of Congress catalog card number 88-060348
Printed in Great Britain by Billing and Sons Ltd, Worceste
Explanations, communication and social cognition 7
Contents
List of Contributors
vi
Preface
vii
EXPLANATIONS
1 Explanations, communication and social cognition

Charles Antaki
2 Whats going on in everyday explanation?
Stephen W. Draper
II
STRUCTURE AND CONTENT
1
15
32
3 Identifying attributions in oral and written explanations John H.

Harvey, Dawn C. Turnquist and Gina Agostinelli 32
4 Integrative complexity coding of verbal behaviour
Philip E. Tetlock and Peter Suedfeld
43
5 Structures of belief and justification

Charles Antaki
60
6 Problem solving and reasoning in ill-structured domains

James F. Voss
74
7 Narrative structures in social explanation

Mary M. Gergen
94
III ACCOUNTS IN CONTEXT
113
8 Accounts on trial: oral arguments in traffic court

Michael J. Cody and Margaret L. McLaughlin
113
9 Explanations as accounts: a conversation analytic

perspective
John Heritage
127
10 Completion and dynamics in explanation seeking

Ivan Leudar and Charles Antaki
145
11 Accountability within a social order: the role of pronouns

Rom Harre
156
8 Charles Antaki
IV RHETORIC AND IDEOLOGY
168
12 Discourse analysis and the identification of

interpretative repertoires
Margaret Wetherell and Jonathan Potter
168
13 Deconstructing accounts
Ian Parker
184
14 Methodology and scholarship in understanding

ideological explanation
Michael Billig
199
References
216
Index
229
List of Contributors
Gina Agostinelli, Department of Psychology, Spence Laboratories of

Psychology, University of Iowa, Iowa City, Iowa, USA Charles Antaki,
Department of Psychology, Fylde College, University of Lancaster,
England
10 Charles Antaki
Michael Billig, Department of Social Sciences, University of

Technology, Loughborough, England Michael J. Cody,
Communication Arts and Sciences, University of Southern California,
Grace Salvatori Hall, Los Angeles, California, USA
Stephen W. Draper, Department of Psychology, University of Glasgow,

Glasgow, Scotland
Mary M. Gergen, Department of Psychology, Pennsylvania State

University, Media, Pennsylvania, USA Rom Harre, Sub-Faculty of
Philosophy, University of Oxford, Oxford, England
John H. Harvey, Department of Psychology, Spence Laboratories of

Psychology, University of Iowa, Iowa City, Iowa, USA John Heritage,
Department of Sociology, Warwick University, Coventry, England
12 Charles Antaki
Ivan Leudar, Department of Psychology, University of Manchester,

Manchester, England
Margaret L. McLaughlin, Communication Arts and Sciences,

University of Southern California, Grace Salvatori Hall, Los
Angeles, California, USA
Ian Parker, Department of Psychology and Speech Pathology,

Manchester Polytechnic, Manchester, England Jonathan Potter,
Department of Social Sciences, University of Technology,
Loughborough, England Peter Suedfeld, Department of Psychology,
University of British Columbia, Vancouver, British Columbia, Canada
Philip E. Tetlock, Department of Psychology, University of California,
Berkeley, California, USA Dawn C. Turnquist, Department of
Psychology, Spence Laboratories of Psychology, University of Iowa,
Iowa City, Iowa, USA James F. Voss, Learning Research and
Development Center, University of Pittsburgh, Pittsburgh,
Pennsylvania, USA Margaret Wetherell, Department of Psychology,
The Open University, Milton Keynes, England
Preface
This book is about how social scientists - broadly defined - deal with
ordinary explanations - again, broadly defined. The contributions
14 Charles Antaki
describe ways of dealing with the explanations that we give and

receive in everyday talk. The range of methods is from the technicality
of content analysis to the interpretative rhetoric of scholarly exegesis.
The methods vary with the authors intellectual background and
particular interests.
Some contributors are concerned with explanation as an index of a

persons mental dispositions (of, say, an individuals proneness to
blame her or himself for failure, her or his disposition to see both sides
of an argument). Others are interested in explanations for what they tell
us about the conventions of ordinary reasoning and the use of
explanation in its social context. Still others are concerned with the
way in which social categories and processes (like, say, gender and
power) are manifested in peoples and groups explanations.
These three families of interests are represented in separate parts of

the book. In the introductory part, Steve Draper and I try to give some
background. Part II, Structure and content, deals with the
identification and measurement of explanation and its relation to
personal variables. Part III, Accounts in context, deals with the
analysis of explanations in their interactive domain. Part IV, Rhetoric
and ideology, deals with the analysis of explanations in the realm of
ideological reasoning and the social structures implicit in language.
After the two introductory essays, each chapter follows the same
structure. Firstly, the authors describe their general area of work
(ethnomethodology, say, or attribution theory), then talk a little about
their particular corner of the field. The middle of the chapter is then
devoted to a close look at a specific set of data which the authors use to
illustrate what they say. Finally, the chapter is rounded off by a candid
assessment of the advantages and disadvantages of the method
16 Charles Antaki
described.
Naturally, the chapters are much more than mere recipes. I hope that
the uniform structure will help the reader to compare not only the
methods on offer, but also the intellectual traditions from which they
spring.
Charles Antaki
Lancaster
PARTI
EXPLANATIONS
1
Explanations, communication and
social cognition
Charles Antaki
Explanation can mean many things, and ordinary explanations can be

sought and interpreted in many ways. Different researchers will have
different ways of netting explanations and different ways of making
18 Charles Antaki
sense of their catch. The differences among researchers are not just in
what they think about explanations, but also in what they think about
people and how they ought to be studied.
In this chapter I want to make some remarks about how ordinary

social explanation is studied by different people. I shall use two
observations about explanations - that they have the power to
challenge social realities, and that they seem to be implicated in
changes in peoples behaviour - to make a distinction between two
approaches to their investigation. After seeing what differences there
are between the two approaches, I shall argue that, if one had to choose
between them, one would choose the tradition that was more able to
cope with the unspoken metacognitive background of explanations as
they are exchanged in ordinary talk.

Explanations in the public domain
The features that make something count as an explanation are far from
simple or uniform, as Draper shows in chapter 2. There is certainly no
mechanical way of searching through a transcript of talk and spotting
everything that could count as an explanation. On the other hand, our
intuitions about what does or does not have the general feel of an
explanation are reasonably sharp.
(1)
We can make a start by looking at explanations in the
public domain, where the conventions of explanation exchange are
familiar enough. The following examples will help us see what
attractions the study of explanations might have for social
scientists:The initial success of the out of town centre appears a
paradox in a region with the highest unemployment in mainland
Britain. Mr Hall ... explains: We havent got 18% unemployment,
weve got 82% in work. (The Guardian, London, 26 August 1985)
20 Charles Antaki
(2) Other people, including several other women, were kicked and
knocked about. One old man who was standing inside a doorway was
told to shut the door and keep out of it and when he protested that this
was not the law he was told by the police We are the Law. (Jobs
not Jail, Bulletin of the South Yorkshire Defence Campaign,
February 1985).
We can choose to look into the explanations in these examples in many

ways, of course: we could ask questions about the social information
processing going on in the explainers or the audiences heads; we
could ask questions about the linguistic and rhetorical form of the
utterances reported; we could ask what symbolic significance the
explanation episode had for the participants; and so on. Let us start
with one of the more obvious and powerful features, which will loom
large in many of the later chapters in this book.
What is immediately noticeable about both explanations is the

power they potentially have to overturn definitions of reality and to
impose the speakers preferred interpretation of events. Explanations,
unlike mere assertions in a stream of discourse, reveal or claim to
reveal what is really the case. The speaker might have been asked to
make the revelation (as he presumably was by the reporter in the first
extract) or may choose to unveil it unilaterally (as the policeman is
reported to have done in the second). In either case, the explanatory
context offers information about the episode which, unlike information
exchanged in other, more neutral contexts, promises to reorient the
framing of the event and the participants place within it.
In the first example, the speaker neatly inverts the emphasis of the
unemployment figures to give the situation a new and more
encouraging aspect. In the second, the speaker has sufficient legitimate
power to warrant his explanation of his behaviour as being self-
22 Charles Antaki
legitimating. In both cases the explanation has the power to redefine

what is going on; and, of course, the new definition suits the speaker
rather better than the original.
We can pursue this feature into explicitly propagandistic literature to

see how the revelatory power of explanations is exploited. It is
common to see articles headed by questions (for example, Why world
food glut amidst hunger and starvation? The Plain Truth, July/August
1987) which promise to explain a very serious and multifaceted human
problem. The answers to such questions in propagandistic material are
often of little surprise to the reader.
The explanation is often a reprise of the general message of the

publication (in this case, human failings consequent on loss of
Christian direction). Reading this kind of publication one gets the
strong feeling that the normal process of explanation has been turned
on its head. Normally, the explanation giver describes something that is
known to be a problem to both explainer and audience, and then
proceeds to reveal something novel which explains it. In the closed
world of explicitly ideological discourse, however, it is the explanation
(it may be human sinfulness, the class struggle, or bad diet) which is
well known and agreed by explainer and audience, and what is new is
the particular problem that the writer has managed to use it to explain.
Seeing the partisan flavour of explanations in literature which is

easily identified as ideologically committed helps us to see the same
thing in ostensibly more neutral writing. The mainstream press is
aware of the attraction that revelation has for readers, and subeditors
24 Charles Antaki
are quick to exploit it in headlines and leading paragraphs. In the paper

I have open in front of me, I can see two examples which might
perhaps border on pseudo-explanation. Why Errol swapped easy sex
for love and Why has Britain deserted the Americans in the Gulf?
(both from The Sun, London, 3 August 1987). Unlike more blatantly
propagandistic publications, the explanations given in answer to these
questions are not completely predictable in detail. However, their
general gist can be foreseen. In a popular tabloid like The Sun, the
explanation of a pop stars new-found monogamy (I decided I wanted
love as well) is consistent with a policy of the sentimentalization of
human relationships, and the identification of civil service mandarins
as the cause of British faintheartedness (There is a telltale mark on the
whole shameful affair: the dead hand of the Foreign Office) is
consistent with its anti-intellectual, robust policy of Proud Britannia.
In certain extreme cases, there is no need for subtle analysis, as this
example about the Libyan embassy shootings of April 1984 shows:

(3) What happened at the Libyan Embassy yesterday was bloody and barbaric because the people involved are bloody and barbaric. (Editorial,
The Daily Mirror, London, 18 April 1984).
A speaker (or in this case, a leader writer) has used the definitional
opportunities of the explanatory form to prejudice the reader into
accepting a certain form of explanation. Once you have described an
event as bloody and barbaric then you have made it easy for your
audience to accept that the explanation is the bloodiness and barbarism
of your chosen target.
26 Charles Antaki
The revelatory power of explanations is something that has made

explanations and accounts of great interest to sociologically minded
researchers since C. Wright Mills. Explanations in this sense appeal to
a very broad range of people interested in the ways in which social
upsets and faults are repaired and glossed over. This aspect of
explanation is perhaps clearest in ordinary discourses of excuse and
justification, where revealing whats really the case has a specific,
usually material, motivation. Courtroom discourse routinely involves
excuses as to why defendants drove too fast, assaulted people, and so
on (see Cody and McLaughlins account in chapter 8). In the press, the
sports pages routinely feature the accounts of hapless sportspersons
and managers explaining away failures and disasters. Here are two
examples, again from the newspaper in front of me (The Sun, 3 August
1987). Tennis: Boris killed off by battler Brad: ... I had problems
with my arms, elbows and legs. Every day this week something has
been hurting. Athletics: What a disaster: ... He confessed: Im
mystified. The training has been going OK but there was no fizz when
I needed it today.

Explanations privately held
So far we have treated explanations as being offered by someone to

someone else or to the world at large. Seeing explanations as things
which achieve ends has oriented us to think about explanations which
are essentially to do with communication (for example, their power to
define realities and to reveal the truth). But there is another side to
the coin.
28 Charles Antaki
Explanations can be objects of interest in their private as well as

their public sense. The account a person privately holds about an event
will colour his or her attitude towards it and can, under certain
circumstances, have an effect on their behaviour. If I think that the
reason why my car is making a funny noise is that I am driving over a
funny road surface, I will carry blithely on without stopping. If you
think that the reason that you feel queasy after eating seafood is that
you are allergic to it, you will stop ordering it in restaurants. And so
on. If our mental apperception of the world includes personal accounts
of why things happened, then our feelings and behaviour are going to
be at least partly contingent on our ordinary theories. Once one has
allowed that, then it becomes interesting and important to examine the
cognitive machinery responsible for the explanations processing and
storage.
It is difficult to flesh out this observation with examples, since if
one pushes a private explanation on to the public stage it will acquire

certain features (of accountability, say) which it might not have had
when held silently in thfe individuals mind. This is not quite the case
with the peoples lay or expert understanding of physical, as opposed
to social, behaviour - for example, the behaviour of systems and
machines. Normally such explanations would be unlikely to be
transformed by being driven into the social arena. They will be given
in a context of accountability only in special circumstances - when
things go wrong, for example (see Reason, 1987 on the failures of
expert explanations in the Chernobyl nuclear disaster). However,
although there are clear overlaps of interest with explanations of
physical phenomena, I shall be staying within the realm of
explanations of social behaviour.
Suppose you were counselling a client who had just lost his job.
After a time, you were fairly confident that you had an accurate picture
30 Charles Antaki
of his private, internal account of himself: that it was his fault, that he
was no good at his work, and that he was generally incompetent. You
might well predict that he would be at some risk of depression. You
would be in agreement with theorists in the cognitive tradition in
psychotherapy, who make just this kind of claim - that the explanations
people have affect their liability to suffer emotional and mental
disturbance (see, for example, Becks 1967 rational-emotive therapy
and Abramson, Seligman and Teasdales 1978 helplessness theory).
The emphasis here is on the effect on people of the account or
explanation that they hold in their minds.
The same emphasis is current in more laboratory-based cognitive

theories which, although they do not set out to correlate explanations
with emotional distress, nevertheless aim to find how they are
processed and how they affect behaviour and judgement. As we shall
see in more detail later, the dominant way of thinking in this area is
called social cognition, and the jewel in its crown is attribution theory.
This is a loose federation of principles which revolve around the notion
that people seek explanations of the world around them, and that they
arrive at these explanations by a process of quasi-rational information
processing.
Traditions of research
The discussion so far has been based on observations on the one hand
about the definitional properties of explanations, and on the other
about their mental effects. The two features suggest that there are two
traditions of work on explanations. In one, the emphasis is on
explanations as they appear in communication. In
32 Charles Antaki
Table 1.1 Features of research on social explanation in two traditions
the other, the emphasis is on their mental representation and how it

affects individuals feelings and behaviour. Table 1.1 sketches some of
the differences.
On the one hand the theorist sees the explanation as a representation

of what the individual thinks, or, if no commitment is made to the
thought being an articulate one, to what is stored in the individuals
cognitive warehouse. On the other hand the theorist sees the
explanation existing in the space between the explainer and the
audience. On this latter interpretation the question of whether the
explanation corresponds to some fixed explanation in the speakers
head is not a straightforward one. We shall see the debate more clearly
as we expand a little on each of the sections mentioned in the table.
Types of explanation
As mentioned earlier, identifying what it is that makes something count

as an explanation is far from easy. Draper in chapter 2 shows
persuasively that, in the appropriate context, any utterance (or even
none) could be an explanation in some sense. If there is a feature that
34 Charles Antaki
all explanations reliably possess, it is that they should reveal something

of the explainers mind to the enquirer. This non-specificity seems to
make sense if we are to cope with the infinite variety of surface forms
a legitimate explanation might take.
Within the experimental social psychology tradition, a line quite

different from this freewheeling relativity has been taken. Here
explanations have been very specifically understood to be one thing:
causal reports. The influential writings of Kelley and of Jones and
Davis (and, to a lesser extent, of their mentor Fritz Heider) have
concentrated on the individuals solution of why problems. Jane
chooses to go to Harvard rather than Yale: why? Monica laughs at the
policeman: why? Attribution theory describes the mental processes
through which an individual must go in order for the answer to emerge.
The best-known model in attribution theory is Kelleys covariance

analysis model, in which the explainer computes the historical
association between the effect and its various causal candidates. The
candidate which enjoys the most reliable association with the effect is
taken to be the cause. So if Monica laughs at all policemen, has always
laughed at this policeman, and no one else laughs at this policeman,
then the policeman is excused, and the likeliest cause of the laughter is
something to do with Monica.
The shortcomings of such a view of explanation are legion and are

well documented elsewhere. For our purposes, we need only note that
the explanation in attribution theory is the attribution of a given event
to one of a number of (usually) given causes. It is a very closed picture
of causal explanation, let alone of explanation in general. It is closed in
the sense that the event for which the explanation is being sought is
preset by the enquirer (usually, an experimenter who poses particular
36 Charles Antaki
and non-negotiable questions to her or his respondents), allowing none

of the definitional work that is so powerful a feature of most normal
explanation. If asked, Why did Monica laugh at the policeman?, I
might, in normal conversation, ask if the questioner was sure that
Monica was actually laughing at the policeman, and not perhaps
giggling nervously, or even laughing with him; and so on. The
closed boundary around forms of expression is perhaps a consequence
of the kinds of research techniques that attribution theorists use; these
are, in their turn, a consequence of the cognitivist tradition in which
attribution theorists find themselves. In this tradition, the important
thing about social beliefs is not their content or their communication
but the way in which they are mediated by cognitive faculties.
It is probably no accident that the picture of explanation in

attribution theory is a close mirror of the form of explanation that
attributionists, as experimental psychologists, themselves favour.
The classic covariance model is no more than a self-conscious prose

version of the statistical model that Kelley and thousands of other
experimentalists use to analyse their own experimental data. Perhaps
attribution theorists have ascribed to other people the kind of analysis
that they use themselves to understand the world; though regrettable,
the practice is at least understandable. What is less understandable is
why they have been so reluctant to test out their theories of what
explanations are like in ordinary talk by looking at actual examples of
verbatim discourse. Had they done so at an earlier stage, the
communicative features of explanation would surely have found their
way into the attributional picture.
For the other tradition, the communicative features have always

been much more salient. Explanations are recognized as statements and
texts which offer some description of events that meet the enquirers
request - or need - for material with which to get a purchase on the
38 Charles Antaki
event at hand. It is a commonplace of certain kinds of social analysis anthropological, ethnological, ethnomethodological, and so on - that
explanations are things which reveal categories. To force the
attributional example into this kind of mould, you could say that
Monica laughed at the policeman separates out, for the enquiring
observer, categories of gender, power and resistance. A less contrived
example would be something like Its only a game, which serves as
an explanation on many levels - it makes enigmatic behaviour
intelligible and makes apparently improper behaviour legitimate, for
example - without any implication of causation in the attributional
sense. To a complete stranger to rugby football, Its part of the game
distinguishes a ruck from merely violent behaviour. The explanation is
parading a category in front of not only the enquirer but also the
inquisitive social scientist who wants to know how this culture
categorizes the world.
Questions addressed
The cognitivist programme features questions about the mental

mechanisms responsible for the selection, storage and integration of
social information relevant to the judgement of causality. Graduating
from the experimental social psychological tradition of person
perception and impression formation, attribution theory has maintained
the emphasis on cognitive process. The two best-known models of
attribution are deliberately and explicitly descriptions of the steps that
must logically be gone through before a judgement of causality is
made. These models have been taken as a working hypothesis about
the ways in which information is selected, stored and integrated, with
researchers tending to assume the basic ideas they embody and
concentrating on filling out particular gaps in the overall structure.
This has led to a huge volume of attribution work which has multiplied
findings at the expense of theoretical coherence.
Within the other tradition, there exists nothing like the umbrella of
40 Charles Antaki
theory and practice under which attribution researchers can so neatly

shelter. There is no single paradigm, and so neither an articulated set of
questions to be asked nor a common way of asking them. Linguists
will address questions in the pragmatics of explanation,
ethnomethodologists will address questions in ordinary reasoning, and
so on. Attribution theory has shortcomings in its limited view of
explanations, but there is no doubting the productivity of the paradigm
it offers its adherents.
The rivalry outside social cognition produces enthusiastic and

radical claims of a kind alien to experimental social psychologists. Two
of the most potentially far-reaching proposals at the moment are that
rhetoric is an appropriate method of argument for psychologists (Billig,
1987a) and that discourse analysis can take us beyond attitudes and
behaviour (Potter and Wetherell, 1987). Both of these, which I pick
for their very clear exposition and their determined targeting of social
psychologists, are represented in later chapters of this book.

Types of theory
Social cognition theories share a general family feature: the belief in a

stimulus-organism-behaviour view of persons. Psychological questions
are posed at the level of the individual, and more specifically at the
level of the individuals mental apparatus. Social cognition theories
involving explanation, reasoning and judgement have an extra feature.
They have a belief in the individuals fundamental rationality, although
they allow, or claim, that it is flawed. In attribution theory, the original
picture was of the individual as a naive scientist, not quite rational
enough to pass muster as a qualified scientist. Although the picture is
changing to some degree, the habit of treating people as poorer
versions of oneself is dying hard in social cognition. The type of theory
it generates naturally emphasizes the differences between the ordinary
42 Charles Antaki
reasoner and the experimenter: hence the ubiquity of such concepts as

the fundamental attribution error, the false consensus bias and so on
(for a full source of such theories see Nisbett and Ross, 1980).
Critics have tried to point out this lack of reflexivity, although

Harres proposal to Kelley to treat his subjects more like himself
(Harre, 1981: 141) probably sounds merely tasteless to social
cognitivists. To others, more in tune with reflexive notions and less
perturbed by the thought of finding themselves sharing their
respondents worlds, it is no more than a moderate and helpful
suggestion. Linguistic analyses of explanation would fail if the
investigators did not believe that they shared the respondents
linguistic resources. The investigator is able to articulate them more
clearly, to be sure, but they are necessarily the same resources, and any
deviation from a common linguistic background between investigator
and respondent makes interpretation more and more difficult.
Ethnomethodology makes a point of displaying the researchers

thorough familiarity with the respondents culture by deliberately
finding ways of upsetting it or, if that is not possible, of setting it in a
context where the normality of the encounter becomes strange and
alien; then its structure and its dynamics can be seen more easily.
Theories in this tradition of work on explanations are explicitly aware
of the roles that explanations play not only between explanation giver
and audience in the usual run of ordinary conversation, but also
between explanation-giving informant and investigative researcher, a
combination of roles not often imagined by social cognitivists.
Methods used
Laboratory manipulations and fixed-format response sheets are the

tools of the experimental social psychologist. This has led to a reliance
on - perhaps a fixation with - explanations that can be presented or
44 Charles Antaki
captured on paper briefly enough for respondents whose time may be

limited and whose attentiveness cannot be guaranteed. Explanations
are either independent variables (that is to say, they are presented,
suitably treated, to the respondent by the experimenter, who hopes that
different versions will elicit different behaviours) or dependent
measures (that is to say, collected from the respondent after she or he
has undergone some experimental manipulation). In either case the
standard procedure in attribution theory is to strip the content of the
explanation down to its causal kernel.
In attribution process experiments, respondents are presented with

information which they have then to combine to form an attribution.
(For example, the information might be about the relationship between
various actors and a given behaviour. The respondents might be
informed about the number of people who purchase Zad rather than
Sno-Wite washing powder; they then have to say why John bought
Sno-Wite rather than Zad.) In attribution studies which treat the

explanation as an independent variable, the respondents might be given
a training regime under which they have to learn a certain attribution
(for example, I am good at maths, but I have to try harder) and are
then followed to see if it has made any difference to their behaviour. In
both the dependent measure and the independent variable case, the
method of presenting or recording the explanation is premised on it
being non-negotiable. This has to be the case in the experimental
method, because the point is to make everything constant and
controlled except for the experimenters intended manipulations. The
experimental set-up is one that privileges a certain mode of thinking in
respondents - rational, documentary-oriented, non-frivolous - which
disallows certain kinds of linguistic behaviour, even if the recording
instruments were sensitive enough to pick it up.
For all its faults, the advantage of the social cognition tradition is
46 Charles Antaki
that it offers a coherent and powerful paradigm in which questions

about certain aspects of explanation can be cast. The experiment is a
legitimate tool and, when used sensitively, is persuasive. In the other
tradition, there is a far greater reliance on the persuasiveness of the
individual researchers interpretations of the data. This is not to say
that each researcher is speaking her or his own language, and can rely
on no common ground with the reader; far from it. Researchers rely on
a shared set of assumptions, a shared set of interpretative devices, and
a shared set of symbols to carry the argument forward. Using these
conventions and relying on persuasive skills is, of course, far from an
easy option. Billig (1987a) argues that the ability to sustain arguments
has become nearly a lost art, and that the modern form favoured by
experimentalists is a pale version of the full-blooded rhetoric practised
in antiquity. His contribution to this book (chapter 14) is an eloquent
example of scholarly persuasion.
Data
In all the interpretative disciplines, language - especially naturally

occurring discourse - is the essential starting point of any method. The
language may be transcribed down to minute aspects of stress,
intonation and paralinguistics, or may be accepted at the level of a
prose approximation to the rush of words in ordinary speech. In any
case, the researcher almost always makes a point of collecting the data
in their natural form. This means a great reliance on unobtrusive
recording, interviewing, document searching and so on, and a great
deal of concern with the procedures used to distil the huge volume of
language down to the categories with which the particular researcher
wants to work.
The source of data in the interpretative tradition emphasizes the

interest in explanations as products of communication. Data are often
extracted from records of public explanations, broadcast or published.
Data can also come from explanations as they occur in private
48 Charles Antaki
exchanges in mundane conversations overheard by the researcher, or

deliberately trawled by a research team logging naturally occurring
conversation. Almost any forum of public exchange can be a source of
explanations. They appear on every page of the newspapers, in every
column of Hansard or the Congressional Record, in conversations at
work or at home, in television advertisements, and so on and so forth.
The source of data for the experimentalist is less wide because she
or he has to bear in mind questions of control and sampling which are
less critical for the interpretative researcher. This has tended to mean,
until recently, that experimentalists limited themselves to the responses
made by the isolated subject in the laboratory. This is changing slowly,
and a few attribution studies are appearing which use written, or even
verbatim, ordinary language explanations. Nevertheless, the bulk of the
data in attribution theory studies, as in other areas of social cognition,
comes from the response sheet.
There are large and obvious differences between discovered,

ordinary language discourse and the preformatted responses available
to subjects in the laboratory. Perhaps the simplest pragmatic difference
is that naturally occurring talk is much more time consuming and
demanding to collect, analyse and make a case with. What social
cognitivists lose on the swings of validity, they gain on the
roundabouts of convenience.
Metacognitive considerations
It would be tempting to conclude by saying that each way of thinking

about explanations has a job to do, and that what are shortcomings to
one researcher are positive advantages to another. True as that
normally would be as a general principle in the social sciences, I think
it would gloss over an important consideration about work on
50 Charles Antaki
explanations.
It is a commonplace of conversation analysis that speakers make

substantial assumptions about what hearers will already know, and that
much of what is said is only sensible against a background of silently
agreed knowledge. That knowledge is updated and customized to the
particular interaction as it develops, making the resulting talk still more
opaque to an observer. A sentence ripped out of a transcript is all but
impossible to make sense of, and even a reasonably long extract is
thought by some to be unintelligible, as Draper reminds us in chapter
2.
The perfectly understandable examples from the press that appeared

at the beginning of the chapter seemed to be evidence to the contrary.
But they were intelligible because they were designed to be so; not
only are we aware of the conventions appropriate to reading and
making sense of newspaper writing, but the journalists consciously or
unconsciously understand that awareness and write accordingly. In less
rigidly structured talk there is a constant stream of negotiation and
trimming going on between speakers, making the overt form of any
given explanation an unreliable guide to what may be going on beneath
the surface.
Social cognition theories are poorly equipped to deal with elements

of explanations not locatable in the workings of a single persons
cognitive faculties. The strengths of the social cognition approach are
the in vitro techniques it can bring to the analysis of an explanatory
atom, or indeed any belief or attitude. On the question of the way the
52 Charles Antaki
atom connects to other atoms, and to the surrounding tissues in vivo, it

has less to offer. The study of social cognition understood as the
examination of the single individuals mental faculties will inevitably
fail to deal properly with explanations understood as joint processes.
The disenfranchisement of the ordinary speakers right to negotiate

meanings and reroute questions does make some kinds of empirical
question straightforward to answer, as we mentioned above. Social
cognitivists have had much to say about peoples use of various kinds
of information, the schemas they carry with them, and so on (see Fiske
and Taylor, 1984 for a comprehensive catalogue). The location of all
these features of explanation in the mental apparatus of one person
does tend to skew the attention away from the dynamics of explanation
in verbatim exchange.
Draper in chapter 2 will expand on the nature of everyday

explanation, and he will be showing that a number of its features
identify it firmly as the product of joint action based on mutual
knowledge. For example, the same request for an explanation can be
answered in a number of different ways, but only one will actually be
chosen for a specific exchange: it will be the one appropriate to the
specific need the explainer guesses the questioner has. This is true even
in the limited case of questions about causation. The example Draper
gives is: Why did the glass bounce rather than break? The answer
could be: Because it was tough; Because I always buy unbreakable
glass; Because it only fell a short distance; Because the floor is
soft; Because we are on the moon in low gravity; and so on. All of
those could be true, but the one that will be given is the one that the
speaker thinks the questioner wants or needs.
If the explanatory response is sensible only against a background of
54 Charles Antaki
the knowledge the speaker ascribes to the audience, then it is tempting

to think that what attribution theorists (as the prime social cognitivists)
do is to play the part of the audience themselves. So that there should
be no misunderstanding, they tell the explainer what they, the
requesters of the explanation, already know. This is the case with all
vignette studies in social cognition, but perhaps most transparently so
in the classic covariance experiments. Here a subject is asked to
explain why Monica laughed at the policeman given that she never
usually does so, but everyone who comes across this policeman bursts
out laughing, and so on. The explainer is told, in this as in other
studies, which element of the event the audience wants to be explained.
By making themselves the enquirers, social cognition workers avoid
the question of what metacognitive resources explainers need to make
sense of normally more elliptical exchanges in normal discourse.
Peoples information processing faculties are

complex and deserve the attention they are
currently enjoying. But peoples skills at
coding and decoding the metacognitions
behind mundane social explanations are
equally, if not more, complex, and to make
sense of them we have to look beyond
orthodox social psychological methods. If this
takes us into new ways of thinking about
people and how they ought to be studied, so
much the better.Whats going on in everyday
explanation?
Stephen W. Draper
Although there has been considerable discussion of the nature of

explanation in the philosophical literature, little if any of it (with the
possible exception of Wittgensteins Philosophical Investigations)
seems to address the nature of everyday explanation by lay people
(that is, what is ordinarily referred to by the word explanation). Most
deals rather with what counts as good or scientific explanation: that is,
it attempts to explicate intuitions about what counts as the best quality
explanation. In contrast, this chapter tries to make some points about
ordinary explanation, including both private explanations and accounts
offered to others, and to outline the implications for empirical work. It
does so mainly by reference to common sense and to examples which,
like those used by linguists, aim to elicit the same intuitions in almost
all readers. It will also perhaps show the authors training in artificial
intelligence in the attitudes to language and mind implicit in the
arguments, though no technical appeal will be made.
The chapter makes four main points about explanation:
Explanation is a broad concept referring to many kinds of

statement, which often have nothing to do with causality, reasons or
accounts.
2 Surface cues and markers (for example, words like because) have
no simple correlation with either the presence or type of
explanation (which makes it hard to do any mechanical analysis).
3 Explanations offering reasons may concern causality or reasons for

belief or reasons for a speech act.
4 Explanations occur within the complex hierarchy of goals involved

in any utterance, and may be intended to satisfy any of a variety of
not necessarily mutually exclusive purposes.
60 Stephen W. Draper
Explanations may contain any kind of
information
Finally the conclusion notes some consequences of these features of

everyday explanation for the possibility of doing useful empirical
work, and for studies of peoples causal beliefs (cf. attribution theory)
and for studies of justifying oneself (cf. giving accounts).The first
point to recognize is that in everyday life almost anything may, in the
right circumstances, count as an explanation. The Oxford English
Dictionary tells us that explain means make known in detail; hence
any information at all may form the content of an explanation. Thus
explain can be prefixed to most kinds of question, as well as to
referring expressions:
information
Explain whom you went to see.
Explain who the queen is.
information
Explain when the boats leave.
Explain where you went.
information
Explain what a knights move in chess is.
Explain the rules of that game.
information
It is not always equivalent to a tacit why. For instance, while a

jealous husband saying Explain whom you went to see may be
mainly asking for a justification or motive, a colleague asking the
same question about a visit to a firm is probably asking for
information about contacts, their position in the firm, and so on.
Certainly Explain who the queen is does not seem to reduce to a
why question, and is nothing to do with causality, nor necessarily
with giving a socially satisfactory account of oneself.
A request for an explanation is a member of the same family that

all questions come from: that is, it is a request for information, and
like questions, it may refer to any kind of subject matter. As we shall
see later, explanations can vary in subject in another way too: they
may be not only about the aspect of the world under discussion, but
information
also about grounds for belief or about the reason for saying something
at a given point in a conversation - that is, about the conversation
itself.
The inquirer does not know whats wanted
An explanation is typically informing about

what kind of answer applies, as well as giving the content itself. For
example: a manager returns from lunch to find the office in an uproar,
and asks for an explanation (Whats going on?). The first and major
part of the information sought will be conveyed not by the detailed
content, but by its type. Is the staff misbehaving? Has vital equipment
broken down? Has senior management visited a crisis on this office?
The feature of explanations is to convey type and structure as well as
content.Whats going on in everyday explanation? 17 Explanatory
descriptions
As Appelt (1985) showed, explanatory functions may lie behind the

choice of ordinary-looking noun phrases scattered throughout a
discourse. For instance, if I say Im going to the church in
Beahampton, the phrase in Beahampton is necessary to specify
which church I mean. However, if I say Im going to climb Helvellyn
in the Lake District, the phrase in the Lake District is unnecessary
for specificity because there is only one mountain called Helvellyn; it
may have been included so that the hearer does not have to inquire
whereabouts Helvellyn is. Similarly if I ask a visitor to pass me the
corkscrew in the second drawer, I may not be clarifying which
corkscrew I mean but rather telling him where the only corkscrew is communicating new information. It seems natural to say that when I
added in the Lake District this was an explanation (a case of making
known in detail), as was in the second drawer. This is connected to
the issue of explaining what you mean when you say it, which is also
discussed later in the section on giving reasons for speech acts:
creating spontaneous explanations is a necessary part of carrying on a
successful conversation. Explanations, then, are widespread in
discourse, and in general are not marked out syntactically.
Surface cues
How are explanation requests expressed linguistically, and can a

categorization of surface cues be related to a categorization of types of
explanation content? This section proposes the answer no. For
instance, there is nothing about the form of noun phrases such as in
the second drawer that labels their function as explanatory.
More generally, it is widely recognized that the meaning of

sentences can depend in part on the context and that things may be
left unsaid if the hearer can fill them in, both syntactically (ellipsis)
and at the level of whole propositions (Gricean cooperativeness
between dialogue participants entails not wasting time by stating the
obvious: Grice, 1981). However, when one is making or replying to
explanation requests, these phenomena predominate to such an extent
that the words spoken, like the tip of an iceberg, hint at rather than
specify what is meant.
Paraphrases
Whats going on in everyday explanation? 69
Many peoples first idea about analysing question types is to

characterize them by the lexical items that appear in them, for example
who questions, why questions, and so on. Amoments reflection
however shows that words like why and how can paraphrase many
different questions expressed using other supposedly contrasting lexical
items, as the following selection of examples from Lehnert (1978)
show:
How often does the bus leave? (with what frequency)
How do you like New York? (what is your attitude)
How did you get here? (by which means)
How did you get here so fast? (what enabled you)
How did the milk get spilt? (who or what caused the spill)
How come youre angry? (what combination of factors)
How do I open this? (what method)
Just as the question word (and in fact the whole sentence) does not
determine which kind of answer is wanted, so conversely in many
situations question words can be interchanged without changing the
meaning. For instance, walking into the kitchen and asking Why is the
milk spilt? and How did the milk get spilt? amount to the same thing.
This is not genuine linguistic paraphrase because there are situations
where the questions would mean different things; but in many cases
they amount to the same thing. Thus question types can clearly not be
based on question words. Likewise explanations do not usually make
explicit the relationship of the answer to the question: because is
similarly ambiguous, as we shall see in a moment.
Analysis of questions about causation
Even if we restrict ourselves to questions with why in them, and even

if we further restrict ourselves to those cases where the question
concerns causality, a typical question is still ambiguous in (at least) two
independent ways (this account follows Van Fraassen, 1980, chapter 5).
The first concerns the contrast class, which can be thought of as an
implicit rather than clause. Why did John go shopping in Beirut?
might mean why John rather than someone else, why did he go
shopping rather than do something else, or why in Beirut rather than
somewhere else. The second issue is that of the implicit relevance
relation. The basic idea here is that all events depend not on a single
cause but on an indefinitely large set of enabling and contributing
conditions, and relevance relation refers to which subset of
preconditions to the event are in question (in focus). Why did the
glass bounce rather than break?: because it was designed to be tough,
because I always buy unbreakable glasses, because it fell only a short
distance, because the floor is soft, because we are on the moon in low
gravity.
This example illustrates firstly the multiple causal antecedents of

ordinary events, any of which may be relevant, and secondly
that because does not specify which of these is being focused on even
in a given physical situation: this depends on which aspects the
discussants are attending to. Thirdly, if taken in isolation, the question
is naturally interpreted as an open-ended explanation request, in which
the inquirer wants to be told which antecedent to focus on - is implicitly
asking which has an unexpected value (is the glass unusually tough, is
the gravity unusually weak, is the floor unusually soft, and so on).
However in some contexts, there may be an implicit relevance relation
to constrain this, independent of the inquirers knowledge. If a
restaurant manager asks a waiter why a table is dirty, she is probably
not asking which customer did it and how, but why the waiter hasnt
cleaned it up. A parent asking the same question, however, probably is
asking who made the mess.
These considerations show not only that a typical question leaves

more unspecified than it makes explicit, but that there is little sense to
the idea that there is ever one true explanation: it all depends on
perspective. When the restaurant manager attends only to the waiters
negligence as a cause of the tables dirtiness, she is in one sense
exhibiting an attribution bias; but it doesnt imply any kind of false
theory about causation. When she goes home and asks her child about a
dirty table she is likely to apply a quite different bias.
Because
As we shall see later, the subset of explanations that seems to have

attracted most attention is that concerned with giving reasons. These are
often marked by words like because but, just as why and how are
ambiguous, so on the one hand because can mean various things and
on the other it can be substituted for by a number of other words or
none at all.
The previous section showed how because can precede any of the
multiple causal antecedents of an event without distinction. The next
section shows how because can occur not only in explanations that
express cause but also in those that give reasons for belief (Mansell
must be going to win because thats what all the commentators say) or
for a speech act (Whats the time, cos Ive got to dash?). Conversely,
because and related connectives like so can be missing from
utterances that are clearly (causal) explanations, for example Why is
he limping? He tripped while staring at a passing girl, or Why is he
off work? He swerved to avoid a child and skidded into a tree.
Causation is often implied by connectives like and and while.
Thus because probably does always signal the presence of an

explanation, but does not tell you much about what kind of explanation.
Furthermore neither it nor the set of related connectives is a reliable
marker: a search of a transcript for their occurrence will not pick out
anything like the complete set of the explanations present.
Explanations that offer reasons
Only a subset of explanations is concerned with giving reasons for

something. However, this subset is the one that most of the literature
has been concerned with. As we shall see, it includes giving causes, but
also other kinds of explanation.
Information about sources and about the world In considering

explanation, it is useful to distinguish between two kinds of content
held in our minds: ideas about the world, and information about the
grounds for those beliefs (typically the source or authority for them).
This is important because so many of the ideas (provisional beliefs) on
which we base our decisions are held on more or less uncertain
grounds, and when a conflicting piece of putative information is
encountered we can try to decide between them on the basis of their
associated sources (their relative authority) if these are known. For
example, suppose I believe that tomatoes are cheapest in Sainsbury and
then a friend tells me they are cheaper in Tesco. If mine is a general
belief about the relative prices in the two shops while his is (he says)
based on a recent price comparison for tomatoes in particular, I shall
change my belief to follow his; whereas if mine is based on a price
comparison, I shall probably ask him when he did his.
In artificial intelligence, this idea has been explored under the name
of truth maintenance or reason maintenance (Doyle, 1978;
McDermott and Doyle, 1980). Although we do not retain the source or
authority for everything we believe, it is a widespread and important
kind of knowledge. In some cases an explanation may consist entirely
of the citation of an authority: Thats what the doctor said, or It was
on the news. This is a general phenomenon, which reflects the socially
distributed nature of much knowledge, including scientific knowledge.
Putnam (1975) has pointed out that concepts such as gold or water
have meanings that are ultimately grounded in the specialist knowledge
of a few, to whom the rest of us could appeal in principle. Gold for
instance is defined by its chemical properties, and that is what we all
mean by gold, even though most of us dont know what they are.
Knowing what authority to appeal to for more information is itself an
important kind of information stored in the mind, and sometimes
conveyed in explanations.
This technique is generally needed sooner or later to close off

chains of explanation requests by referring to some public, authoritative
information source. This matches the reality that knowledge is socially
distributed, and solves the problem of infinite regressions of why
questions (such as children of a certain age tend to embark on).
Retaining knowledge of the source of an idea also allows rival

explanations (theories) to be entertained, transmitted and discussed.
Van Fraassen (1980, chapter 5) reminds us that scientists often talk of
some theory explaining one observation but not another: explanations
have a relationship to what they explain that does not intrinsically
involve truth (although very often our interest in them ultimately does).
Similarly an individual might say I found an explanation for that today,
but I dont know if its the right one, thus showing that constructing,
like eliciting, an explanation is distinct from accepting and believing it.
Explanations then may contain either an elaboration of propositions
about the world or reasons for believing a proposition. An important

kind of reason is an authority - who asserts the proposition.
The structure of argument
Retaining reasons for belief is what makes rational argument and

persuasion possible. Toulmin (1958) has described some of the
structures underlying arguments (and Voss offers empirical evidence in
chapter 6 of this book). These consist of the roles that propositions play
within an argument - the relationships between them, rather than their
logical status. For Toulmin, a claim (an asserted proposition about the
world) is typically supported by data, and the warrant is the general
rule by virtue of which the data are a support for the claim; for
example, Petersen is not a Roman Catholic (claim) because he is a
Swede (datum) and very few Swedes are Roman Catholic (warrant).
Of Toulmins seven roles, three commonly appear after because as
reasons for belief: data (as in the above example), warrants and
backings. A backing is a reason for accepting a warrant: it may be a
recursive justification (treating the original warrant as a claim to be
justified), or it may be an appeal to authority or some other reason for
belief. To continue the example: Very few Swedes are Roman
Catholic, as you can see from any almanac.From now on I shall use
warrant in an extended (and more usual) sense to refer to any support
for a claim - any reason for belief. Broadly speaking there are three
kinds of warrant: those connected with the consistency of a claim with
other accepted beliefs (this is the domain of logic in its broadest sense,
and was Toulmins main focus); those that introduce new facts about
the world as support; and those that state the grounds for accepting a
claim as a fact without reference to consistency, such as authority or
personal experience. In artificial intelligence, truth maintenance is
mostly concerned with the detection of inconsistencies within a belief
system, and methods of resolving these by comparing the origins of the
claims (for example a default rule) for reliability. Arguments between
people in addition involve determining what beliefs are held in
common, and perhaps trying to introduce new facts that may have a
bearing on the subject. The type of a warrant is not usually explicitly
marked; for example, A because B might be said when B was
Newton said or I saw it myself or it is a corollary of the law of
conservation of energy or C caused it.
Warrants, then, are a major type of explanation. Since they

sometimes consist of an assertion of causation, they can simultaneously
be a warrant and an assertion of fact about the world. Warrants
(grounds for belief) and causal statements are not mutually exclusive
categories; for example, You wont be able to use the car because its
run out of petrol. A cause is a first-rate warrant.
An example of ambiguity
This example is adapted from an utterance reported in Antaki and Naji

(1987). It illustrates the ambiguity between cause and warrant in an
explanation, and shows how, in the absence of a detailed knowledge of
the beliefs of the interlocutors, we cannot tell just what kind of
explanation is being offered. Following the utterance, I list alternative
interpretations of its meaning and status as an explanation. This will
emphasize the multiplicity of interpretations of an utterance taken from
a protocol in the absence of quite detailed knowledge of a particular

kind of context: the beliefs of the interlocutors.
The utterance is:

Hes a good producer because he used to run 'Open Door for the
BBC
The alternative interpretations are as follows:
1 Causation: making Open Door caused goodness by training him.

(This presupposes they both know hes good, and answers How did
he become a good producer?)
2 Warrant based on causation (from programme to producer): you can

believe he is good because making Open Door caused goodness by
training him. (This, and the following interpretations, answer why
should I think him a good producer?) The causal premise - the fact
that such experience causes goodness in the producer - is
presupposed.
3 Warrant based on personal witness (of a fact about value) plus

causation (from producers goodness to choice of producer): you
can believe he is good because the speaker saw Open Door and
saw that it was good; therefore its producer was good because
goodness in a producer causes a programme to be good. Again the
causal premise (though different from 2) is presupposed.
4 Warrant based on public knowledge: as 3, but using the assumption

that everyone, not just the speaker, knows Open Door is good.
This interpretation would be more likely if Open Door were a
well-known and widely admired programme.
5 Warrant based on authority (about a fact about value): Open Door

is evidence that the BBC think hes good. That is, you can believe
he is good because the BBC are authorities on which producers are
good, and they must think hes good because they only assign good
producers to important programmes, and Open Door is an
important programme (presumed public knowledge).
You can believe he is good because Open Door is a good job, but
hes moved onwards and upwards, so hes even better. This is a
different inference path, which presupposes that his reassignment is
a promotion, but doesnt require any knowledge of his subsequent
jobs or their value. It presumes that Open Door was evidence of
his worth in one of the above ways.
7 In addition to these ambiguities is the possibility that the speaker is

deliberately exploiting them: not knowing which the hearer will
accede to, he may be deliberately offering all of them. This kind of
logical haziness is familiar in everyday justifications like Alecs a
good chap because he went to Cambridge, where the speaker wants
to imply that Alec must be good because Cambridge only accepts
good people, and that Cambridge gives superior training and
therefore makes anyone good, and even that wanting to go to
Cambridge shows good values.
Most of these interpretations involve an appeal to causality (as well as

to other things), but to different implied causal rules. In one his
goodness caused goodness of the programme; in another the
programme caused goodness in him by giving him experience; in
another his goodness caused his assignment as producer to an important
programme.
Note how these possible interpretations each involve several other

tacit propositions, as well as having a different status in terms of the
analysis developed in this chapter. If we knew which of the possible
ancillary propositions were held by the interlocutors, we might be able
to deduce which of the interpretations applied. For example, if we knew
that both interlocutors already knew he was good, then the first
interpretation (explaining an accepted fact) would apply. Or
alternatively, if the speaker were well known (to the hearer at least) as a
fanatical admirer of Open Door, then the third interpretation might be
salient. If the interlocutors were television insiders who however

regarded the BBC management as no judge of worth, then
interpretations based on the BBCs authority would be unlikely and
those based on presumed common knowledge of a programmes worth
would be likely.
Thus this example illustrates both the ambiguity (to outside

observers) between explanations asserting cause and those asserting
reasons for belief, and how the resolution of that ambiguity depends on
the knowledge of the interlocutors (which is typically not recorded in
transcripts). Explanations are relative to what is known already.
Reasons for speech acts
We have explored two kinds of reason, but there are in fact three: cause
(that is, another fact about the world that is a reason for the occurrence
to be explained); warrant (a reason for believing a claim); and
justification for a speech act (a reason for making a particular
utterance). An example of the third is:
When are you leaving? Because I need a lift.
Here the reason given is an explanation for the speech act - in this case
asking a question. Although it is possible to argue that the need for a lift
caused the question, or that the explanation was given in order that the
speaker not appear to be given to irrational utterances, the real issue is
that without the explanation the hearer will probably not know how to
respond to the question, because most responses depend on the hearer
being able to recognize the speakers intention (Grice, 1957).
The issue is not to give more detail about the world, or to give
reasons for believing some assertion about the world, but to explain
what is being done (or attempted) here and now in the conversation. In
the example, without the explanation the hearer might answer When
are you leaving? by I dont know, or object to the abrupt change of
topic, or even take it as an insulting hint that he should leave; whereas
with the explanation he can understand the question as an indirect
request for a lift, show (if he wishes) a willingness to negotiate the time
of departure, and so on. Thus the explanations function is to help make
the conversation successfully cooperative.
A problematic example
In everyday conversation (as opposed to philosophical analysis) these

three types of reason can seem to blend into each other. For instance,
the utterance
Its difficult because its hard to do two things at once.
could be viewed as:
1 A causal statement: doing two things at once causes difficulty for

me. This treats difficulty as an objective state of affairs which we all
perceive directly and similarly. On this interpretation an objection
would be: You can walk and talk, so not all pairs of activities cause
difficulty.
2 A warrant for my calling something difficult. This treats difficulty as

a belief about the world which has to be supported by indirect
evidence. An objection might be: You dont really find that difficult
because you never complained before, and you were humming
while you did it.
3 A reason for calling it difficult within this conversation, that is

defining how the term difficult is being used here. This treats
difficulty as a personal judgement whose utterance needs to be
explained for the purposes of this conversation. An objection might
be: You cant call that difficult if in the same document we tell
people that typing and checking spelling simultaneously is routine.
Such ambiguity applies to many common assertions about what is hard,

attractive, nice, long, and so on. The difficulty of analysing such
commonplace statements may indicate that whether an assertion is
true or about the real world is often not of concern to interlocutors; it
may often be more important to establish a way of speaking that both
accept and understand for the moment. Its a long way to Glasgow
because the journeys boring, but Majorca seems quite near. How long
is long? If this view is correct then analysis is difficult not only because
surface forms do not clearly mark the type of reason being offered, but
also because the type (at least with concepts like long) may not be
distinct in the interlocutors minds.
Conversational goals
Explanations occur within dialogues, and as such are embedded in the

hierarchy of goals which underlie utterances - a view of language
developed in the field of artificial intelligence by Power (1979) and
Cohen and Perrault (1979) among others. On this view, I may wish to
know what the time is, and to achieve this I get you to tell me, and to
do that I ask a question, which in turn involves a number of decisions
and actions such as beginning a conversation and seizing a suitable
moment for asking. Many phenomena can be located in this hierarchy
of goals, including topics, speech acts, and turn taking. Note that what
can be described as a single action may serve goals at a number of
levels simultaneously. For instance, in catching a strangers eye and

saying Do you know the time?, I simultaneously open a conversation,
take a turn, announce a topic, ask a question (the direct speech act), and
imply a request to be told the time (the indirect speech act).
There are (at least) three issues about the goals directly and
indirectly associated with explanations. Firstly, explainers may or may
not believe the content of their own explanations to be true: for
instance, when scientists discuss the relative merits of alternative
theories, or when someone reports an explanation asserted by a
thirdparty. Secondly,evenwhentheexplainerbelievestheexplanation, the
inquirer may or may not accept this (the explainers) valuation of it,
that is, be convinced. (This is the gap between illocutionary act and
perlocutionary effect in the terms of Austin, 1962). Thirdly, the enquirer
may or may not be concerned to learn more about the world (that is,
intend to accept the explainers valuation); the inquirer may just want
to learn about the contents of the explainers mind, for example when a
teacher asks a pupil. Thus explaining is an illocutionary not a
perlocutionary act, and a request for explanation does not in general
imply a request for a perlocutionary effect.
Thus Explain X to me does not necessarily mean Try to convince

me but may instead (for example when uttered by a teacher) mean
What do you know about X?. Its root meaning is something like
Inform me what you think and believe to be true about X (perhaps
including reports of possible theories and the beliefs of third parties).
In a particular case it may additionally imply a request to inform about
the world by convincing the inquirer; this will probably affect the
detailed content and form of a satisfactory response and so must be
inferred by the explainer from the context, but it is not part of the root
meaning. The same distinction applies to pedagogic and sincere
questions. Both teachers and students give explanations. Their overall
goals are opposite (complementary), but their immediate goal (used as

a means to their overall goals) is the same: to expose part of their minds
to the other.
Thus even though Elizabeth explained calculus to him is used to

imply that Elizabeth thereby changed his beliefs and knowledge, it does
not necessarily imply this. There is nothing problematic about saying
Elizabeth explained how she got the black eye, but no one believed
her. Thus when you identify an explanation in a transcript you cannot
necessarily conclude that either participant believed it or intended it to
be believed.
Effects and intentions
We must look at the relationship of goals, actions, and effects. Most

things we do have multiple effects, only one or a few of which were
intended - that is, were the reason for our taking that action. For
instance riding my bike to work gives me exercise, saves the train fare,
makes me warm, and splashes mud on my clothes. Furthermore most
goals can be achieved by any of a number of actions: riding a bike is
not the only way to get warm, to go to work, or to get splashed with
mud. The relationship of goals to actions is therefore many to many,
since a given goal can be achieved by more than one alternative action,
and a given action could be serving any of several goals (one for each
effect).
This means that when you observe an action, you cannot reliably
infer why it was done - the goal of the agent. Nevertheless such
inference is both widespread and necessary in daily life. As scientific
observers we must do it too, but we should be aware of the likelihood
of error.
Explanations have multiple effects, and may therefore be serving

any of a number of possible goals. The only essential aspect of an
explanation is that it purports to expose to the hearer (H) a part of the
mind of the speaker (S). That part may or may not be believed by S (it
may only be a possible explanation out of a set of alternatives). Seeing
it may or may not convince H of some idea about the world and thereby
inform H. Thus even if H takes the explanation to be a true
representation of Ss mind, it may or may not be intended as or taken to
be a belief of S, and (independently) a fact about the world. Further to
all of this, H may draw some conclusions about S - that S is
knowledgeable, rational, defensive, showing off, helpful and deserving
of gratitude, and so on. In different cases, different amounts of these
chains of effects will occur; even when they occur and S expected
them, it still may not be the case that S intended them and spoke in
order to achieve them, just as I do not ride my bike in order to get

muddy even though it is an expected consequence. Thus while
explaining how something works might be done to impress you (for
instance in showing you round my place of work), it might be done just
to help our joint activity along (for instance explaining how my bottle
opener works so you can use it). Estimating what social effects are
likely or intended requires knowledge of the situation and of the
participants: it will probably not be flagged in the words used.
Conclusion
This chapter has surveyed from a common-sense viewpoint the topic of

ordinary explanation. In general, explanations include any kind of
making known in detail: for instance, the phrase Helvellyn in the
Lake District might have been chosen to explain where Helvellyn was,
and two moves forward and one sideways is a (poor) explanation of a
knights move in chess. Studies of explanation are typically interested
in only a small subset of kinds of explanation: for instance, philosophy
of science tends to concentrate on causality and the relationship of
hypotheses to data. Social psychology is mainly interested in how we
explain peoples behaviour to ourselves and others, and in how these
explanations appear in utterances. This chapter accordingly
concentrated on these aspects of explanation, but it should not be
forgotten that they comprise only a small part of explanation in general.
Causes of human action are only part of causality in general,

causality is only one kind of reason, and reasons are only one kind of
explanation, so cases where human behaviour is the explicit subject

form only a subset of explanations. However, in principle we can
always ask the question of why the speaker is offering an explanation.
Explaining, like speaking, is a social action and reveals personal goals
which are frequently social in nature: I may be explaining to gain
credit, to appear rational, to convince you of something, or to change
your view of my behaviour. Thus as observers we can use every
explanation as the occasion for an enquiry about the social function of
the explanation, but we must recognize that there may in fact be no
social motive behind it - although conversely it might be socially
important without this being intended by the speaker.
All studies of explanation based on records of utterances, however,

face two basic problems. Firstly, the surface structure
of natural language offers few cues about the presence or type of

explanations. Secondly, what is stated is selected against a
background of what is mutually known by the interlocutors: there is
usually no point either in stating what is known or in stating
something which does not attach to what is known (for example, in
using terms like Helvellyn without explanation). However, without
knowing that unstated background we cannot understand the role and
status of what is stated, as the multiplicity of possible interpretations
of the Open Door example illustrated. Explanations are implicitly
but intrinsically interactive in character in the sense that they are
constructed to build on and add to what the other already knows: thus
what gets stated in an explanation is crucially dependent on what is
already shared knowledge, and cannot be understood by an observer
without that. The difficulties this makes for work on transcripts
deserve stressing.
Problems of transcripts
Linguists who study syntax work on the basis of isolated sentences

and judgements of their grammaticality. This is valid because such
judgements are consistent among almost all who read the examples,
and because they turn out not to depend on context; that is, a shift in
context does not usually make someone change their mind about
whether a sentence is or is not grammatically correct. However, this
does not hold so well for judgements of meaning (for example,
phrases like colourless green ideas can begin to seem meaningful in
the right context), and not at all for the kinds of interpretations of role
inherent in looking at explanation. Consequently isolated passages
taken from a transcript cannot be reliably analysed, as Goffman
(1981:32ff.) discusses and as the examples above illustrate.
This is basically because a large part of the origin of shared

meaningcomesfromknowledgesharedbytheinterlocutors,partlyonthe
basis of earlier utterances (so isolated sentences do not carry all aspects
of their meaning explicitly), and partly from special knowledge they

may have of each other which the analyst may not share. One
consequence of this is that you cannot do surface-driven counts of key
words like because, since the function of an assertion is likely to be
not marked explicitly but clear to the interlocutors on other grounds.
Another is that you cannot expect to understand in detail

eavesdropped conversations where the two speakers know each other
better than the experimenter does. For instance an analyst could not
tell whether the phrase my parents house in London was an
explanation of where the house was or a simple description clarifying
which of the parents houses was being referred
to, whereas the participants would. Thus recording telephone

conversations between friends will probably not be fruitful, while on
the other hand conversations between strangers would put the analyst
on the same footing as the participants. Getting agreement among a
panel of analysts does not in any way address this problem: if the panel
are all constructing an interpretation based on general knowledge of the
culture but ignorance of the particular circumstances of the
conversation, then they may well invent the same interpretation, but
still not understand what it meant to the participants. This is like asking
a panel what the most salient meaning of a word like round is and
then using that to analyse a transcript of a boxing commentary.
Beliefs about causality
If you are interested in peoples beliefs about causality (as attribution

theorists are), then a study of the subset of explanations giving reasons
is relevant, although causes will be mixed with reasons for belief (for
example, authority or other evidence) and reasons for speech acts (why
did the door slam? Because were trying to hold an exam). As the
Open Door example illustrated, which causal belief was implied by an
utterance may only emerge from a full explication of the rationale
behind a given interpretation. Thus while there seems to be a rich
potential for exposing beliefs about how peoples attributes and actions
affect each other, these are not made explicit in ordinary speech.
Although difficult, it seems conceivable that a technique might be
developed to draw extended explanations from people (perhaps by
asking them to explain and justify spontaneous explanations that they
have given) which would expose a chunk of their causal belief
structure.
Accounts
Similarly, if you are interested in the accounts people offer of themselves, then the above analysis offers various starting points. Firstly,
since any explanation is a disclosure of ideas in the explainers mind,
then any explanation might serve as, and be intentionally offered as, a
reason for seeing the explainer as rational, that is as holding views
which are coherent even if the inquirer (or even the explainer as well)
does not believe them. (Only the context can determine whether a given
explanation is being offered with this as its aim.)
When the subject of the explanation is an action of the explainer,

then it is the explainers behaviour rather than rationality that is at
issue. The distinction between excuses and justifications developed by
Cody and McLaughlin in chapter 8 may be linked to whether the
explainer is offering a reason that the explainer does not believe is
necessarily valid or that will not be accepted by the inquirer, or whether
the explainer does believe it and hopes to convince the inquirer that the
action was in fact the correct one to take. This is parallel to the
distinction between scientists discussing possible viable explanations of
a phenomenon, or expounding one they are committed to.
Finally, the distinction between causes and warrants as types of

reason is related to whether, in giving an account, the explainer is
presented as an object being acted upon by causes (I was forced to do
it) or as making a choice for a reason (It seemed best to do it because
the baby was crying so hard).
Note
The writing of this chapter was supported by SERC fellowship

B/ITF/94. The inspiration was in large part due to the two Alvey
workshops on explanation held at Surrey University in March 1986
and January 1987.PART II
STRUCTURE AND CONTENT

3
Identifying attributions in oral and
written explanations
John H. Harvey, Dawn C. Turnquist and

Gina Agostinelli
Attributions refer to peoples understandings of themselves and their

environment; more broadly, Attribution is part of our cognition of the
environment (Heider, 1976: 18). Someone who has just passed a
driving test at the first attempt may attribute his or her success to good
fortune; another might attribute it to a happy choice of driving school;
a third might attribute it to their own natural driving talent; and so on.
Attributions such as these are often made about matters of moment in
our lives.
Most attribution research has involved the use of structured, Likert

response scales designed to examine the extent to which people
attribute causality, responsibility or disposition to self, other or
environmental entities. Recently, however, a strand of work has begun
focusing on unstructured techniques for probing individuals
attributions. Attributions identified by these methods, termed free
response or unstructured attributions, presumably better reflect
common-sense ways of understanding people and events because they
are not influenced greatly by the format of research questions. Further,
these methods allow the study of attribution in instances when
traditional measures are not easily administered (for example, resistant
populations, sensitive topics).
116 Harvey, Turnquist and Agostinelli
This chapter is concerned with how such attributions are investigated.

Issues in the measurement of unstructured attributions will be
presented, followed by examples of oral and written accounts for
combat stress. Using these examples, we will describe a method for
identifying and quantifying attributions, and will conclude with a
discussion of the advantages and disadvantages of investigating
unstructured attributions. But first, a brief review of attributional
concepts is in order.Theoretical background
Interest in attribution stems from its crucial role as a mediator of

perceptions, emotions, motivations and behaviours. Indeed, how
people see cause and effect has implications for or may influence their
interpersonal relations (for example Fincham, 1985; Harvey et al.,
1982; Yarkin, Harvey and Bloxom, 1981), their psychopathology (for
example Brevin and Antaki, 1987), their response to psychotherapy
Identifying attributions in explanations 117
(for example Antaki and Brewin, 1982; Harvey and Galvin, 1984),
their decision making (for example Jones and McGillis, 1976) and
their adjustment to illness (for example Janis and Rodin, 1979; Taylor,
Lichtman and Wood, 1984). Further, the act of confiding, which often
involves causal explanations, has been linked to physical health
(Pennebaker, 1985). The confiding process allows an individual to
reorganize and assimilate a traumatic experience, apparently reducing
stress on physiological systems and improving health. Such a view is
compatible with the presumed merit of attributional activity.
Attributions may serve many other functions, including enhancement
of ones perception of control, preservation of self-esteem, presentation
of a particular picture of the self, and emotional release (Forsyth, 1980;
Harvey and Weary, 1984).
Under what conditions do we make attributions? Spontaneous

attributional activity is particularly evident after traumatic experiences
and life events, presumably because one is trying to understand or to

control these changes (Harvey and Weary, 1984). Other research has
demonstrated that individuals make attributions in situations when
their expectancies are disconfirmed (Hastie, 1984) or when an
achievement outcome is attained. However, attributional activity may
occur when one must understand any state or make any judgement, and
thus most situations can elicit attributions.
How do we select a particular attribution? There is an infinite

number of potential explanations when one is faced with an event,
behaviour or internal state to explain. All of these explanations are
likely to be retained as alternatives as long as they are consistent with
the evidence. One may rule out many alternatives as one searches for
evidence, but there may still remain more than one rival explanation.
According to Kruglanski (1980), at some point an individual stops
generating hypotheses and attains closure on a given belief. This
frozen belief is then no longer evaluated against alternatives or

competing pieces of evidence, and one becomes more confident of its
validity. Thus, according to this view, one may expect to hold many
spontaneous attributions early in the process, but retain only one after
freezing.
What types of statements might we consider attributions? While

some authors limit the term attribution to judgements of causality, we
suggest that attributions may be conceptualized as statements about
such states or relations as causality, responsibility or personal
dispositions. Similarly, Brewin and Antaki (1987) have classified
attributions into four typologies: (1) description or labelling, (2) moral
responsibility, (3) causality and (4) self-presentations.
For any type of attributional statement, several dimensions have

been identified. Firstly, an attribution may be stated implicitly or
explicitly. Secondly, an attribution may reflect an internal or external
locus of control. Thirdly, attributions may be stable (consistent across
time) or unstable. Another dimension is one of responsibility, or
whether an attribution is situational or dispositional. Note that the
responsibility dimension is related to both the locus of control and the
stability dimensions. It is included here as a separate category because
one can generate attributions which are internal, stable and situational
as well as ones which are internal, stable and dispositional. Finally,
attributions may be global (pervasive across many types of situations)
or specific to a particular situation (see Harvey and Weary, 1984).
Unstructured attributions
The quest of our work has been to study these attributional processes
and concepts by identifying, coding and analysing naturalistic

attributions. Such attributions are not presented in isolation, but rather
are contextualized and embedded within plots, stories and commonsense explanations (Antaki, 1981). Included in this domain is the study
of accounts, the story-like constructions people frequently make when
trying to provide meaning for confusing, significant circumstances
(Harvey et al., 1986; Weber, Harvey and Stanley, in press; Weiss,
1975), as well as investigations of vivid memories for personal events
(Harvey, Flanary and Morgan, 1986) and confiding after traumatic
experiences (Horowitz, 1976; Pennebaker, 1985).
With some exceptions (for example Elig and Frieze, 1979; Miller,
Smith and Uleman, 1981), research has established that individuals
make unsolicited causal or explanatory statements and that these can
be identified in oral or written free responses. Common-sense
psychology (Heider, 1958) certainly suggests that attributions occur
spontaneously and frequently. In general, unstructured attributional

data may be obtained from (1) archival sources, (2) unstructured
queries of participants in naturally occurring events, or (3) unstructured
probes of subjects in experiments. However, all require consideration
of methodological issues.
Archival sources
Newspapers, diaries, books, written correspondence and taped lectures

or conversations are all potential archival sources for the investigation
of unstructured attributions. In these instances, both the oral or written
response and the event or state which elicited the response are naturally
occurring. In fact, the experimenter may have no contact with the
respondents except through their historical traces.
For example, Lau and Russell (1980) scored 107 newspaper articles
on sporting events for outcome, expectancy of outcome, and causal
ascriptions of outcome. Lau (1984) used this same procedure, but
followed particular teams over time. Unlike the former study, events
sampled were not independent. Schoeneman and Rubanowitz (1983)
also tapped newspapers as archival data sources, identifying
attributions from advice columns. The potential for sampling bias is
perhaps the greatest drawback in using newspapers as a data source.
In addition to newspapers, written correspondence and diaries have

been analysed for attributional content. A novel approach was taken by
Bettman and Weitz (1983), who reasoned that annual corporate reports
would be less subject to response withholding than public statements,
and examined causal reasoning in relation to outcome and expectancy.
Weiss (1975) documented the nature and poignancy of diary accounts
for marital separation.
Oral archival sources are less commonly used. Nisbett, Harvey and
Wilson (1979; cited in Weiner, 1985) randomly selected and
surreptitiously recorded conversations. Sentences with causal content
(expressing or requesting) accounted for 15 per cent of all statements.
Similarly, Antaki and Naji (1987) found a high frequency of the word
because in unobtrusively recorded conversations. They further
categorized the different types of events explained by the because
statements.
Naturally occurring events
Contact with the respondent allows tighter control over sampling

procedures and standardizes response cues. In utilizing this source of
data, events of subjects lives are probed (in an unstructured fashion)
by the experimenter. For example, Harvey, Wells and Alvarez (1978)
elicited and studied verbal accounts and attributions of recently
separated or divorced individuals. In another study, Carroll and Payne

(1977) analysed attributions in the spontaneous verbal reports of parole
officers making parole decisions.
Several issues should be taken into account when using archival

sources or when probing for attributions about naturally occurring
events. Most of these issues arise from the lack of experimental control
over the naturalistic event. Firstly, one must consider the
representativeness of the chosen event or sample when drawing
conclusions. For instance, would attributions of Vietnam War veterans
be generalizable to attributions of Korean War veterans? In other
words, were these events sufficiently similar to generalize? Secondly,
an investigator may be interested in the verity of the account (as
compared with the participants perception or recall) as well as in the
convergence of the account (with or away from another individuals
account). Harvey et al. (1986) argued that the accounts individuals
hold for relationship dissolutions may or may not contain a high degree
of accuracy, and may often express views divergent from those of their
mates.
A third issue to consider is the independence of the event (see also

Brewin and Antaki, 1987). Events in which outcome expectancy is
dependent to some extent on a prior event often occur naturalistically,
and may affect attributional activity. For example, while many repeated
laboratory tasks start anew each time, one divorce may establish
interpersonal behaviours which predispose an individual to a second
divorce.
Finally, the amount of time which has passed since the natural event
should be weighed. Memory has been considered to be a possible
mediator of the relationship between attribution and some dependent
variable (for example Fiske, Kenny and Taylor, 1982; Harvey et al.,
1980). Further, a dispositional shift in causal attributions has been
identified. Subjects explained an action in which they had just engaged
in terms of situational demands, but several weeks later explained the
identical behaviour in terms of their own dispositional characteristics
(Moore et al., 1979). Thus, restricting the event of all subjects to a
particular period or merely recording and analysing the impact of time
passage may be necessary.
Experiments
Most of the aforementioned measurement issues are no longer relevant

when one has experimental control over both the event and the
response. Several investigators of unstructured attributions have

chosen this paradigm. They have asked subjects to talk aloud during
task performance (for example Brunson and Matthews, 1981; Diener
and Dweck, 1978) or to relay their thoughts after completion of a task
(Gilovich, 1983; Harvey et al., 1980; Wong and Weiner, 1981).
Some studies involved relatively direct attribution probes such as

Why do you think you succeeded or failed on this task? (Elig and
Frieze, 1979), whereas other studies attempted to examine attribution
more indirectly. For example, Wong and Weiner (1981) asked subjects
to imagine various outcomes for a questionnaire set of scenarios.
Following these responses, subjects were asked What questions, if
any, would you most likely ask yourself (given the situation
imagined)? In an even more indirect type of probe, Harvey et al.
(1980) asked respondents to indicate in writing what they had seen and
felt in watching short relationship vignettes on videotape. The
attributional activity elicited in this fashion was labelled unsolicited

attribution by the investigators.
Empirical examples: accounts for combat stress
An arena in which the study of attribution using unstructured

techniques is both relevant and necessary is that involving the reports
of combat war veterans. The Vietnam War was traumatic for thousands
of young US troops who for many months were exposed to terror,
death and almost constant peril, but who survived and then returned to
the United States. These veterans report considerable difficulty in
understanding or making sense out of events in their lives since the war
(Brende and Parson, 1985; Figley, 1984). Their accounts are filled with
freely occurring, unsolicited attributions about themselves and their
environment, and thus this population is particularly useful to us for
studying attributional processes and adaptive functioning.
While attribution investigations are relevant to these veterans,

traditional measurement with structured questionnaires is problematic.
During our first contact with a sample of veterans, they told us of their
strong dislike of paper-and-pencil instruments which other
investigators had attempted to use previously. With consideration to
participation rate, we opted to use free response techniques.
The oral account presented below is extracted from a study we are

currently conducting on the effects of attributions, vivid memories and
personality characteristics on the physical, occupational and social
functioning of combat veterans (Harvey, Agostinelli and Claiborn,
1987). Our sample is composed of men who have seen active combat
during the Vietnam War, some of whom have been physically injured.
Subjects are interviewed by two Vietnam veterans who are trained in
interviewing techniques, but who are blind to experimental hypotheses.
The standardized interview format includes global questions regarding
subjects impressions and feelings since their return from the war and
requires approximately 45 minutes for completion. Subjects responses
are recorded on audiotape, transcribed, and coded for attributional
content. As our methods for this particular study do not involve written
accounts, the second illustration is a reprinted example of archival data
which may be analysed in the same manner as our transcribed oral
accounts.
Oral account. The following narrative is a transcribed segment from a

recorded interview with a former Army infantryman:
I still can feel the sheer terror. I still have flashes of the scenes of death
and dying. ... Things have a way of reminding you of a situation you
encountered there ... being unable to relax. ... Sometimes I startle and
want to hit the deck when I hear a car backfire. ... Sometimes I wake
up in panic that I cant get my gun loaded fast enough. ... I have a great
fear of the unknown.
I was comfortable before I left for Nam. I had been married for two
years and had many good friends. But when I returned, it was as if I
had changed greatly and had no control over what I found here. My
wife was waiting for me, and soon after returning our relationship
started to deteriorate. ... I was in a daze. I didnt know what was going
on. ...
I encountered hostility when I went back to college. I had no place
there. I wasnt much older but I was different... and they made me feel
that way. And the students were into various kinds of power - brown
and black and so on. Before I left and in the service, I didnt encounter
such groups; you just had friends regardless of colour. I was dealing
with kids in college who didnt know anything about what I had
experienced or what it was like over there. ... All that I could
communicate with them was second level bathroom bullshit.
I wanted to be off by myself and not around anyone - wife, parents,
former friends, you name it. I felt then and still feel today that I can
only talk about Nam with other vets who were there. ... Some of my
friends were even frightened of me and didnt want anything to do with
me. We were all in the same place but separated in time and by vast
experience. I grew older and they had stayed the same. I not only was
ostracized by them, I also was placed in a different category.... I and
vets like me were and still are somewhat totally separate from the rest
of the world. ...
So for years after I returned, I literally had to re-learn all about the
society - the music, what had happened in this country. ... And I grew
further and further from my wife until we divorced a year after I
returned. ... So it was as if I no longer had a home, a wife, people who
knew and loved me. Now I think that it is easier as time goes by - my
understanding of that period in Nam and soon after is growing and I
feel more control - Im closing the distance between then and now.
Written account. In the following written account, a veteran writes of

his experiences and their continuing effects (Brende and Parson, 1985:
46):
My girlfriend told me she didnt know how to relate to me. ... I had
expected things to be the way they were; but they werent. She said she
thought I had been killed in the war, because I stopped writing to her.
Honestly, I didnt know how to relate to her now either. I dreaded
going to bed with her. ... She also said that I wasnt the loving guy she
used to know and love, that something horrible must have happened to
me over there to change me so completely. I told her I didnt know
what she was talking about. She said the look in my eyes was the look
of a deeply terrorized person, with a long-distance stare. ... She also
mentioned that my frightened look and pallid complexion, my uptight
way of sitting, talking, walking, you name it, my aloofness, and all
that, made her too uncomfortable for us to continue our relationship.
She said that besides, she had found somebody else anyway. That
really hurt me. ...
When it came to my family, my mother told me that I wasnt as
considerate and sweet as I used to be. My dad felt I wasnt as diligent
and committed as he remembered me to be prior to Vietnam. I didnt
know what any of these people were saying. I knew I was getting
pissed off more and more by hearing all of this bullshit.
Techniques for analysis of unstructured

attributions
As is true with all forms of data, conceptual and empirical issues are
not separable. They are there whenever one decides what to count as
data, interprets them, draws theoretical or practical inferences from
them and frames the next research question in terms of a particular
method (see Oyamai, 1985 for an enlightening discussion of this view).
Consistent with the foregoing point, the procedures for analysing and
interpreting free response attributional accounts data should be guided
by theory. None the less, it is likely that the following steps will be
pursued in transforming oral or written free response attribution raw
data protocols - like those above - into quantifiable and statistically
analysable data.
Identifying attributions. In the previously described work by Harvey et

al. (1980) on unsolicited attribution, attributions were identified by
phrases or clauses denoting or connoting causality or responsibility or
ascribing traits. Thus, further reduction of the process would be
necessary to identify causal, responsibility or trait statements or

phrases. For example, consider the veterans recorded statements above
regarding the wars effect on him and his relationships. A coder may
determine the following attributions made by the veteran from the
narrative: (1) a feeling of general lack of control in dealing with the
environment upon return to the US; (2) deterioration in marriage upon
return to the States; (3) a perception of isolation and of being an
outside among peers in college; (4) a feeling of little similarity with or
ability to communicate with peers or family (including wife); (5) a
feeling of identification with and orientation toward Vietnam veterans
upon return to the States; and (6) a feeling that it was necessary to
relearn aspects of the culture and that an understanding of the period in
Vietnam is growing over time.
Defining dimensions. In addition to criteria for recognizing attributions, criteria for identifying theoretically meaningful characteristics of
attributions should be specified. These dimensions may be coded on a

continuous or dichotomous scale. The positive-negative valence of
attributions may be determined by a priori ideas of the experimenter,
by post hoc content analysis and scaling within the sample, or by
reference to empirical guidelines such as Andersons (1968)
likeableness ratings of trait words. The stable-unstable dimension may
be judged either retrospectively (change over time is inferred from
subjects report) or longitudinally (with repeated samples of a subjects
attributions). Similarly, globality of an attribution may be inferred from
the pervasiveness of its appearance in the subjects descriptions or it
may be observed in the subject across different situations. Locus of
control is generally judged by the object of explanation statements, and
focus of responsibility is reflected in the dispositional or situational
nature of the explanation. Finally, although there has been little work
on the structure of attribution (Yarkin and Harvey, 1982), the
organization of attribution should also represent an important
dimension that is amenable to coding.
For instance, the explicit attribution of the first example, I wasnt

much older but I was different ... and they made me feel that way,
evidences (1) negative valence, (2) instability (inferred by contrast
with the last statement regarding his growing understanding), (3)
globality, (4) external locus of control, and (5) situational
responsibility, using dichotomous scales. Of course, with modified
criteria, different conclusions may be drawn.
Coding attributions. A necessary procedure in any study of naturalistic

attributions, this step usually is carried out by two or more practised
coders, who are naive to the experimental hypotheses. Based on the
guiding theory, a count of the total number of attributions as well as the
dimensional qualities of each attribution discussed above will be of
interest. Eliciting conditions of the attributional statements and subject
characteristics might also be included. These data are meaningless,
however, if coders are not reliable in their judgements. One would
expect coefficients of agreement of at least 0.70 to argue for high

reliability of judgements.
A final point is that approaches which are quite different from that
outlined above may be pursued. Firstly, our approach essentially was
designed to quantify all theoretically meaningful dimensions of the
data so that inferential statistical tests might be applied. However, it
might be as tenable to develop a descriptive analysis only.
It also may be viewed as useful and appropriate to stop at the stage of

nominal-scale statistical analysis (for example, chi-square analysis of
number of internal versus external control attributions).
Secondly, whether attributions are recorded or written may have to

be determined by research feasibility issues. Transcription problems
will be reduced when they are written, but respondents may not wish to
write or fill out questionnaires as part of their participation. Writing
may permit more organization and/or precision, but it may also require
a degree of practice using this medium that many respondents do not
possess.

Advantages and disadvantages
As has been suggested throughout this chapter, we believe that

unstructured attributional methods may yield some of the richest
naturalistic evidence that it is possible to collect. Also, they provide the
only means of studying some populations and probing sensitive topics.
Giving people the opportunity to make attributions in their own words
may afford them an occasion for extended reflection and self-report
that other methods constrain. Some reflection and report may be more
natural and less subject to reactivity of instruments than other
approaches requiring more condensed, questionnaire responses. As an
illustration, people can respond to diary-type probes at times when
their feelings about the topic in question are most intense and their
thoughts are most penetrating (for example, awakening heavy with
thought and feeling in the early morning hours).
The disadvantages of unstructured attributional methods certainly

include the possible difficulty of the quantification of data. More
specifically, it is usually a laborious task to train coders to make useful
discriminations and thereby to secure reliable judgements. Also, as
with structured methods, response bias in the form of selfpresentational dynamics may be prominent. Investigators should be
particularly aware of this bias during face-to-face interviews when the
respondent attends to the interviewers subtle social cues.
A further difficulty of an unstructured approach is the potential for

sample bias. Overall, there are likely to be several different selfselection biases found in this type of method as compared with
structured approaches. The effort and work required of respondents
may be such as to reflect a self-selection bias in and of itself.
Respondents who elect to participate may be people who find the task
of talking or writing about their thoughts and feelings in their own
words quite compelling. Archival data are also difficult to sample

without introducing some degree of bias (Weiner, 1985). Therefore, it
is important to test the generalizability of ones findings by comparing
the descriptive characteristics of ones sample with the characteristics
of the population of interest.
These possible limitations notwithstanding, we

conclude by heartily recommending free response
attribution methods to scholars in the field. At
their best, we believe that they can be
systematized so as to permit an instructive
inquiry into peoples continual search for
meaning in critical facets of their lives.Integrative
complexity coding of
verbal behaviour
Philip E. Tetlock and Peter Suedfeld
Theoretical background
Integrative complexity theory was originally developed to explain

individual differences in the complexity of the cognitive rules people
use to analyse incoming information and to make decisions (Harvey,
Hunt and Schroder, 1961; Schroder, Driver and Streufert, 1967). The
theory focused on two cognitive structural variables, differentiation
and integration. Differentiation refers to the number of dimensions of
a problem that are taken into account in evaluating or interpreting
events. For instance, a politician might analyse policy options in an
undifferentiated way by placing options into one or two value-laden
categories, for example the good socialist policies that promote
redistribution of wealth and the bad capitalist policies that preserve
or exacerbate inequality. A highly differentiated approach would
recognize that different policies can have many, often contradictory,
effects that cannot be readily classified on a single evaluative
dimension of judgement - for example, effects on the gross national
product, the governmental deficit, interest rates, inflation,
unemployment, the balance of trade, and a host of other economic and
political variables. Integration refers to the development of complex
connections among differentiated characteristics; differentiation is
thus a prerequisite for integration. The complexity of integration
depends on whether the decision maker perceives the differentiated
characteristics as operating in isolation (low integration), in first-order
or simple interactions (the effects of A on B depend on levels of C:
moderate integration) or in multiple, contingent patterns (high
integration).
146 Philip E. Tetlock and Peter Suedfeld
Early proponents thought of this construct as a trait which they called

conceptual complexity. Measurement of the trait relied heavily on the
semiprojective Paragraph Completion Test for assessingindividual
differences in cognitive functioning. Subjects were presented with
sentence stems (for example, Rules When I am criticized ... ) and
asked to complete each stem and write at least one additional sentence.
Trained coders rated subjects responses on a seven-point scale
designed to measure the integrative complexity of subjects thinking in
the topic area. The higher the score, the greater the evidence of both
conceptual differentiation and, above scores of 3, conceptual
integration.
Integrative complexity coding of verbal behaviour 147
One change in vocabulary must be understood here, although it

actually emerged after about a decade of research in this area. This is
the use of the term integrative complexity. In contrast with conceptual
complexity, which explicitly refers to stable individual dispositions,
integrative complexity is defined by the level of complexity
(differentiation and integration) evidenced in the verbal material or,
conceivably, other behavioural aspects of the individuals or groups
under study. Thus, people who are high in conceptual complexity may
be defined as those who typically behave at a high level of integrative
complexity; however, the latter phrase does not imply anything about
dispositional characteristics. One implication of this view is that people
of different levels of conceptual complexity may actually behave at the
same level of integrative complexity under particular circumstances,
and that an individual of any given level of conceptual complexity may
show different levels of integrative complexity at different times or in
different situations. Integrative complexity can, in short, be viewed as a
measure of a state level of functioning - although this stance does not
preclude identifying interesting patterns of cross-situational or crossissue consistency in functioning (Tetlock, 1984, 1986).
Two points concerning the coding system deserve mention here.

Firstly, with adequate training (two to three weeks), coders can rate
verbal responses for integrative complexity with high levels of
reliability (Pearson product-moment correlations between 0.85 and
0.95) Secondly, the complexity coding system focuses on the structure
and not the content of expressed belief, and is not biased for or against
any particular philosophy. One can be simple or complex in the
advocacy of a wide range of political positions.
Early laboratory research using the Paragraph Completion Test

supported the view that systematic individual differences exist in
conceptual complexity. The test demonstrated predictive power in a
variety of experimental contexts, including inter-nation simulations of
crisis decision making (Driver, 1965; Schroder, Driver and Streufert,
1967; Streufert and Streufert, 1978), studies of bargaining and
negotiation behaviour (Pruitt and Lewis, 1975; Streufert and Streufert,

1978) and studies of attitude change (Crano and Schroder, 1967;
Streufert and Fromkin, 1972). Relative to simple subjects, subjects
classified as conceptually complex utilized a broader range of
information in forming impressions of others and in making decisions,
were more tolerant of dissonant or incongruent information, and were
more likely to be successful in achieving mutually beneficial
compromise agreements in bargaining games.
Such empirical successes notwithstanding, it became clear by the

late 1960s that a static trait model of conceptual complexity was inadequate. Complexity of cognitive functioning at a given time was not
just a function of stable dispositional variables; several experiments
indicated that situational factors also influenced behavioural complexity (Driver, 1965; Schroder, Driver and Streufert, 1967). Some
environments were much more conducive to complex information

processing than were others. The findings demonstrated that (1)
moderate levels of threat, time pressure and information load are most
likely to promote integratively complex styles of thinking; and (2)
individual differences in conceptual complexity influence how people
react to changing levels of these environmental variables. Schroder,
Driver and Streufert (1967) and Streufert and Streufert (1978)
explicitly recognized these points in their interactionist theories of
integrative complexity, and this issue has become a major concern in
the area (see Streufert and Swezey, 1986).
For some time, research on the conceptual complexity construct was

primarily limited to experimental studies that examined the interactive
effects of dispositional complexity (assessed by the Paragraph
Completion Test) and situational variables (environmental stressors) on
subjects selection of low-involvement response options (endorsing

attitudes or making decisions with no important consequences for
subjects own futures or those of others). The external validity
limitations of such studies are well known (see Janis and Mann, 1977;
Tetlock, 1983a).
A critical development in the evolution of integrative complexity

theory and research occurred in the mid 1970s. In a study of
revolutionary leaders, Suedfeld and Rank (1976) used archival
documents as the data base. The study showed that, unlike other
measures of cognitive style that are linked to specific paper-and-pencil
tests (for example, the Dogmatism or Tolerance of Ambiguity Scales,
the Embedded Figures Test), the integrative complexity coding system
is not tied to the coding of Paragraph Completion Test responses
(Suedfeld, 1978a). The data indicated that not integrative complexity
per se, but an appropriate shift in complexity as roles changed, was a

powerful predictor of whether a particular leader was successful in
retaining power after the success of the revolutionary movement.
Since then, a large number of studies have used the integrative

complexity coding system to analyse a broad range of archival
documents and to test an even broader range of hypotheses (for
example Levi and Tetlock, 1980; Porter and Suedfeld, 1981; Suedfeld,
1985a; Suedfeld and Piedrahita, 1984; Suedfeld and Tetlock, 1977;
Suedfeld, Tetlock and Ramirez, 1977; Tetlock, 1979, 1981a, 1981b,
1983a, 1983b, 1984, 1985, 1986; Tetlock, Hannum and Micheletti,
1984; Tetlock, Bernzweig and Gallant, 1985). These novel
methodological applications of the coding system have enormously
expanded the data base of integrative complexity theory. Among other
data sources, the coding system has been used to analyse diplomatic
communications exchanged during major international crises,

transcripts of Japanese cabinet meetings prior to the 1941 decision to
attack the United States, the correspondence and speeches of eminent
non-governmental figures, pre- and post-election speeches of American
presidents in the twentieth century, confidential interviews with
members of the British House of Commons, American and Soviet
foreign policy statements in the period after World War II, and
magazine editorials.
The theoretical network (Cronbach and Meehl, 1955) surrounding

the integrative complexity construct has expanded to include not only
individual difference predictions, but a wide array of hypotheses
concerning situational determinants of complex information processing
that even the later interactionist theories of integrative complexity had
not anticipated (for example, hypotheses concerning the effects of role
demands, societal crisis, life span, accountability, groupthink and value

conflict on integrative complexity of functioning). For example,
Suedfeld (Porter and Suedfeld, 1981; Suedfeld, 1985a; Suedfeld and
Piedrahita, 1984) has found that complexity declines as a persons
death approaches; and Tetlock (1981a, 1984, 1985) has argued that
politicians strategically shift the integrative complexity of their
utterances to achieve a variety of impression management objectives
(for example, rallying electoral support, appearing conciliatory or
tough, pre-emptive self-criticism).
An integrative theory of integrative complexity has yet

to emerge. Given the wide range of apparent determinants of
integrative complexity, it may not be realistic to expect the rapid
emergence of such a theory. A comprehensive explanation of the
findings will probably have to draw on a number of theoretical
traditions, including personality theories (to explain systematic
individual differences in conceptual complexity), role theories (to
explain the impact of institutional and role variables), and cognitive
theories of judgement and choice (to explain the impact of stress and
value conflict).Integrative complexity coding of verbal behaviour 47
Coding procedures
Whatever theoretical position one adopts on the determinants of

integrative complexity, it is critical that investigators agree on the
methodological procedures for assessing the construct. It is not
sufficient simply to provide coders with abstract definitions of
conceptual differentiation and integration and tell them to assess the
integrative complexity of texts. Coders require much more detailed
guidelines to perform their task with acceptable reliability (for such
detailed guidelines, see Bluck et al., 1985; Tetlock and Hannum,
1984).
One critical distinction that coders must make is between

scorable and non-scorable materials. Statements are treated as nonscorable if they contain purely procedural or factual statements with
no causal inferences or evaluative claims. For example:
Mr Reagan and Mr Gorbachev have agreed to meet on 11 and 12
October in Reykjavik, Iceland. The agenda will include arms control,
the mutual balanced force reduction talks in Europe, human rights and
Third World conflicts such as Afghanistan, Angola and Nicaragua.
To be treated as scorable, a statement must go beyond the information

given and either evaluate the event in question (for example, The
unrealistically ambitious agenda ... ) or advance a causal explanation
for the event (for example, in order to disarm domestic critics, Mr
Reagan and Mr Gorbachev ... ). Among other unscorable materials
are passages that consist entirely of quotations (these may be scored to
measure the behavioural complexity of the source of the quotation, but
not that of the quoter), cliches or proverbs, satire and sarcasm, or
definitions of particular terms. A scorer may also reject paragraphs that
appear incoherent or incomprehensible, or where the level of
uncertainty of the scorer is so high as to be unresolvable.
Coders are instructed to assign the lowest possible score (1 on a sevenpoint scale) to conceptually undifferentiated responses - that is,
responses that could have been generated by a single, fixed rule.
Coders can rely on a number of specific indicators or guidelines to
determine whether a score of 1 is warranted. Firstly, does the statement
place events into value-laden (good-bad) categories with a high degree
of certainty? For example, This summit meeting is the latest sign of
the spinelessness of Reagans foreign policy. Reagan acted in a
cowardly and contemptible way in the Daniloff affair; there is every
reason to suppose he will continue to do so on arms control and other
issues vital to the security of the free world. Secondly, does the
statement imply that absolute solutions to policy problems can be
found? For example, We should close down our embassy in Moscow
and give top priority to setting upan impenetrable space shield against
Soviet ICBMs. Thirdly, does the statement deny the existence of value
trade-offs? For example, There is no good technological, political or
economic argument for not pushing full speed ahead with the original
Reagan Strategic Defense Initiative plan. This is indeed a case where
all the pluses are on one side and all the minuses are on the other.
Fourthly, does the statement provide a unicausal account of events?
For example, Soviet agriculture is a disaster because of the destruction
of individual incentives to work in a totalitarian socialist system.
Finally, it should be noted that more than one perspective can be
voiced; the crucial criterion then is that only one is accepted as
legitimate.
Level 2 is intermediate. It is characterized by the recognition of a

potential for legitimate alternatives or exceptions to the fixed rule.
However, the alternatives are not explicitly delineated, and conditions
under which one may be preferable can be implied. A somewhat higher

level of tolerance for ambiguity than at level 1 is demonstrated.
For scores of 3, the coder must decide that the response indicates
both awareness and tolerance of two different interpretations or
perspectives on an issue. Statements that would receive scores of 3
typically have three features. Firstly, they recognize that reasonable
people view the same problem in different ways. For example, The
Soviet build-up of ICBMs in the 1970s looked very threatening to
many American observers. The Soviets, however, may have seen
themselves simply as catching up to a still strategically dominant
United States. Secondly, they distinguish two or more causes for
events but fail to recognize interaction between (among) causes. For
example, Gorbachev hopes to increase Soviet productivity by
cracking down on laziness and corruption on the one hand and offering
incentives for improved labour efficiency and quality control on the

other. Thirdly, they acknowledge that decision making involves
making difficult choices in which no option is better than its
alternatives on all possible criteria. For example, Mr Bush supports
the Presidents request for funding SDI because he deems it critical not
to fall behind the Soviets in anti-ballistic missile technology. He
recognizes, however, that the SDI programme will be extremely
expensive and may trigger a new destabilizing round in the arms race.
The intermediate level of 4 is the threshold for scoring passages that

imply the recognition of integration as well as clearly evidencing
differentiation. At this initial stage, the author must indicate that
several perspectives or dimensions exist and also hint that they
interact. For example, the author may state that further information is
needed before explicit statements about the relationship between
alternatives could be specified, may present alternatives in a way

suggesting that there is tension between them, may indicate the
probability of interactive outcomes, and the like.
To assign a score of 5, the coder must decide not only that the
response indicates awareness of alternative interpretations or
perspectives on an issue, but also that the response clearly indicates the
use of integrative rules for understanding the underlying sources of
these different ways of looking at the world or for understanding the
conditions under which one or another way of looking at the world is
more appropriate. Integration could take the form of mutual influence,
synthesis, and negotiation or compromise. Three typical specific
indications that a score of 5 should be assigned are as follows. Firstly,
there are explicit attempts to explain why reasonable persons view an
issue in different ways. For example, Soviet and American nuclear
forces have offsetting strengths and weaknesses. The Soviets have

advantages in the number of land-based ICBMs and throw weight; the
Americans have advantages in the number and accuracy of submarinebased ICBMs and in accuracy of land-based ICBMs. Arms control
negotiations are extremely difficult because of the tendency of each
side to focus on those aspects of the strategic equation in which it is
inferior and, as a result, to suspect the other side of bad faith and
striving for superiority. Secondly, there is recognition that one needs
to take into account the interactive and not just the separate effects of
the causes of events. For example, The Nixon-Kissinger policy of
detente consisted of both carrots (trade inducements for good Soviet
behaviour) and sticks (a more confrontational American policy posture
in response to bad Soviet behaviour). The Jackson-Banik amendment
took away many of the carrots; Watergate and Americans postVietnam malaise took away many of the sticks. The combination of
these events killed detente. Thirdly, there is recognition that decision
making involves trade-offs in which one must assess the relative
importance of competing values (how much of value r 1 am willing to
give up for this amount of value y). For instance, Developing weapons
systems with first-strike potential may increase the credibility of
Americas commitment to extended deterrence (the nuclear umbrella
over Americas allies). It may also, however, reduce crisis stability - by
increasing the incentives for the Soviets to use their ICBMs before
those missiles are destroyed in an American attack. The key question is
whether our commitment to extended deterrence raises the risk of
nuclear war in a crisis to an intolerable level.
The score of 6 involves signs of a high-level interaction indicating

the working of several levels of concepts. Alternatives are expressed as
plans, processes or courses of action; there is indication of several
moving components viewed as systems or networks. The recognition
of at least two such components, and how they affect each other or the
system, is the criterion for a score of 6. Alternatives are accepted,
compared and integrated so as to describe one or more outcomes,
frequently on the basis of global overviews or organizational
principles.
The score of 7 requires an overarching thesis or conclusion

pertaining not only to the existence but also to the nature of the
relationship or connectedness between alternatives. These alternatives
are clearly described and held in simultaneous focus. Outcomes are
compared; the conditions necessitating a global view, and that view
itself, are described. For example, Although progress is being made

toward ensuring peace between the superpowers, there are
shortcomings within this generally significant achievement. One of the
many reasons for this is the competing demands of peace and security.
We must concern ourselves with the intricacies of resolving this
dilemma. In the Soviet Union, a strong military hierarchy and the
memory of the devastation caused by World War II must be balanced
with the growing desire of the population for an improved standard of
living - perspectives that imply different policies for the management
of the economy and power. In the United States, a wide diversity of
political and economic priorities must be combined with the
accommodation of the views of allied nations whose own widely
varied needs the leader of the alliance must harmonize. The question is
how the leaders of the two countries, working within this intricate
network of the world economic and political allegiances, can
accurately judge each others needs and limitations while at the same
time balancing the competing demands of groups within their own
nations.

Empirical example: cognitive adaptation versus
rhetorical style
The very first study to apply integrative complexity scoring to archival

materials (Suedfeld and Rank, 1976) set out to test a hypothesis related
to changes in complexity as an adaptive response to changes in the
environment. The hypothesis was that revolutionary leaders had to
view the world through a simplifying filter: the movement was all
good, the government all bad; others were allies or enemies; events
that favoured the revolution were to be fostered and commended, those
that did not were to be opposed and denounced.
However, once the revolutionary movement was victorious, and its

leaders themselves became the government, demands changed. In most
cases, a large proportion of the population had been either opposed or
neutral (passive) during the revolution; these people and groups now
had to be treated in such a way as to minimize their opposition or
scepticism. Foreign governments that had not supported the revolution
still had to be dealt with. Problems of rebuilding the domestic
economy and foreign trade, revamping policies in agriculture,
education, security and so on had to be developed and sold to the
population. These goals could not be reached with simple
undifferentiated and unintegrated approaches. Consequently, continued
success would require the leader to move to a more complex
integrative mode.
Analysing the speeches and writings of 19 revolutionary leaders,
from Oliver Cromwell to Fidel Castro, the researchers found that there
was indeed a difference as predicted. Successful leaders (those who
remained in power until natural death or an unforced retirement)
showed a mean change from 1.67 before the victory of the revolution
to 3.65 afterwards; unsuccessful ones (those who were deposed,
executed, imprisoned or otherwise driven from office) changed from
2.37 to 2.22. Individual scores showed that low levels of complexity
before victory were just as important as high levels after it. The authors
hypothesized that integratively complex revolutionaries tend to be
perceived by their colleagues as insufficiently devoted to the cause.
After all, a rebel who professes to see some good in the government
against which he is fighting may have his fervour doubted.
The following examples illustrate the fairly high levels of integrative complexity at which two successful revolutionaries - Cromwell
and Castro - functioned after gaining power. The Cromwell passage

argues for tolerance of different viewpoints and notes the unfortunate
consequences that may flow from intolerance (score value of 3):
Sir, the State, in choosing men to serve it, takes no notice of their opinions; if they be willing faithfully to serve it - that satisfies. I advised
you formerly to bear with men of different minds from yourself: if you
had done it when I advised you to it, I think you would not have had so
many stumblingblocks in your way. It may be you judge otherwise; but
I tell you my mind. I desire you would receive this man into your
favour and good opinion. I believe, if he follow my counsel, he will
deserve no other but respect from you. Take heed of being sharp, or too
easily sharpened by others, against those to whom you can object little
but that they square not with you in every opinion concerning matters
of religion.
The Castro passage also reveals cognitive differentiation, although in a
different way. In praising Che Guevara, Castro distinguishes among a

variety of Guevaras achievements and character traits (his experience
as a soldier and the fruits of his intelligence; his skill as a writer and
the profundity of his thought):
And because of this he has left future generations not only his
experience, his knowledge as an outstanding soldier but also, at the
same time, the fruits of his intelligence. He wrote with the virtuosity
of a master of our language. His narratives of war are incomparable.
The depth of his thinking is impressive. He never wrote about
anything with less than extraordinary profundity; and we have no
doubt that some of his writings will pass on to posterity as classic
documents of revolutionary thought. And thus, as fruits of that
vigorous and profound intelligence, he left us an infinity of memories,
an infinity of narratives that, without his work, without his efforts,
might have been lost forever.
These data do not indicate, however, whether the change in successful

leaders was deliberately managed by them (that is, whether the leaders
themselves were aware and in control of the change). The position
taken by Suedfeld and Rank (1976) was that structure is not likely to
be manipulated because it is less salient than content. Further studies
showed changes in the opposite direction (from high to low levels)
with societal stress, particularly impending war, among both
governmental leaders and other eminent individuals such as scientists
and writers (Porter and Suedfeld, 1981; Suedfeld, 1981, 1985a;
Suedfeld and Piedrahita, 1984). The disruptive stress hypothesis
advanced by Suedfeld and his colleagues (for example Suedfeld,
1978b, 1985b) explained these shifts on the basis that environmental
stressors such as information overload, time pressure, danger to
important values and the like, which frequently also lead to individual
stresses such as fatigue and anxiety, reduce the complexity of
information processing.
An alternative view was proposed by Tetlock and his group, based

on findings that in some circumstances the change in integrative
complexity was so rapid as to imply deliberate impression
management or at least changes in rhetorical style, rather than
adjustments in the actual level of information processing. For example,
eventually victorious US presidential candidates (somewhat analogous
to pre-victory revolutionaries, in the sense that they were in
opposition) showed lower levels of integrative complexity in policy
statements during an election than they did immediately after assuming
office (Tetlock, 1981b). Tetlock argued that simple rhetoric helps to
rally support for attacks on existing policies, whereas complex
reasoning is often most useful in defending those policies.
Another study, using as the data base the speeches of US senators

during periods when control of Congress shifted between liberals and
conservatives, provided mixed support for the rhetorical style

argument (Tetlock, Hannum and Micheletti, 1984). The prediction was
that each side would show higher complexity when it was in control
than when it was in the minority. The results showed reductions in
complexity among liberal and moderate senators when the control of
the Senate moved to the conservatives; but there was no corresponding
increase among conservatives during these periods.
Thus, the question is not yet answered. On the one hand, stressrelated decreases in complexity have been found among individuals
who have no particular reason to adjust their rhetoric, since they have
no direct role in either attacking or defending a particular policy; on
the other, the rapidity and strategic nature of the change cast some
doubt on explanations involving major shifts in cognitive style. Of
course, there are no data showing the time requirements for such shifts
to occur. It may even be that the question is not answerable, since

cognitive and rhetorical style may be so confounded as to defy
separation (cf. Tetlock and Manstead, 1985); but some suggestive
studies can at least be imagined. One might, for example, analyse
speeches delivered by the same person, on the same topic, to hostile
and friendly audiences in rapid succession on several occasions. The
two hypotheses lead to contrary predictions: disruptive stress would
cause lower complexity when one is addressing ones opponents
(assuming that the issue is important to the speaker, the audience is
both important and strongly hostile, and so on); the rhetorical style
hypothesis would posit higher complexity when defending ones policy
before its critics. The same situation could be simulated in an
experimental context: individuals could be asked to argue for particular
policy positions to audiences with supportive, hostile or ambiguous
views. Again, the disruptive stress hypothesis would predict simpler,
and the rhetorical style hypothesis more complex, arguments when
individuals address hostile audiences. Other aspects of persuasive
rhetoric could also be explicitly measured (for example, one- versus
two-sided, extreme or moderate, firm or conciliatory arguments).
Experimental studies would also permit more refined tests of
individual difference and situational boundary conditions for the
applicability of the competing hypotheses. For instance, modifications
of rhetorical style may be more often a valid explanation for
complexity changes among persons high in self-monitoring (Snyder,
1979); the degree to which subjects feel personally committed to the
policies they are asked to defend may partly determine whether they
rely on simple versus complex styles of rhetoric to defend those
policies (cf. Staw, 1980).

The advantages of the integrative complexity approach reflect some of

the reasons for its development. Complexity is related to information
processing and problem solving in almost any area of life. While
conceptual complexity may indeed exist as a stable (or relatively
stable) personality trait, the construct is not necessary for examining
the level of complexity at which people function. The great range of
subjects studied with this approach, which includes historical figures,
politicians, generals, distinguished scientists, novelists, business
executives and diplomats as well as more standard research
populations, is only one index of its versatility and applicability.
One example is the accessibility for research of individuals who

cannot be reached by more intrusive methodologies. Thus, for
example, studies based on the archival scoring of integrative complexity have included as subjects such people as Queen Elizabeth I,
Oliver Cromwell and George Washington. The ability to assess the
integrative complexity of long-dead historical personages is only one
issue; another is the fact that contemporary leaders who are not likely
to accede to requests for research participation can also be reached.
Tetlocks study of US presidents, and Suedfelds research on such
individuals as Fidel Castro and Andrei Gromyko, exemplify this point.
Several checks on the validity of scoring translations from the original
language (by comparing the scores from translations and from the
original) have affirmed that the level of complexity is not significantly
affected, at least in the case of official, authorized or scholarly
translations.
The archival measurement of integrative complexity uses random

sampling of the available materials. Information that might serve to
identify specific individuals, periods and other variables crucial to a
particular hypothesis can to a great extent be eliminated or obscured.
These two features make the technique less susceptible to experimenter
bias than most content analyses. As a contrast, we may consider
psychohistorical techniques. Psychohistorians typically look for crucial
events in the subjects childhood and family life, which when
associated with the persons adult behaviour are tested against certain
psychodynamic hypotheses about such relationships. One of the most
frequent criticisms of psychohistory is that some incidents in childhood
are emphasized while others are ignored, and that the researchers are
likely to pay more attention to those incidents that fit psychodynamic
preconceptions. As a result, the total early experience of the subject
may be seriously distorted. This problem is exacerbated by the fact that
very few if any biographies or autobiographies provide equal detail
about the various periods or experiences of the subject, tending to
mention dramatic incidents and to ignore many others. Whether the
resultant characterization of the persons childhood is valid has been
severely questioned (for example Greenstein, 1975; Tetlock, Crosby
and Crosby, 1981). By contrast, in the integrative complexity scoring
procedure a person who does not know the hypothesis locates the
corpus from which samples are to be drawn, and then by the use of a
random numbers procedure selects the sample to be scored. Once this
is done, such information as specific names and dates is removed to the
greatest possible extent before the samples are actually scored by
another person. Ideally, the scorers are also unfamiliar with the specific
hypothesis being tested.
The integrative complexity approach does not assume that cognitive

behaviour is rigid. In fact, it can monitor changes in the level of
complexity characterizing a subjects behaviour quite finely. For
example, written materials may be produced throughout several days
or weeks of a particular decision process or crisis situation; in many
cases, these are precisely dated and researchers can trace changes day
by day. When the event involved is also well documented, as in
historical and some autobiographical records, one can further establish
the relationships between such changes in integrative complexity and
the developing situation in which the source is functioning.
It should also be remembered that the theory and the research

methodology are neutral as to content. Thus, the same position on any
particular issue may be held at any level of complexity; conversely, the
same level of complexity may underlie diametrically opposed views.
This relative freedom from content bias makes integrative complexity

measures more objective and wider in applicability; but perhaps its
most important advantage is that it is less subject to either deliberate or
unconscious manipulation. Suedfeld, Tetlock and Ramirez (1977)
demonstrated this in the case of speeches about the Middle East
problem in the UN General Assembly. During periods of growing
crisis, several months before major wars broke out in the area, the
speeches of the delegates were as full as ever of references to the
importance of peace and their own countrys dedication to maintaining
it; in fact, one content analysis of these speeches had indicated that
such professions appeared to intensify in the several months prior to
each
outbreak of war. Overt content is easily selected by the source to

convey the message desired; but levels of integrative complexity,
which in the speeches showed consistent and dramatic decreases

during the periods leading up to war, implied a hardening of positions,
a reduced willingness to consider more than one point of view as
legitimate, and the other characteristics we associate with integrative
simplicity. In the aura of international crisis, these same characteristics
seem likely to be associated with the probability of a breakdown in
peaceful negotiations.
Unlike some content analytic techniques, integrative complexity

scoring can be reliably performed on a relatively small data base,
which makes it easier to perform replications and reliability checks.
When compared with such massive analyses as, for example, those of
Holsti on governmental documents just prior to World War I, the effort
and time required to score a sample of paragraphs for complexity are
relatively small. It would be very difficult to perform complete content
analyses on, for example, the documents produced before and after the
success of several revolutions between the seventeenth and twentieth
centuries, as was done in the study by Suedfeld and Rank (1976). Nor,
if one wants to eliminate the possibility that a particular score or
analyst is inaccurate, is it very difficult to select another random
sample from the same data base and have it scored by one or more
other trained raters.
Given the nature of sample selection, one also need not worry
unduly if the archives are not entirely complete. Whereas in content
analysis one or a few documents may be crucial, and their
unavailability may distort the data and conclusions, this is not likely to
happen in the case of integrative complexity scoring. When gaps exist,
the remaining material stands on its own; when the gap may be only
part of the document (although in terms of content a crucial part), the
rest of the passages may be sufficient to establish complexity level.
We now turn to the disadvantages of the procedure. One drawback

is that the process of becoming a trained scorer is still difficult. It
requires participation in a scoring workshop or seminar, most of which
last at least one week and frequently more. Furthermore, such seminars
are at the moment available in only two major places, the University of
California at Berkeley and the University of British Columbia. It
should be mentioned that there is work going on at present to develop a
training system and scoring manual through which individuals can
learn the procedure at a distance, and then submit samples of scored
test materials to establish their reliability as raters.
It is generally impossible to establish whether changes in

complexity are causes or effects of environmental or situational
developments (or even whether that dichotomy is applicable). For
example, there is a replicated finding of decreasing integrative
complexity in the public position statements of governments prior to
the outbreak of war (Suedfeld and Tetlock, 1977; Suedfeld, Tetlock and
Ramirez, 1977; Tetlock, 1985). It is probable that the worsening of a
crisis and the reduction of complexity are cyclically interactive: as the
crisis worsens and stress increases, it disrupts the complexity of
integrative functioning, which in turn leads to statements and decisions
that further exacerbate the situation. But while this hypothesis seems
intuitively plausible, there is no way to test whether it is accurate. One
may equally well argue that the decrease in complexity is merely a
consequence and symptom of the deteriorating situation; or conversely,
that the situation gets worse because the protagonists are taking
increasingly simple positions.
Tetlocks argument that the level of integrative complexity may be

manipulated as an impression management strategy may further
complicate the relationship between complexity and situational factors.
Although this hypothesis still requires further research, perhaps in a
more direct methodology (for example, one that allows the researcher
to find out whether subjects are intuitively aware of the level of
complexity at which they operate), it is certainly one that needs to be
kept in mind as a warning in research relating public events to
cognitive processes.
Another problem is that one does not always know the actual author
of particular materials. This is especially true with documents
produced by anonymous writers, even when there are strong historical
guesses as to the real source; and for material possibly or definitely
produced by speech writers and other assistants. Some studies have
shown no reliable differences between the latter kinds of material and

those known to have been produced personally by the subject being
studied. It may be argued that any major leader will insist that he or
she feel personally comfortable with a speech produced by someone
else before endorsing it. Speech writers and ghost writers are probably
selected with that criterion in mind, and their submissions have been
known in many cases to be rejected or edited accordingly. Still, the
potential problems raised by the use of such assistance cannot be
definitely dismissed in every instance.
In some cases, written material has not survived to make complexity scoring possible. Recall, for example, Suedfeld and Ranks
(1976) data showing that revolutionary leaders who evidenced low
levels of complexity during the revolution and then attained significantly higher levels when they were in governmental office tended to
remain powerful longer than those who failed to make such a shift. It
would be interesting to know whether the former group had showed
high or low levels of complexity prior to being involved in a
revolution; however, for most of them no correspondence or other
writings are available from the pre-revolutionary period. Similar cases
arise with illiterate leaders, non-literate societies or subgroups, and
individuals whose materials were produced in a relatively little-known
language and have not been translated.
Where do we go from here?
The integrative complexity coding system is a cost-effective and wellvalidated methodological tool for testing psychological hypotheses in a
remarkable range of contexts: domestic and international politics,
literature, science and business. The price of this empirical
productivity has, however, been a degree of conceptual confusion.
Data generation has outpaced theory generation - a state of affairs

reminiscent of the cognitive dissonance literature in the late 1960s. We
now have numerous interesting demonstrations of individual difference
and situational correlates of integrative complexity of functioning. We
know, for example, that the integrative complexity of verbal behaviour
is influenced by the personality of the perceiver (demonstrations of
significant consistency across times and issues), the issue domain
under consideration (demonstrations of the impact of value conflict),
ones political role (demonstrations of the impact of changes in role or
political audience), and environmental stressors (demonstrations of the
impact of information load and threat). Determinants of integrative
complexity can be viewed from a correspondingly wide variety of
theoretical perspectives: psychodynamic theories of political belief
systems (for example Adorno et al., 1950; McClosky, 1967; Stone,
1980), information processing theories of cognitive functioning under
stress (for example Schroder, Driver and Streufert, 1967) or under
value conflict (for example Janis and Mann, 1977), and theories of
impression management and self-presentation (for example Tetlock
and Manstead, 1985).
We know little about the conditions under which

these very different theoretical points of view are
more or less useful. It has been easier to build up
a differentiated body of empirical knowledge
than it has been to come up with integrative
theoretical principles. The major challenge now,
however, is theoretical integration. It may not be
possible to resolve the theoretical dichotomies
that social psychologists often try to impose on
their data (for example, impression management
versus intrapsychic, cognition versus motivation);
but it is possible to organize the observed
covariates of integrative complexity into
empirically defensible and theoretically
meaningful categories. Much remains to be
learned about the stability and consistency of
individual differences in integrative complexity.
Why is the complexity of some individuals much
more stable across situations than others? Much
remains to be learned about person-situation
interactions in integrative complexity. Why, for
example, do marked individual differences exist
in the cognitive strategies that people use to cope
with accountability demands and value conflict?
We doubt that archival research by itself will be
able to answer such questions. Laboratory studies
will be necessary to isolate key variables and to
test critical hypotheses. But we are optimistic that
a research programme that builds on the
distinctive strengths of both the archival and the
experimental traditions will advance our
understanding of the antecedents, mediators and

consequences of integrative complexity.Structures
of belief and justification
Charles Antaki
In this chapter I want to describe one way to get at how people

organize their explanations - how they set them out and how they
defend them. The technique I shall be reporting is a compromise
between getting at someones completely natural and spontaneous

explanation (if such a thing could exist) and the demands of
standardization and convenience. It is also a compromise between on
the one hand getting a complete catalogue of everything a person
could say in an explanation, and on the other being satisfied with the
reductionist programme favoured by standard social cognitive
techniques. It is best used for looking at the explanations people give
for events they care about and know about; its a technique that
encourages a certain rationality, and a certain commitment to a clear
and definite view of the event under discussion.
5
The point I want to champion is that peoples explanations are

coherent wholes, more or less well structured in and of themselves.
Your explanation of (say) your choice of holiday destination or your
position on nuclear weapons will have ramifications, elaborations,
justifications; the causal links in the explanation will be more or less
secure, and the whole thing will be more or less well moored in the
network of your beliefs about life in general.
192 Charles Antaki
This notion of an articulated causal belief structure may seem

reasonable enough, but it is rather a change from the tradition in
experimental social psychology. Attribution theory - the umbrella term
for experiments on ordinary explanation - is very far from being
interested in the subparts of a persons explanation structure or how
they knit together. On the contrary, the strength of the attributionist
position is the way it reduces the complexity ofordinary explanation to
a small set of clear conceptual categories. The attributionist is less
interested in the building blocks of your explanation of (say)
unemployment than in how the whole collection could be boiled down
to two or three dimensions of meaning.
Structures of belief and justification 193
Reducing explanations to simple causal categories Those who

followed Heider (1958) had the thrifty idea of reducing the complexity
of ordinary explanation to a manageable number of categories of cause.
There is far more to Heider than this reductionism, but that has been
his legacy for all practical purposes. What Heider did was not to say
that causes were necessary or sufficient, enabling or inhibitory,
counterfactual or abnormal; his intention was not to philosophize
ordinary explanation, but rather to find philosophical distinctions in
ordinary talk. His prime gift to attribution theorists was to point to a
profound difference between causes that the individual perceived to be
located within agents of action and those perceived to be located
outside them. Broadly speaking, all ordinary explanations of social
behaviour - and non-social behaviour - could be ranged somewhere
along the line from internal to external.
The idea of distilling an explanation, and the idea of distilling it
194 Charles Antaki
down to a point on an internal-external line, found (for reasons we need

not go into here) a responsive chord in the mood of developing
cognitive social psychology. Apart from anything else, it gave a
blessedly simple dependent measure which could be used to chart the
output of a complex social judgement. By the late 1970s the literature
was alive with hundreds of experiments designed to discover just what
made an explainer choose one point rather than another on the internalexternal line. The guiding models of Jones and Davis (1965) and of
Kelley (1973) suggested that rational information was important; and
indeed, so it was. To caricature, but not unfairly: if a respondent was
told that John consistently laughed at a certain comedian, and at many
other comedians, that information disposed the respondent to judge the
cause of that event to be internal to John.
The value of this reductive programme depended on two things:

firstly, that the internal-external division to which explanation had been
boiled down was a sensible and productive one; and secondly, that not
too much of the flavour of ordinary explanation was destroyed in the
process.
It would be foolish not to concede that the internal-external

distinction is a profound one. As Totman (1982) points out, it is
enshrined in our legal and moral worlds; it would be astonishing if it
had no psychological counterpart. As to sense, then, the programme
passes. Productivity? Once again, it seems obvious enough that, in
some circumstances, the single most important thing about an
explanation is indeed what it all boils down to. Say two students fail a
class test. Both give long and elaborate accounts of the whys and
wherefores of the failure. We ask them to boil it down to one thing.
Jane says My alarm clock didnt wake me up in time; John says Im
just no good at maths. From that stark difference we know that Jane is
likely to do better next time, since clocks can be changed; John is not,
196 Charles Antaki
since presumably ability cannot. Here the single-category explanation

does us well enough; there is, in fact, a large literature on how such
explanations affect work and happiness in class (see Weiner, 1986) and
are implicated in depression (see for example Peterson and Seligman,
1984).
But what have attributionists lost by the reduction? Firstly, they

have lost sight of how people reason in making their explanations. No
one needs to reason or show their working in an attribution
experiment. Indeed no one is allowed to. All the working is assumed to
be done by the hypothetical constructs (the analysis of covariance and
so on) that are the business of Jones and Daviss and Kelleys models.
The reasoner has been displaced by her or his cognitive system. What
is lost is respondents evidence, backing, the justification of their
causal beliefs. Attribution theory can be a theory about cognitive
processing, but it can never be a theory of ordinary reasoning.
Secondly, reduction means stripping away the context of

explanations - to whom they are given, by whom and for what purpose.
Reduction to a cognitive product denies the possibility of explanations
as social processes with a point and a purpose. Respondents have been
freed of their need to persuade, and they have, in any case, been denied
a language in which that persuasion could be done. Attributions on
rating scales are offered to nobody, they are in a reductive language
foreign to the respondent, and they force the respondent to use a
vocabulary of thought which, for all its occasional predictive value, is
on the whole a travesty of the richness of ordinary expression. Put the
explainer in a social context where she or he wants to explain
something important to someone, and you will not see a bald and
vulnerable attribution to internal or external cause. You will see
argument, defence, justification, and a display of a complicated and
articulate explanation.
The structure of beliefs and warrants
198 Charles Antaki
What, then, should we be doing if we want to get at peoples reasoning,

and at the social goals of their explanations? Clearly the first step is
just to listen to what they say, and record it as faithfully as possible. We
could stop there and do content analyses of what concepts they use.
That would be a perfectly reasonable start, but I think one can do more.
We could sketch ourselves an idealized picture of what an

articulated explanation could look like. The criteria would be that it
showed how the person reasoned, and that it had a place for the social
purposes that the person had in mind. Then, armed with this sketch, we
could do some empirical data collection and try out the data against the
ideal model. Let us take a look at what the model should feature.
Structure. When one says that individuals have a certain structure to

their beliefs, one is saying that there is some network of (at its loosest)
association or (at its tightest) logical implication among a certain set of
their views of the world. This kind of idea is most familiar in
psychologists personality theories, where the paradigm case of (say)
an authoritarian personality (in the terms of Adorno et al., 1950) was
thought to believe in X and Y and Z. But note that in this sense
structure here is really no more than a constellation; and the pattern,
like the Great Bear or the Plough, is in the eye of the expert beholder.
The kind of structure I have in mind is the structure that the persons
themselves see. At its most basic, it will be something like the toe bone
is connected to the (pause) foot bone, the foot bone is connected to the
(pause) shin bone and so on. It doesnt have to be anatomically
accurate to be what the person genuinely believes.
Justification. Once one allows that respondents can make something of
200 Charles Antaki
organizing their own beliefs, one asks what they might want to use to
cement crude belief statements together (why is the toe bone connected
to the foot bone?). In some discourses, the justification for claiming
that A caused B will be a mechanical one; in another discourse, it will
be a motivational one; and so on. But whatever our respondents say,
they will want - if they want to abide by canons of rationality - to say
something about why they think the two beliefs are linked.
We now have a very rough sketch of what might be in an ordinary

explanation. Let us now have a look at some data to fill in some of the
detail.

Empirical examples
I shall give two illustrations of what one can do with the idea that a
persons explanation has a reasonably elaborate causal skeleton which
is fleshed out with justification. One example is from a study in which
respondents talked, and I constructed their explanatory structure from
the transcripts of taped speech; the other is from a study where the
respondents wrote down the structure themselves.
Example 1: conversational structures
202 Charles Antaki
In the first case, I talked to members of the Social Democratic Party

(SDP) about their decision to join the party, their explanations of
current social problems, and the state of the political scene in Britain.
All were academics who were articulate, thoughtful and - by virtue of
their membership of a consciously new political grouping - keen to
account at length about such political matters. Here is a longish extract,
in which JG explains British unemployment.
CA: What from your point of view are the causes of unemployment?
JG: Im sure the world recession has had a lot to do with it, and I think
on top of that the very tight monetary policies of the government
have exaggerated this to some extent. Whether one can believe
half the effect being caused by the world recession Im not sure. I
would like to know much more about economics, but when I think
that the professional economists themselves cant find a uniform
approach I wonder if I ever will.
CA: Lets take the world recession first. Obviously one is speculating
here, but how would you unpack the causes of the world
recession?
JG: I dont honestly know. What happens in this country largely stems
from what happens in the States. Were very dependent - just as we
are for the weather, incidentally. I think there have been similar
monetarist policies, which Ronald Reagan possibly copied from
Mrs Thatcher. Im not sure, but they were fairly adjacent in timing,
so thats perhaps unfair. But fairly restrictive policies causing high
interest rates and so on.
CA: What would the Reagan intention be, in having those restrictive
practices?
JG: I think he wanted to some extent to undo the social fabric of the
state ... by that I mean social medicine, these sorts of things; [that]
is my impression. To save money - you know, on the good
housekeeping type of theory of Mrs Thatcher. Thats my
impression of it.
CA: Lets move to Mrs Thatcher and the government policy. You were
saying that government policy has had some effect on unemployment. What is the government policy in your perception?
JG: I think principally its designed round the good housekeeping idea
of only having so much money to spend, and Ive never really
followed this through to the bitter end, so I cant claim to
understand it. Because people talk about the ability to borrow
without making the country any worse off and certainly, from what
little I know of business, businesses borrow money in order to
expand and produce more and thereby make more money, and I
cant help feeling that the country must have a similar sort of

ability to do this. Where you draw the line, of course, I dont feel
that I know enough to say, except that I have this gut feeling that
Mrs Thatcher has drawn the line too low.
CA: Do you think shes drawn it even from her own point of view on
rational lines?
JG: I think shes following Geoffrey Howe in this. I suspect that hes
the man who has really drawn the lines, and she seems to have
great trust in him - possibly alone of her cabinet - and I dont
altogether trust Sir Geoffrey Howes logic. Again Im not an
economist enough to know whether this is wrong or not. I suppose
I ought to say that what opinions I have on economics are based
on reading The Guardian more than anything else. That gives
away a lot, doesnt it?
CA: One can read The Guardian with very different interpretations.
The way Ive set it down here it looks as if ones attributing
government policy, if pushed, to Howes analysis?
JG: Yes, I think that wouldnt be too unfair. I have a personal dislike of
Mrs Thatcher for what I consider her dictatorial, overbearing
manner, but thats something separate in a way.
204 Charles Antaki
Here we are, then, with a bare transcript. If we were attributionists

looking for attributional categories, we would be concerned to boil
down the words to a point where we could see clearly where JG fell on
our causal dimensions. This would be, as I noted above, the sensible
thing to do in some circumstances; and, if we felt the circumstances
were right, we would turn to the sophisticated and sensitive work
exemplified by Harvey, Turnquist and Agostinelli in chapter 3 to guide
us.
However, rather than pinning JG down to the attributionists

language, let us instead take him at his own estimate. How does he see
the structure of the causes of unemployment? What is the skeleton, at
least, of his reasoning?
Causal structure. Stripping away everything else, JGs causal story is

that unemployment is caused by two kinds of thing: world recession
and government policies. The world recession is caused (largely) by
events in the US; specifically, restrictive monetarist policies. These in
turn are caused by President Reagans intention to reduce public
spending. On the other hand (or down the other path), government
policy (good housekeeping) is determined by the Prime Minister, who
in turn is being guided by Geoffrey Howe (the Chancellor of the
Exchequer at the time).
If we set all that out diagramatically, it would look like Figure 5.1.
You can see that the reduction to the skeleton of causal structure does
some violence to the subtlety of JGs expression, and others would
treat it quite differently: see especially the contributions by Heritage
(chapter 9), Wetherell and Potter (chapter 12), Parker
206 Charles Antaki
British unemployment
British
government
policies
n
World
recession
US
monetarist
policies
Prime Minister
Chancellor
of
the Exchequer
President's intention
to save money
Figure 5.1 A diagram of the causal structure in a respondents
explanation of British unemployment (the causes read upwards)
(chapter 13) and Billig (chapter 14). But let us keep to it for now and
come back to subtleties in a moment.
Suppose we then went through the same kind of skeleton stripping

in a large sample of similar transcripts of peoples explanations of
unemployment. We would have in front of us a large set of diagrams of
causal structures. We could then ask some interesting questions. Are
some peoples structures more articulated than other peoples? Do
people use lots of causes in a single, unwinding string, or do they see
(as JG does) two or more independent paths of causes leading up to the
event? Do some types of cause occur more reliably closer to the event
than others? Do some types of cause predict other types of cause? If
there is a tendency to attribute unemployment externally, does it
appear in causes near the top of the paths, or in fundamental causes at
the bottom of paths?
208 Charles Antaki
I dont want to foreclose on all the possible questions, but I can

answer one or two of them on the basis of the study from which JGs
transcript actually comes (Antaki, 1985). The academic SDP
respondents I talked to were more likely to put things to do with
persons or society right at the bottom of their causal pathways. Groups
(like the Conservative Party or left-wing radicals or traditional
economists) would appear anywhere; person causes (Mrs Thatcher)
and society causes (world trade, free collective bargaining), on the
other hand, appeared proportionately more often as termini. There was
a significant relationship among the types of causes in any path; one
could predict, with greater than
chance accuracy, what cause was coming next. The pattern

was that a personal cause would be unpacked into a further
personal cause, a societal cause into another societal cause,
and a group cause into a group cause. There were few
examples of society to personal steps, or vice versa.
One could spend time interpreting any of these structural findings,

generating ideas to test with new samples of structures; but let us leave
it there.
210 Charles Antaki
Justifications. So much for demonstrating the idea of looking at the

skeletal structure in a transcript. The next thing to do is to look for the
way the explanation proceeds at an interpersonal level to get across
whatever social purposes the interaction requires. What is JG up to, and
how does it emerge in his words?
As you can see from the transcript, JG is doing much more than
neutrally reporting the bare skeleton of the diagram. Other chapters in
this volume - see especially Parker (chapter 13), Heritage (chapter 9)
and Leudar and Antaki (chapter 10) - make much more of the
obviously interactive nature of what is going on, and some things look
obvious on a first pass. JG occasionally depreciates his competence to
answer the question (Ive never really followed this through to the
bitter end, so I cant claim to understand it; I would like to know
much more about economics, but... I wonder if I ever will; I dont
honestly know). He is nevertheless cooperative and articulate in
answering the probes and he is at pains to distance himself from merely

emotional reactions (I have a personal dislike for Mrs Thatcher ... but
thats something separate in a way).
Such aspects of the data paint a more vivid picture of JGs image of
the world. We can capture some of that vividness, I think, by making
space in our diagram for the justificatory links, or warrants, that he
makes between his causal claims. The idea of warrants is described by
Draper in chapter 2, and more thoroughly laid out by Voss in chapter 6.
The idea we need to note here is that we are taking it that JG is giving
reasons or justifications for the causal claims that he is making.
212 Charles Antaki
Take the claim that world recession is caused by US economic

policies (What happens in this country largely stems from what
happens in the States). The warrant he gives is a meteorological
analogy: Were very dependent - just as we are for the weather,
incidentally. The use of an analogy is neat, but JG is clearly not fully
confident that it will do; he adds incidentally at the end of the
sentence to pull away from a full commitment. The tone continues in
his warrant for claiming that Reagan possibly copies from Mrs
Thatcher: Im not sure, but they were fairly adjacent in timing. Here
there is a stronger, more rational piece of warranting: adjacency in
time, on a Humean model of causality, is consistent with A causing B.
On the other hand, JG immediately stands back: Thats perhaps
unfair.
It is clear that JGs warrants of his causal beliefs are at the mercy, in
this interaction at least, of worries about his competence to give them.
Nevertheless, he does back up what he says, so we can begin to read

off a little bit about what kind of evidence he thinks appropriate to
various claims, what authorities he might be citing, and what kind of
reasoning he thinks appropriate to display to this particular interlocutor.
(Of course, at home or in the pub he might use quite a different
discourse.) As we said above when looking at JGs structure, one could
collect data on all of these things from more people like JG (in this
case, middle-class members of the SDP) to get a feel for the
justifications they consensually use. Carrying on further with this
example of a project, one might then compare them with Conservative
or Labour voters, or with working-class SDP members, and discover
interesting things about the kinds of evidence different groups invoke
to back up their beliefs about causes.
The drawback to the conversational method is that it requires a fair

degree of subjective interpretation both in identifying the respondents
214 Charles Antaki
causal structure, and then in spotting just what it is that is meant to

count as evidence or warrant for the various causal links. It goes
without saying that this kind of subjectivity can go horribly wrong; in
interpreting JGs talk, I might simply have misunderstood what he
meant, made one or two spurious links, and got the whole structure of
causes and justifications quite wrong.
The alternative is to get the respondents themselves to make the

causal links - on a piece of paper, if necessary - and to say clearly and
specifically what warrant they have for making each link in the
structure.
Example 2: written structures
How shall we get our respondents to write down their explanations?

One could ask them to write an essay. But we would still have the
problem of seeing through to the skeleton underneath the
idiosyncrasies of expression. And it would hardly help at all in making
the justifications any clearer: people do not typically write and I think
thats the case because ... . No; if we have decided that we are
interested in structure and justification (and we have first checked that
such things really are there in peoples explanation), then we shall have
to ask for them explicitly.
Let us look at a procedure that turns the conversation into an explicit

pencil-and-paper task. I asked students at Manchester University to
explain to me the causes of the students occupation of university
buildings (the occupation and its aftermath was at the time a live issue,
attracting a good deal of media attention and causing strong feeling
around the university).
216 Charles Antaki
Once the students had been recruited, they were left to complete the
task in their own time and at their own pace. Each person was given an
envelope containing the instructions. First you had to write down, on
one card per cause, as many causes as you wanted to explain the
occupation. Then you had to take those cards and arrange them so that
they showed which causes caused which other causes, and how they all
related to the causing of the occupation itself. This meant shuffling the
cards about on a table or something of the sort. Then the respondent
drew the pattern he or she had made on to a blank sheet of paper, with
an arrow between every cause-effect pair (notice that the instructions
did not mention paths, steps, branching or anything else; the
respondent was entirely free to write as many cards as she or he
wanted, to put them in any causal order, to branch them in any way,
and so on). All this, of course, was merely breaking down the task of
drawing out a causal structure in the respondents own words.
To get at his or her justifications for causal beliefs, the next thing the
student had to do was to write down, on a separate card, the
justification for each link in the causal structure. The instruction for
this was simply to say: Why you think there was that causal
connection.
What the respondent gave me, then, was a sheet of paper on which
was drawn his or her own causal structure, some cards describing the
causes and some cards describing their justifications. I could then
reconstitute the diagram and the cards into a full-blown causal
structure. The general form of the structure was a chronological narrative. The start of the event was located at the visit of an allegedly
racist Home Office minister, which was known as the Waddington
affair after the violence of the police protection tactics. The next phase
was the universitys action in disciplining four students who had been
involved in the demonstration, and the last phase was the students
218 Charles Antaki
unions decision to protest against the disciplining.
Figure 5.2 shows part of one typical respondents causal skeleton. It

shows one causal link, and what the respondent said about why she
thought the antecedent caused the effect.
Causal structure. If we wanted to concentrate on the structure of the

explanations of the students occupation, we could ask
220 Charles Antaki
Effect
Cause
'Extreme left-wing
attitudes'
The occupation
of University
buildings
Justification for this
claim
'Extreme left-wing
(eg RCP) especially
using expulsions as
an excuse to gain
favour'
Figure 5.2 Part of a respondents explanation structure showing the

justification of a causal claim
all the types of question of the students that we asked of the SDP
members in the previous study: for example, whether people who were
for or against the occupation gave different structures (they didnt,
although they put different causes at different points in similar
structures); and so on.
Justifications. But let us concentrate instead on the justifications, since

the value of this paper-and-pencil technique is that they are now very
clear. In the diagram, the justification appears next to the cause-effect
pair. Our respondent claimed that Extreme left-wing attitudes caused
the occupation of the university buildings, and writes, in defence of the
claim: Extreme left wing, e.g. RCP [Revolutionary Communist Party]
especially using expulsions as an excuse to gain favour.
222 Charles Antaki
What does one do with these justifications once one has them
written down clearly by the respondent in this fashion? Clearly one has
to do some kind of paraphrasing to get the sense of what was meant. In
this case, what the student seems to mean is that she feels justified in
claiming that left-wing attitudes caused the occupation by the fact that,
as she saw it, the ostensible expulsion issue was merely an excuse for
the activist policies of an extreme minority.
One can search among the whole corpus of justifications to see if

some kind of identifiable categories of justification emerge. In this
particular study, what emerged was that the respondents used four
kinds of warrant to justify the claims they made:
1 Justifying the link by revealing an extra cause. Here the respondents

revealed an extra level of cause that mediated between the two
steps they had set down. For example, many respondents thought
that outside pressure caused the university to discipline students.
Their justification was to say that outside pressure brought by
government caused the universities to be concerned with their
image of campus discipline, which caused them to make an
example of this particular case.
224 Charles Antaki
2 Justifying the use of terminology. Here the respondents made sense

of their use of a description of the state of affairs brought about by
the cause. One quite common claim was that the universitys action
had caused the loss of student union autonomy. A student making
this claim might give, as a warrant, something that would make it
clear that what happened was genuinely a loss of autonomy - for
example, For the university to discipline the students was to
encroach on the unions traditional right to apply its own
discipline.
3 Justifying by unpacking a hidden aspect of the cause. Here the

respondent would be revealing a new or unsuspected aspect of the
cause which made the link with the consequent clearer. For
example, one respondent claimed that Dissatisfaction with
government policies caused a history of incidents. Her warrant was
to unpack the cause to reveal that there was general student anger
at the way government is running down the educational system and

especially treatment of students.
4 Historical background. Occasionally the respondent would use the

opportunity to write a paragraph or sos worth of background
information which covered a good deal of ground, and was difficult
to melt down into one particular category of cause.
226 Charles Antaki
It must be said, however, that identifying a statement as being

unequivocally one or other of the above justifications was hard. Partly
this was because of the compact, almost telegraphic style adopted by
most of the respondents. But it was partly attributable to the possibility
that the respondents simply could not cope with the instruction
formally to justify their claims. Respondents knew that something had
to be written about the links they had made, and every respondent
accordingly wrote something. One respondent was clearly au fait with
what was asked of her, and produced responses of the form A justified
B in virtue of the fact that C is a precondition of B, and A brought
about C. But this was unusual. What is likely is that the formality of
articulating what is normally a highly implicit part of dialogue was just
too contrived, and made some respondents produce spurious responses
in an unsuccessful effort to bring into conscious expression the
warrants that they silently invoked in more normal talk.

As I mentioned at the start of the chapter, the advantage of this way of

working with explanations is that it respects peoples ability - in fact
their right - to organize their explanations in ways that a reductive
approach would simply not allow. The respondents are using their own
language in which to express their thoughts; they are expressing as
many causes as they like; they are putting them into whatever
organization they like (within the boundary, of course, of causation);
and they are showing their justifications in whatever sense they feel is
appropriate.
228 Charles Antaki
One thing I would like not to do is to make any strong claim that
this way of working reveals peoples cognitive representation of a
causal sequence. This is the point of view expressed by Kelley (1983)
and by Bowerman (1981), who want to claim that people store
representations of causal sequences in something that looks, at first
blush, rather like the tree structures in Figure 5.1. That view is a
regression to thinking of peoples explanations in a social vacuum.
Here the point is that the structure is thought up for a specific purpose
and is addressed to a specific person. I would have no particular
commitment to expecting to see the very same structure again, and I
would fully expect the same person to draw up quite a different
structure, with different causes differently justified, when they were
doing it for someone else, or less formally, and so on. That is not to say
that what they say to me is meaningless; far from it. There is nothing
meaningless in saying that these students (for example) expressed
themselves to a lecturer in a certain way (and that they varied in how
they expressed themselves according to, say, whether they were for or
against the occupation). The point is to acknowledge, and make part of
the story, that this was how they addressed themselves to one particular
person at one particular time, and that they might well address
themselves differently to each other, to the police, and so on.
On the debit side there is the question of validity.

I started off by saying that we must respect
ordinary talk; yet I have presented a paper-andpencil task quite different from the ebb and flow
of conversation. I have to admit that this is a
problem. There is little doubt that the technique
encourages a rationality that may not be there in
other media, and requires of the respondent a
certain commitment to understanding the event
under discussion and to looking as if they
understand it - which might, again, not be the
case with a different medium. Moreover, the
technique is perhaps too committed to what
Harre and Secord (1972) call the open souls
doctrine: taking what respondents say too
literally and making little allowance for, amongst
other things, irony, facetiousness, exaggeration
and the complex use of metaphor and allusion. It
is perhaps too rigidly literal in its treatment of
what people say to be a good net for catching
subtleties of explanations in actual exchanges.
But it has useful qualities for explanations and
accounts which can be expressed more
formally.Problem solving and reasoning in
ill-structured domains
James F. Voss
6
In the most recent decades psychological research has been in a period

of transition and, with respect to the study of higher mental processes,
the transition has included an increase in research on problem solving
and reasoning. Moreover, the transition has involved both conceptual
and methodological trends. Conceptual trends have included attempts
to develop an understanding of mental structure and function.
Expressions commonly used in relation to this concern are
representations, declarative knowledge, procedural knowledge,

and strategies. Methodologically, procedures have been developed
which enable the investigator to study mental processes in a relatively
wide variety of task situations. This chapter is concerned with one
type of task situation that is found in the study of problem solving and
reasoning, namely that of collecting and analysing verbal protocols,
and in particular with protocols that are collected in relation to illstructured problems.
The collection of protocols is of course not new. Ericcson and

Simon (1984) have summarized research involving protocol analysis,
considering the virtues of the procedure as well as its limitations.
However, the current interest in protocol collection was given strong
impetus by the development of computer simulation techniques, for
investigators wanted to relate the solution steps found in the protocol
to a model of those processes (for example Newell, Shaw and Simon,
1958). Newell and Simons book Human Problem Solving (1972) contains examples of protocol analysis taken from a number of domains,
including geometry and cryptarithmetic. Moreover, the protocols
presented are portrayed in terms of problem behaviour graphs; such
graphs represent the step-by-step development of the solution.
The problems typically used by Newell and Simon (1972) as wellas by

a number of other investigators were well structured: that is, the goal
of the problem was well defined, the givens of the problem were
presented in the problem statement, and the constraints of the problem
were provided in the problem statement or were known to the solver.
Problems that are well structured include solving algebra or other
mathematics problems, puzzle problems, and physics problems such
as those found in undergraduate textbooks. Furthermore, the solutions
of well-structured problems typically have been determined, and they
have consensual agreement in the sense that the experts in the field
agree upon a particular solution.
In contrast to well-structured problems, in ill-structured problems

the goals are vaguely defined, the constraints may not be stated but
may be generated in the course of the solution, and the givens may be
minimal. (Because the extent to which the goals, constraints and
givens of a problem may vary, the well-structured/ill-structured
distinction should be viewed as a continuum rather than a dichotomy.)
In addition, ill-structured problems typically do not have solutions
which have consensual agreement among experts in the field. Illstructured problems tend to be found in fields such as social policy,
political science, sociology and, to some extent, economics, although
cutting edge research in the natural sciences also involves dealing
with ill-structured problems.
An additional aspect of the information processing model involves

the use of the concepts of task environment and problem space. The
task environment consists of the situational context of the problem.
The problem space consists of a mental space of the solver which
contains all possible states of the problem, correct or incorrect, the
problem operators, and the problem constraints. In addition, the
solvers understanding of the problem statement constitutes the initial
state of the problem, while reaching the goal - that is, solving the
problem - constitutes the final state of the problem solution. Thus,
according to this view, the problem solving process consists of the
solver proceeding from one state of the problem to the next until the
goal is achieved. Moreover, the movement from state to state is
presumed to take place via the use of an operator. For example,
considering the problem of adding a column of numbers, the solver
would use the operator add and apply it to each number as the
individual was cumulatively adding the column.
In the solving of more complex problems, and especially in the case

of the ill-structured problems, the solver develops a representation of
the problem and subsequently provides a solution to the problem. In
physics, for example, the solver, especially the expert solver, may read
a problem, interpret it by examining the concepts and relations of the
problem (perhaps by constructing a drawing), and conclude that the
problem is perhaps a kinetics problem (cf. Larkin et al., 1980). Such
a classification constitutes the representation of the problem, and upon
classifying the problem the expert solver then applies the appropriate
equations. With respect to ill-structured problems, developing the
problem representation may be quite complex, with the solver
delineating the causes of the problem as well as its constraints. The
solution then consists of providing a solution and justifying it;
justification occurs as the solver indicates why the proposed solution
will work and considers the possible difficulties of the proposed
solution and how these difficulties may be met.
6
Analysing protocols of solutions to ill-structured
problems
The question of how to analyse protocols obtained in the solving of illstructured problems does not have a simple answer. In our work on
this issue (Voss et al., 1983; Voss, Tyler and Yengo, 1983) we spent
considerable time trying to capture the contents and structure of the
protocol, approaching the question via use of a number of different
methods. The expert protocols are relatively lengthy and, having been
working in the area of text comprehension, we were predisposed to

use a form of propositional analysis. However, it became clear quite
quickly that such an analysis would not capture the organization and
structure of the protocol. We tried other means such as classifying
sentences in terms of their content, but this also was not fruitful.
Moreover, from a problem solving perspective we tried to determine
whether particular strategies were used such as top down, breadth
first, but this also failed to capture the protocols structure and
contents.
Problem solving in ill-structured, domains 11
A more positive outcome occurred when we applied Toulmins

(1958) model to the protocols. Toulmins model consists of taking a
single assertion and viewing it in terms of a datum and a claim
component. For example, consider: A Soviet-United States arms
agreement will likely produce a reduction of the United States troops in
West Germany. The datum involves the occurrence of an arms
agreement, while the claim is the effect that presumably will be
produced by the datum. In addition, the Toulmin model includes the
concepts of warrant and backing. The warrant is what enables a person
to make the claim; it is a sort of minor premise. In many cases warrants
are not stated explicitly; the argument structure then embraces what
Aristotle terms an enthymeme. However, support for the assertion
and hence the warrant is often provided in the form of backing.
Backing refers to some type of supportive statement which may take
the form of a fact or an assumption. Thus, in the present example, one

may state that the need for United States troops in West Germany
would be reduced.
The final two components of the Toulmin model are the qualifier
and the counter-argument. Qualifying involves asserting that the
datum-claim relationship holds only under a particular set of
conditions. Expressions typically used in such a case are under such
conditions, given that or in these circumstances. The counterargument refers to a content which raises a counter-position to the
datum-claim relationship, and it is often signalled by but or
however. In the present example a counter-argument would be that
with a decrease in nuclear arms there would be an increased need for
conventional forces.
The unit of analysis in the Toulmin model is the single assertion.

Applying it to our protocol analysis, we made two assumptions which
extended the model to cover argument structures longer than one
statement. One assumption was that the claim of one assertion could
serve as the datum of another assertion. The second assumption was
that backing could consist of datum-claim assertions. These two
assumptions thus permitted us to delineate argument chains (cf. Voss,
Tyler and Yengo, 1983).
While the adaptation of the Toulmin model constituted a significant

step in the protocol analysis effort, it also became apparent that one
thing the Toulmin-type analysis did not do is capture the problem
solving nature of the protocol. Furthermore, reading of the protocols
made it clear that the solution was being driven by use of general
problem solving strategies. In order to capture this, we viewed the
protocols in terms of goals and subgoals and examined them in relation

to the strategies used to accomplish the goals. What became apparent
from this analysis is that, in solving ill-structured problems, the
solution process was driven by a problem solving control structure and
an argument or reasoning structure which was subordinate to and
driven by the problem solving structure. Moreover, the strategies
employed in the solving of ill-structured problems were found to be
largely those described by Newell (1980) as weak problem solving
methods, for example: analogy; generate and test; means-ends analysis;
decomposition; and problem conversion.
Given our experience in the analysis of protocols generated in the

solving of ill-structured problems, three conclusions regarding how
much analyses should be conducted may be stated. These conclusions
do not constitute a formal system; nor can we even state that there is a
single way to analyse. Yet the conclusions seem to be warranted.
244 James F. Voss
Firstly, the analysis of protocols collected in relation to the solving

of ill-structured problems must be model driven. By model driven we
mean that the analyser needs to make some basic assumption about the
model of protocol structure. The protocol contents do not jump out at
the analyser, clearly indicating the structure and content organization.
This is an extremely important point. When individuals collect a
protocol and then ask how to analyse it, they often expect that a
magical coding system can be employed which makes the cogent
components obvious.
Problem solving in ill-structured domains 245
A question then that may be raised is how the model of protocol

structure should be developed. The answer is that as a starting point the
investigator should look at the goals of the study. Sometimes the
investigator may want the protocol to provide nothing more than an
answer to a question. In this case the analysis may be relatively
effortless, although we have observed individuals over-analysing the
protocols, attempting to squeeze out information irrelevant to the issue
at hand. At other times individuals may want to develop a highly
detailed analysis of structure and content. This goal requires a detailed
exposition of the model. Such a process, moreover, is likely to be
iterative, proceeding until one believes the structure and contents are
successfully captured. In fact, the need to be iterative is more the rule
than the exception.
Secondly, reliability is of course an important issue. One needs to be
246 James F. Voss
able to develop a coding system which permits analysis by two or more

individuals to yield equivalent outcomes. To accomplish this task we
have found that the best procedure is to structure the analysis as much
as possible into basic categorizations and rules. This is also an iterative
procedure in that categorizations and rules need to be generated until
the contents and structure are appropriately and hopefully sufficiently
described. When one is finally satisfied that the coding system is well
articulated, then reliability may be established. This is done by having
two or more individuals analyse a large sample of protocol contents.
Providing training or practice to individuals naive to the coding system
is not uncommon, with reliability checks being made after such
training.
Thirdly, whenever publishing work related to protocol analysis, the

reader should be provided with an original protocol or segments
thereof as well as the analysis of the protocol, so that the reader may
follow the logic (or the lack thereof) of the protocol analysis. If it is not
feasible to print such an account, then an address should be provided to
which the reader can write in order to obtain copies of the original
protocols and the analyses thereof. This is important because what we
do not need in the field of protocol analysis is an increasing
multiplicity of protocol analyses by which the reader must accept on
faith what was done. Such a state of affairs will destroy the scientific
credibility of the entire endeavour.
Taking into account these three principles, the general conclusions

are that the analysis of protocols in the context of the solving of illstructured problems is model driven, a fact that makes it incumbent
upon the investigator to think about the nature of the questions that are
being addressed. In addition, once a model is determined that is useful,
248 James F. Voss
the investigator must maintain the criteria of reliability and

communicability.
Empirical example
A protocol of the Voss et al. (1983) study is presented below. The

protocol was generated by a US expert on Soviet Union affairs who
was responding to the question of how he would improve crop
production in the Soviet Union if he were head of the Ministry of
Agriculture of that country. The whole protocol is reproduced so that
we can see the extent of integration and elaboration of the respondents
problem solving.
I think that as minister of agriculture, one has to start out with the realization that there are certain kinds of special agriculture constraints
within which you are going to work. The first one, the most obvious

one, is that by almost every count only 10 per cent of the land in the
Soviet Union is arable. This is normally what is called the blackland in
the Ukraine and surrounding areas. And secondly, even in that arable 10
per cent of the total land surface, you still have climate for instance,
over which you have no direct control. So that is sort of the overall
parameter in which we are working.
Now, we have traditionally used in the Soviet Union three kinds of
policies to increase agricultural production. Of course, agricultural
production has been our Achilles heel and something that we have
inherited from the time the tsars freed the serfs. Even before then, the
agricultural production was low because historically the aristocracy had
no need to fend for itself, as it turned to the tsar for its support and
hence never, like the English aristocracy for instance, introduced
modern methods of fertilization, never went to enclosures or
consolidations of lands, never experimented with crop rotation. That
was passed on to the peasants and throughout the period when the
peasants had been freed to do what they willed with the land. They are
responding with the old, rather inefficient ways. At any rate, we have
had three different ways by which we have tried to increase agricultural
production.
The first one might be labelled exhortation. The Soviet approach to
agricultural production is to mount campaigns continually to call for
more effort on the part of the peasants and agricultural workers, and to
put more effort into their labour activities for agricultural production.
Those things are mounted periodically. Quite frankly, I think that they
are a waste of time and energy and they really, as minister of agriculture
I must say, that they really do nothing more than give the party a sense
of false importance, because it is normally incumbent upon the party to
develop these ideological indoctrination campaigns and the notion of
mind over matter, in this case, has not paid off; and it leaves the party
with the belief that ideological, and if you will excuse the term, spiritual
policies can overcome objective limitations. So, I would not emphasize
exhortation very much.
It seems to me that the second way that we traditionally go about
trying to increase agricultural production is through constant
reorganization that leads to confusion, that leads to mismanagement
and that has forced a mind set upon the peasant agricultural worker of
sort of laying back and waiting because this too will also pass. We have
gone through collectives, state farms, and machine tractor stations. Our
latest attempt at reorganization - which has knocked down, by the way,
the number of collectives from about 250 000 to 30 000 in the last five
years - is through the development of what are called agroindustrial
complexes, in which the former collective farmer becomes a wage
earner.
So, I think we have to tend to the nature of agricultural production. I
want to say one thing, and I have to recognize that this is clear as day,
and that is that in all of these cases more or less except for stringent
250 James F. Voss

ideological periods we have always allowed the private crop to exist
even though we take it to be a much more primitive, less historically
progressive form of agricultural production. We must realize that in
terms of our food staples, even until today, roughly 40 per cent of the
food staples are grown on the private plots.
The third thing that we have done - and this is where I would like to
start off in terms of turning around agricultural production - the third
thing we have done is we have tried to mechanize, and I want to use
that in the broad sense because it is not the word I want, we have tried
to mechanize industrial production. Not just mechanize it but also
introduce scientific advances in Soviet production.
I think as a starting point as minister of agriculture, my first aim
would be to get monies to invest in this further mechanization and
further application of scientific techniques of agriculture, to agriculture
situations. Even though we have mechanized to some extent, it has
been a rather crude form of mechanization. It has been rather low level.
It is not coherent, it is not consistent, it is - we have the same old
problem that we, if we develop tractors or we produce tractors, we
produce a thousand tractors, we have no parts to service them when
they break down. We do not have adequate transportation supplies or
transportation networks to carry the product we do have to the urban
markets. We have been woefully lacking in a methodical application of
fertilizers to our agricultural sectors in society. We are much more like
a Third World than an industrial world in terms of the lack of use of
fertilization. We still do not have very scientific management in terms
of crop rotation, and because of all this we still have a rather labour
intensive agricultural production system and therefore production per
unit is very marginal.
First of all, what I would like to do is to - I would have to fight very
strongly in the party and the government - to redirect the investment
ration going to agriculture over heavy industry and even light industry
in the way it has been in the past. Though we have in the last ten years
shifted investment politics so that more and more is coming into
agriculture and away from industrial production per se, it still is not
anywhere near the break even point, even though for both international
political and domestic reasons, agriculture is our Achilles heel. If we
simply do not produce enough we will become more and more
vulnerable to dependency
upon the West for agricultural production and we will, if we do not
have enough foodstuffs internally, develop a more unstable regime.
There will be lack of support. We have developed a programme where
the support from the people is going to stem at least in some basic way
from demand satisfaction, and yet we have not been able to satisfy a lot
of the basic food demands of large parts of our population.
So, I would first have to fight in decision making circles, to once
and for all recognize that the construction of socialism is the

construction of an entire society that is self-sufficient and does not have
to depend upon especially potential adversaries for crucial parts of its
basic resources. That means that we have a high enough industrial base,
we have the war technology equivalent to the West, we must now
redirect in a major way our investment policy towards agriculture,
towards mechanization, towards the infrastructure around
mechanization, towards the transportation programme, towards the
plastic bag programme so that whenever the damn fertilizer is
packaged it does not sit out in the lot as it does. We lose one-half of our
fertilizers because it rains on paper packages. What I mean by
infrastructure, there is a whole series of secondary enterprises that we
have to make. Secondly, we have to develop an education programme
among the agricultural workers to teach them the more rational use of
mechanization, mechanized planting and harvesting and fertilizer and
crop rotation. Even in this day and age its shocking how little they
know and how backward many of our farmers are. As minister of
agriculture, I would like to say as an aside, since I probably will be
defecting to the West with the next ballet, that I blame the party for
this. The party has always insisted that the management of state farms
and even the collectives be people who have strong party credentials
and who, quite frankly, do not know anything about agriculture. So, I
would like to see the development of a special institute that trains
managers of agricultural enterprises and trains them in the most up-todate, modern techniques. Even if we start without investment policy,
we start to reap the benefits of fertilizing and mechanization if we have
administrators in these three kinds of farms: we have now the
agroindustrial enterprises, which are becoming the largest; the state
farms; and the smaller and smaller number of collectives. If we do not
have the right agricultural managers, all the most modern techniques
and methods of farming will go out the window.
The second thing that I must do is to raise, rather dramatically, the
wage levels of the agricultural worker. Ideology aside, and I want to be
pragmatic for a minute, what we have is one hell of a large agricultural
working class. We can call them collective farmers, we can call them
state farmers, but there, it is an agricultural working class. And it
compares shockingly low in terms of wages, in terms of security, in
terms of insurance, in terms of welfare benefits, in terms of living
conditions. It is like a second class, a lower class, an underclass
compared with the urbanized working class in our country. We are not
going to get anywhere in terms of agricultural production until we, the
state, show them that we consider them equal citizens to the urban
proletariat and we treat them equally in terms of our service, welfare,
insurance and wage policies. Again, this is going to rankle with the
ideological facts of those people in the party who for some reason see
agricultural production as a free, as a traditional
kind of production, and I guess we would like to make synthetic foods
so they would not have to have farmers. But, to deal with the problem
so that we do not increase our vulnerability, our dependency upon the
252 James F. Voss

West, we are simply going to have to raise the agricultural worker to
the same level and bring him in as a full citizen, both participating in
policies the same as our other citizens and also sharing the distribution
of goodies and rewards in our society to the same extent.
The third thing that I would do would be to work out a system
whereby the private plots are not considered a temporary historical
phase but are an integral part of our agricultural system. Whether we
like it or not, that is where we get a lot of our foodstuffs, and I think
that we could work it into our system so that the agricultural,
agroindustrial enterprises we are now calling the private lands could be
given to those farmers who are most productive as an incentive to work
even harder for the agroindustrial enterprise. As well, I think we need maybe I mentioned it before but I am going to go over it again - as
well, I think we need strong investment, not only in the day-to-day
things that increase production, we also need more investment in
research and technology. So that is one area to deal with.
The second area that I think that we have to deal with is the question
of what are the marginal payoffs for trying to increase the amount of
arable land. I think that by doing a series of studies we can identify
those lands where a massive irrigation would bring them up almost to
the productive level of the blacklands, and I would target especially
those where just moderate irrigation or little irrigation would be needed
to bring them up to such a level of production and I would develop a
very, very extensive irrigation network for those lands. I think
irrigation is a major problem that we really have not dealt with
adequately, sufficiently, coherently, and in a centralized manner,
believe it or not.
The third area that I think we have to think about is not only the
private plots, but also the use of the private market as a mechanism, as
an incentive mechanism for further agricultural production. Frankly,
right now, a farmer that works on a collective and has a little private
plot on the side and sells produce privately in the market makes more
from the little bit of tilling of the private plot and the sales on the
market than he does from his full days work on the collective. If we
are going to get rid of this private notion of agriculture, then we have to
make collective work much more rewarding, even if it means further
subsidization by giving inflatably high, artificially high prices to our
collectives, to the state farm workers, to the agroindustrial enterprises,
to those workers and subsidized foodstuffs. We have one hell of a
problem because what we have been doing is (given how low the
productivity is per person) we have been paying them wages - even
they are low by urban working class standards - we have been paying
them wages that are higher than the prices that we get for the food
when we sell it to our people in the urban areas. We have been
subsidizing our whole food programme to the tune of 20 or 30 per cent.
We have to increase that subsidy if we want to get that production up to
the point where at some point we would not need a subsidy at all.
As previously noted, our model assumed that the protocols had
two structures, a general problem solving structure and a reasoning
254 James F. Voss
structure, the former being assumed to drive the reasoning structure.

Table 6.1 presents the operators used in reference to each of these
structures. An operator here is a link between statements; for example,
Historically, the aristocracy had no need to fend for itself is given as a
reason (RREA) for Agricultural production (being) low. The first
three G operators listed in table 6.1 are the basic problem solving
operators and are self-explanatory. The GIPS operator was applied
when the solver represents the problem in relation to how it is to be
solved. The GSUP operator was applied when a subproblem or
constraint had already been identified, and the GEVA operator was
applied in reference to specific solutions which are accepted or
rejected.
Table 6.1 Operators of the problem solving and reasoning structures

Problem solving
operators (G Structure)
GCON State constraint
GSUB State subproblem
GSOL State solution
GIPS Interpret problem statement
GSUP Provide support
GEVA Evaluate
GSUM Summarize
Reasoning structure
operators (R structure)
RARG State argument RSAS State
assertion RFAC State fact RPSC
Present specific case RREA State
reason ROUT State outcome RCOM
Compare and/or contrast RELA
Elaborate and/or clarify RCON State
conclusion RQUA State qualifier
The most cogent operator of the R structure is that of stating an

argument (RARG), which frequently is a stated subproblem or
constraint. The use of other operators then follows, including RSAS,
which also involves statement of an argument but the argument is not
part of the problem solving structure. The RFAC, RPSC and RREA
operators are used to support a previous statement , while the ROUT
operator is used when an outcome of a previously stated action is
described. The RCOM and the RELA operators are self-evident, while
the RCON operator is used in stating a conclusion, which in turn
usually ends a line of argument. The RQUA operator produces a
qualification of the preceding statement.
I Given the above operators, the protocol analysis began with segmentation of the protocol
into idea units, with each unit typically consisting of one sentence. Following
segmentation, our version of the problem behaviour graph was constructed. This was done by
256 James F. Voss

determining the operator that applied as the solver moved from oneGIPS (RARG)
Historically, agriculture has been a problem in the Soviet Union
(RFAC) Problem has been inherited from the time the tsars freed the
serfs (RFAC) Agricultural production was low even before then
(RREA) Historically, the aristocracy had no need to fend for
itself (RCOM) Was not like English aristocracy
(RPSC) Never introduced modern methods of
fertilization (RPSC) Never went to enclosures or
consolidation of land (RPSC) Never experimented
with crop rotation
(RFAC) Agriculture problem was passed on to peasants so they
could do what they willed with the land
(ROUT) They responded with old, inefficient ways (RFAC)
USSR had three different policies to increase agricultural production
(RPSC) Exhortation
(RELA) Campaign for more effort on the part of the
peasants (RCON) Was waste of time and energy
(RREA) Only gave the party a sense of false importance(RREA) It is

incumbent upon the party to develop these campaigns (but they
havent paid off)
(RREA) Party believes that ideological policies can pvcrcome objective
limitations
(RCON) I would not use exhortation (RPSC) Reorganization
(ROUT) Leads to confusion, mismanagement, makes peasants laid back (RPSC)
Reorganized collective state farms, machine tractor stations
(RPSC) Now have agroindustrial complexes, reducing number of collectives and
making farmer a wage earner
(RQUA) Have always allowed private crops to exist
(RQUA) Except in astringent ideological periods (RELA) It is taken to
be a more primitive form of production (RFAC) Private crops account for 40 per cent of food
staples (RPSC) Mechanization
(RCON) This is where I would start my solution
(RELA) Have tried to mechanize agricultural production more
(RELA) Tried to introduce scientific advances in agricultural production
Pr
ob
le
m
so
lvi
ng
in
ill
str
uc
tu
re
d
do
m
ai
II
GEVA
(RARG) Mechanization low level, crude, inconsistent

(RPSC) If we make tractors, then we have no repair parts
(RPSC) No scientific management of crop rotation (ROUT) Because

of all this, we have labour intensive system
(RREA) Have developed programme where support from people comes from demand
86
Ja
m
es
F.
Vo
ss
satisfaction
(RQUA) But still haven't been able to satisfy force
demands
III
GEVA
(RARG) Support infrastructure around mechanization (RPSC)

Support
transportation
programme
(RPSC)
Support plastic bag programme

(RFAC) Lose when it sits out in paper bags (RELA)
Support series of secondary enterprises
IV
GEVA
(RARG) Need to teach agriculture workers rational use of mechanization

(RELA) Mechanized planting, harvesting, fertilizing and crop rotation (RREA)
Farmers are backward
(RREA) Blame the party for this
(RREA) Management tire people that have strong party credentials, and dont know
about agriculture
(RCON) Need special institute to train managers
(RREA) If don't have right managers, investment in agriculture useless
Pr
ob
le
m
so
lvi
ng
in
ill
str
uc
tu
re
d
do
GEVA
(RARG) Large agricultural working class which compares low in wages, security,
insurance, welfare benefits, and living conditions
(RELA) Can call them collective farmers or state farmers, it is the working class
(RCOM) It's like a lower class compared with the urban working class
VI
GEVA
(RARG) Make private plots integral part of system
VII
(RREA) Because it is where a lot of foodstuffs come from (RPSC) Use as incentive for farmers to work harderGEVA
(RARG) Identify lands to irrigate

(RREA) To raise levels of production
(RELA) Target areas where little to moderate irrigation would he needed (RELA)
Develop extensive irrigation system there
(RREA) Irrigation is major problem not dealt with adequately, sufficiently,
coherently, or in centralized manner
VIII GEVA
(RARG) Have to make collectives more rewarding
(RREA) Now worker gets more from private plot than whole day's work on collective (RPSC)
Subsidize collectives to give artificially high prices (RREA) To get rid of private notion of
agriculture (RFAC) Wages paid are lower than urban working class (RFAC) Have been paying
wages higher than what is received from food (RFAC) Have been subsidizing 20-30 per cent
88
Ja
m
es
F.
Vo
ss
262 James F. Voss
Pr
ob
le
m
so
lvi
ng
in
ill
str
uc
tu
re
d
do
m
ai
ns
(RCON) Have to subsidize until production is up enough not to have to

subsidizeidea unit to the next, the idea units thus representing the
respective states of the problem as the solver proceeds along the
solution path. The analysis of the protocol in terms of the states and
operators is provided by the problem behaviour graph presented in
figure 6.1. The reader may compare the protocol, as presented in
paragraph form, with figure 6.1. The protocol begins with a statement
of constraints, followed by interpretation of the problem statement. The
problem is represented in GSUB and the solution is presented GSOL.
The reader is asked to note that the contents of the boxes that have
brackets are broken down and presented in tables 6.2-6.4. The boxes
containing brackets are denoted by roman numerals I-VIII. One other
aspect of the protocols is that a dashed line describes an implicit
statement, as indicated by the contents.
263 James F. Voss

A theoretical issue that may be raised is how the protocol contents

should be interpreted. To what extent, for example, do the contents
reflect the knowledge structures of the individual? We have gathered
data germane to this question, and the most reasonable answer is that
the protocol contents indicate how the individual would solve the
problem at that particular time in that context. Two results provide for a
more complete answer.
264 James F. Voss
We did request that a few experts return approximately two weeks

after their initial session. We asked one to solve the problem again and
the other to recall his solution. What we found is that the solutions were
shorter, but that they maintained the general structure. What was
different in content, but not structure, was the argument development.
The experts, for instance, did not remember very well what they had
provided as examples.
In another case
an
expert
on
Central America
agreed to attend a
graduate seminar,
at which time he
was given two
problems
and
asked to provide a
think
aloud
protocol to each
of them in the
presence of the
class. He received
one problem and
provided
a
solution,
then
received
the
second problem
and provided a
solution. One of
the
problems
essentially asked
him to suggest a
new US policy
regarding
El
Salvador if he
were a US State
Department
official. (This was
before the US
preoccupation
with Nicaragua.)
He began with the
solution
phase,
basically without
the verbalization
of
a
representation
phase. When he
had finished, I
asked him why he
did not discuss
the causes of the
problem
and
related matters.
He immediately
answered
that,
first of all, from
the
problem
statement
one
could infer that
the current policy
was not working
because
the
problem asked for
a
change.
Secondly,
he
pointed out that
Fi
gu
re
6.
1
G
ra
ph
of
th
e
pr
ot
oc
ol
pr
es
en
te
d
PROBLEM
STATEMENT
he had
;
GCON
10% arable
land
1 GSUP
Blackland in Ukraine
Pr
ob
le
m
so
lvi
ng
in
U
ns
tr
uc
tu
re
d
do
m
ai
ns
been giving much thought to the problem, having appeared on TV

and given speeches on the issue, and he felt he was well aware of
the background. This answer is quite instructive, for it
demonstrated how contextual components can influence the
protocol contents. But in addition, the answer also suggested that
experts, when dealing with a problem they have thought about only
infrequently or not at all, use the problem representation phase to
think through the problem; this process involves an extensive and
quite directed memory search. Incidentally, the second problem
this expert received pertained to Mexico, and in this case he began
by representing the problem via the more typical type of
representation development.
The findings described in the immediately preceding paragraphs

thus indicate that ill-structured problem solutions may vary with
context. At the same time, such results should not be interpreted as
chaotic responding. Instead, the basic solutions will generally not
change unless new information becomes available, either
externally or via further search. But experts would be expected to
be flexible when providing a solution to another expert as
compared with a novice. Indeed, it is the novice who would likely
not be capable of such flexibility.
We turn now to the advantages and disadvantages of the

analysis presented in the preceding section. We found this analysis
useful in our efforts to determine the nature of problem solving
activity as it is evidenced in relatively long protocols that are
solutions to ill-structured problems. This particular type of analysis
has a number of virtues, including the following:
1 There is high inter-rater reliability with respect to both segmentation and operator classification.
2 The analysis provided for a description of the protocol which

was meaningful psychologically, in the sense that problem
solving and reasoning components were explicated.
3 The analysis was possible with the use of relatively few

operators, an interesting commentary on the processes of mental
search that take place during problem solving.
4 The analysis was applicable to individuals having large

differences of knowledge related to the problem area, and to a
variety of problems.
5 The analysis produced data which could be employed to make

comparisons across individuals and problems.
As to disadvantages, the analysis is time consuming and labour

intensive, especially in the earlier phases. In addition, the analysis
provides one type of data, and the investigator may readily want to
look at the protocol in different ways to address other questions.
It is important to reiterate a point made earlier in this chapter,

namely, that the presented analysis is appropriate only if the goals
of the investigator are essentially the same as we had in developing
the analysis. To illustrate this point, we recently completed some
work in which we collected protocols that were answers to
questions in economics (Voss et al., in press). The participants were
naive (having no formal training in economics) or novices (having
had one or two courses in economics). One question asked for
factors that can influence a change in interest rates. One person
began his response with: I guess, from an idealistic viewpoint, I
look at interest rates as the cost of money. Again, its a supply and
demand type of thing. If money is in short supply, it costs more. In
that particular study we were interested not in providing the
sentence-by-sentence analysis of a protocol, as described in this
chapter, but in specific questions such as Did the individual
transform the problem, Did the individual use any type of rule at
arriving at an answer? Thus our purpose led us to a different type
of analysis. Of particular interest is the fact that it is often

necessary to study the protocols in order to consider the nature of
the model that will provide for extracting the most interesting
information. Our experience suggests that, as stated, protocol
analysis is model driven and serves not as an end in itself but as a
tool to address issues of interest to the investigator.
Note
Preparation of this chapter was supported by a

grant of the Office of Educational Research and
Improvement of the United States Department of
Education for establishing the Center for the
Study of Learning at the Learning Research and
Development Center of the University of
Pittsburgh. The viewpoints expressed in this
chapter are not necessarily those of any of these
organizations.Narrative structures in social
explanation
Mary M. Gergen
In this chapter, I would like to present the idea of accounts as stories not random sequences of action and reaction, but coherent unfolding
narratives of human conduct. The idea of looking for narrative
structures has a long history in literary theory, of course, and even in
the social sciences it can be traced back to the early 1920s (and
specifically to the seminal ideas of the Russian folklorist, Vladimir
Propp). Nevertheless social psychology has been somewhat immune to
this way of thinking, and we can start by having a look at why this
might be so.
Narrative structures in social explanation 283
The
origin
the study ofdeveloped
social explanation
in social psychology
was based
the
notion
that of
explanations
out of
a common-sense
psychology.
As toonthe
conceptualized
by
Fritz
Heider
(1958),
everyday
explanations
were
integral
ordinary
business
of dominant
getting along
with others.
Heiders
were
later pressed
into
conformity
with the
experimental
paradigm
inideas
social
psychology,
and what
had
been a rather
loosely
construed
approach
to everyday
explanation
became
reconstructed
as
tightly
organized
theories
of
cognitive
process
(Antaki
and
Lewis,
1986).
As
fitted the laboratory
modeofofeveryday
operation,life,
studies
ofonly
causal
attribution
became
decontextualized
simulacra
bound
to
the
brief time
frame
ofversus
an experiment.
Variables
were developed
that
classified
explanations
into
the
internal
external
and
actor
versus
observer
cells.
The
area
of
attribution
theory
flourished,
and
helped
cognitive
social
psychology
achieve
its
present
hegemony.
From
the(cf.
1960s
to the
present,
attribution
research
exerted
powerful
influence
the
Kelley,
1973;
Jones
and Nisbett,
1971;has
Ross,
1977;a and
see chapter
3 inon
this field
book).
Criticisms of the experimental paradigm and the associated array of mechanistic theories have
been widely voiced in recent years. One particularly relevant line of criticism has challenged the
ahistorical and decontextualized nature of traditional inquiry (Gergen, 1982). Such inquiry remains
generally blind to temporal transformation in phenomena, and it is removed from concerns with the
sociocultural contexts of the people under study. A number of efforts to surmount these problems
are contained in the book Historical Social Psychology (Gergen and Gergen, 1984). This contains a
variety of efforts by European and American scholars to analyse cross-time or diachronic social
processes. This attempt to develop a contextually embedded form of analysis has led in a variety of
directions, one of the most important of which is inquiry into everyday explanations of human
action.
284 Mary M. Gergen
The focus of this chapter is on the analysis of narrative explanation. For many scholars
narratives are seen as providing a sense of coherence and directionality in ones life (cf. Cohler,
1979; Kohli, 1981). As the psychologists de Waele and Harre (1979) have indicated, life events are
made intelligible by locating them in a sequence or unfolding process. If events are not tied to a
larger story they lose significance. So, for example, if someone mentions that it snowed 10 inches
on 14 February, this remains an inconsequential fact until connected to a story about ones efforts to
travel that day to meet a loved one and the end of the relationship precipitated by the failed tryst.
Narratives are often shaped so that culturally valued activities and endpoints are highlighted for self
and others. Events that are encapsulated in stories are presumed to have some significance in the
lives of the characters and those who tell and hear the story. For example, Rosaldo (1986) has
suggested that the anthropologist may gain an understanding of which activities have great
importance to the people by listening to the stories that are told and retold among themselves.
Joseph Campbell (1956) has suggested that the most basic myth for humankind, or monomyth as
he calls it, is the story of the hero who overcomes obstacles to reach some transcendent goal. A
similar claim for a universal plot was made by Lord Raglan (1956). While these myths are framed
out *One might well question to what extent the requirements of what defines a well-made story or a universal plot are not
influences by androcentric forces in society. For example, metaphorically speaking, the ideal Western narrative follows the
dramatic form of the prototypical male performance during intercourse - linear motion, with increasing intensity, directed to an
endpoint, with a climax and fade-out. One might envision a prototypical narrative form based on female sexuality that might be
less goal directed and more episodic and have less emphasis on a climax and more on the foreplay, that is, the process of the
interactions.
of a peculiarly male perspective (the lone male hero leaves home), the basic story is a well-known
framework frequently employed by people to lend meaning to their lives.1
The concern with narrative account of human action is shared by scholars throughout the social
sciences and humanities. The social significance of myths, legends and tribal stories has been a
focal consideration for folklorists (Propp, 1968; Young, 1987), anthropologists (Campbell, 1956;
Rosaldo, 1986), political theorists (Tololyan, in press), psychiatric investigators (Spence, 1982),
developmental psychologists (Bettelheim, 1976) and philosophers (MacIntyre, 1981). Of particular
relevance, a number of investigators have discussed the ways in which narratives define reality for
groups of people. Narratives are seen as critical to the ways in which people understand themselves
and their lives, past and present, and how their futures might be (Sarbin, 1986; White, 1980).
Tololyan (in press), for example, has analysed the lives of Armenian terrorists; he argues that their
plans and activities are the result of being imbued in a cultural tradition where the stories, songs
and religious litanies create heroes of young martyrs who die fighting for the honour of the
Armenian people. Tololyan believes that the cultures narratives are primary factors in promoting
present day terrorist activity among the Armenians.
286 Mary M. Gergen
Our own line of inquiry has extended this concern with the importance of narratives in peoples
constructions of their lives (Gergen and Gergen, 1983). From our perspective, traditions of story
telling, dramatic performance, literature and the like have generated a range of culturally shared
forms of emplotment, or narrative forms. When the individual attempts to understand him/herself,
these culturally embedded forms furnish a repertoire of sense making devices. It is through
embedding ones actions within one or more of these forms that ones actions take on meaning;
they belong to a person with a certain past, heading in a certain direction, and with a future that will
represent an extension of this past (Gergen and Gergen, in press). Yet, as we have also proposed,
narrative constructions are not the mere product of cultural history. The particular form that they
acquire for any person is an outgrowth of the social relationships in which one is currently
embedded (Gergen, in press). Ones narratives typically include the positioning of others in
relationship to self. Without their particular functioning within the narrative, ones own position or
identity is threatened. Likewise, others self-narratives contain constructions of other individuals
embedded in their mutual social surrounds. Thus, the narrative constructions within a community
of interlocutors may be viewed as a communal achievement. Finally, these narrative constructions
appear to have a variety of functions over and above rendering the individuals actions sensible
both to him/herself and others. For example, narratives may reinforce or alter others selfperceptions. Being a worthy friend, for instance, must be supported in the narratives of ones
friends. Other functions of the narrative will be treated later.
Individuals will typically possess a variety of narrative potentials, which may variously be
deployed as occasions permit or invite. Narrative presentations may thus be used to signal a future
set of actions, to indicate positive potentials, to invite others nurturance, to generate a sense of
drama, and so on. In each case, these social functions are prepared by the accumulated lore of
cultural history.
Theoretical analysis of narrative lines
A certain number of our analyses have been concerned with the way in which narratives structure
individual lives. However, to carry out such an analysis a preliminary elucidation of the major
narrative forms available within the culture is required. A narrative that is nonsensical to its
listeners is a failure. In order to be intelligible a narrative must conform to certain rules of what
constitutes a reasonable story within the culture. As our analyses have shown, basic to a good story
in Western culture is the establishment of a valued endpoint or goal toward which the action in the
story is directed. For example, stories about winning the lottery, being promoted or getting married
might all be seen as having positively valued endpoints, while those about breaking an arm, losing
a bet or separating from a friend may all be seen as having negatively valued endpoints. A coherent
narrative line is achieved by selecting and ordering events around this endpoint. Events not
288 Mary M. Gergen
influencing the course of action related to the endpoint are usually not included in the well-made
story.
Events within the narrative abide by culturally prescribed notions of order. Often they are
arranged chronologically. If events are randomly described, the narrative may not be seen as
credible (Lippman, 1986). Events in the story are also described evaluatively with respect to the
goal or endpoint. Events that lead toward the achievement of a valued endpoint are positively
evaluated, and those that lead away from it are negatively evaluated. Other subtle rules of what
constitutes a realistic narrative have been elaborated but are unnecessary to our present purposes.
Yet these various ingredients only tell us what is required for an account to possess a storied
sense. Still required is a framework for describing how these various components are organized to
yield the range of plots which we today consider sensible or intelligible. For example, if someone
were to describe her life story as one in which each positive event was followed by a negative
event, and vice versa, the narrative would be viewed with suspicion by contemporary listeners.
Such a story would seem unreasonable by current social standards, as would a story in which one
wonderful success followed unceasingly upon the preceding one, or one catastrophe followed upon
another without end. How then, are we to characterize the major narrative forms derived from our
cultural heritage?
Some help in answering this question is furnished by traditional accounts of emplotment. The
early Greeks formulated the basic difference between the tragedy and the comedy. Tragedy
involved the sudden demise of a noble person through the elaboration of some tragic flaw that this
person possessed. Oedipus and Antigone were both tragic characters in the Greek sense. Comedy
was often about less socially elevated persons. After confrontation by the characters with various
mishaps, the plot was resolved with a happy ending. Aristophanes play, Lysistrata, is exemplary.
While the terms tragedy and comedy remain vital today, the concepts have also undergone shifts
in meaning. Today we tend to think of tragedy as involving a sad ending, regardless of how it has
290 Mary M. Gergen
occurred, and comedy as being humorous, usually with a happy ending. The social class origins of
the characters is also less relevant in contemporary drama but not entirely erased, as tragic
characters are generally more respectable than comic ones.
Northrup Frye (1957) has provided a more elaborate theoretical summary of narrative forms. He
describes four basic forms of narrative related to the cyclical changes of the seasons. Spring is
represented by the comedy. Here we find a challenge or threat which is overcome to yield social
harmony and a happy ending, as in the Greek formulation. Summer is the season of the romance,
defined as a series of episodes within which a major protagonist is faced with numerous challenges
and threats. These obstacles are overcome, and the protagonist emerges victorious. The autumn
dramatic form is the tragedy. The happy days of summer are overturned, as the approach of winter
and the death of living things is forewarned. The tragic form is similar to the ancient Greek one.
Winter is the season of satire. Beyond hope, these narratives are the representations of unrealized
expectations and dreams.
From the present perspective, the traditional typologies of narrative form seem unsystematic.
While they do pinpoint traditions of importance, the types have either little relationship with each
other or, in the case of Frye, the relationship through seasonal variations seems more poetic than
analytic. In the earlier analysis, narrative construction was said to require the establishment of an
endpoint or a goal, with events selected so as to make the goal more or less probable. In this
context it is possible to view each event in a narrative as moving through a two-dimensional
evaluative space. For example, as one succeeds in approaching the valued goal over time, the story
becomes more positive; as one goes through a series of negatively valued steps toward a negative
endpoint, the story moves in a negative direction. All narrative plots may be converted from a story
form to a linear form with respect to their evaluative shifts over time. These linear forms are called
story lines. Examples of these are shown in a later section.
By conceptualizing narratives in this way, one can begin to see variations in narrative form, from
the rudimentary to the more complex. On the simple level, consider the stability narrative. This
narrative is characterized by an unchanging story line with respect to the evaluative dimension. As
depicted in figure 7.1, the stability narrative (A, in the figure) might be, for example, the story of a
highly successful business trip where everything was perfect from start to finish. The reverse
narrative, in which nothing went right, would be represented by a negative stability narrative N2).
292 Mary M. Gergen
The stability narratives may be contrasted with two other simple narratives. Within these, the
narrative may be structured in such a way that things get continuously better or worse over time.
When the events become increasingly positive over time, the narrative may be called a progressive
narrative. When they are increasingly negative over time, it may be called a regressive narrative
(see figure 7.2). The individual who describes a rise from mailroom clerk to president of the
corporation, with no detours or set-backs, would be telling a progressive narrative. Conversely, the
once wealthy heiress
+
Time ------------------
Figure 7.1 Stability narratives
Figure 7.2 (a) Progressive and (b) regressive narrative forms
who describes a steady procession of events leading to the gutter is constructing a regressive
narrative.
These three prototypes - stability, progressive and regressive - exhaust the basic vocabulary of
possible narrative projections over time. They are the rudimentary forms from which more complex
variations are constructed. The classic forms described above, such as the Greek tragedy and
comedy, are derived from these three simple forms. The tragedy is the story of the rapid downfall
of one who holds a high place in society. A positively evaluated stability narrative is followed by a
rapidly descending regressive narrative. The comedy begins with a regressive narrative followed by
a progressive one. That is, life events become problematic and then, through a succession of events,
happiness is finally achieved. When a progressive narrative is followed by a positive stability
narrative, the well-known fairytale ending of happily-ever-after is obtained. The romance, usually
the tale of a heroic character who overcomes obstacles, may be diagrammed as a series of
regressive and progressive story lines. The satire, we may note, involves a particular style of
depiction emphasizing ridicule or parody and is not, in fact, another type of narrative form. Figure
7.3 illustrates the narrative forms of tragedy, comedy, happily-ever-after and romance.
We may also use these graphic representations to speak to issues of dramatic tension. The way in
which many people characterize their lives is flat and unexciting; others can create a sense of
drama in their lives, even when events in themselves may seem to be of little intrinsic interest. The
dramatic tension that shapes these experiences of life may be attributed to the type of story line
294 Mary M. Gergen
followed within the self-narratives. In particular,
Figure 7.3 (a) tragedy, (b) comedy, (c) happily-ever-after and (d) romantic saga
drama is first injected into a story line by increasing the incline of the slope in either the upward or
downward direction. The steeper the incline of the story line, the more sharply events change in
their evaluation. This shift is, in part, responsible for increased dramatic impact. Thus, within
Western culture, the stability narratives would be the least dramatic and perhaps most boring of
narrative forms. The tragedy is highly compelling because of the radical shift as the regressive
narrative unfolds.
A second contribution to dramatic impact can be traced to the direction of the story line. In
particular, alterations in the direction of the slope line are important ingredients of dramatic tension.
In this sense the comedy has dramatic power because the downward trend of the regressive
narrative is, at a critical point, abruptly shifted to a progressive narrative. This point also creates a
break in the emotional tension generated by the previous downward slope of the line. During the
final progressive phase emotional satisfaction can be achieved, while in the tragedy this process
must be delayed until the narrative is ended. Thus the comedy is most immediately satisfying and
the happily-ever-after form is most soothing - an apt form for a bedtime story.
By asking questions about these various aspects of a narrative, the researcher may develop
several ways to extend psychological inquiry. This chapter will describe one effort to explore self narratives that may be fruitful for others as well.
296 Mary M. Gergen

Empirical example
Most of the research exploring self-concept and self-disclosure in social, personality and clinical
psychology rests on the assumption that, under the proper circumstances, self-narratives are stable
and true. People are presumed to have the capacity to reveal valid biographies to others. From the
social constructionist perspective, this basic assumption is questionable. A recent critique of
traditional personality psychology also supports this social constructionist perspective (Potter and
Wetherell, 1987). Rather, people are not capable of revealing their true selves or their actual life
stories because such things do not exist. People are only able to construe their lives within the
confines of linguistic and social conventions. These constructions will depend on their particular
situation at the moment. The social constructionist position challenges established wisdom from a
theoretical perspective. Thus, the data one gathers when collecting self-narratives are viewed not as
kernels of truth about a persons life, but as temporary constructions of what seems most
appropriate from the perspective of the narrator at that time. (This is not to say that the same
narrative may never again be repeated, but merely that it need not be considered foundational.)
What is constrained for the narrator is the form that a narrative must take. As
illustrated in the introduction to this chapter, narratives are composed of a limited
vocabulary of forms. Story tellers may individuate their narratives by being more or
less dramatic about various events, but even these shifts in evaluative dimensions and
alterations in story line are regulated by social conventions. In the study to be reported
here, college students were asked to create narrative expositions of their lives. We were
interested in finding out whether there were any trends in how this group of people in
late adolescence would tell the story of their lives, and also to see how consistent they
were in their descriptions of their overall lives, and of various subelements of their
lives. The study was also designed to experiment with a story line methodology, which
will be illustrated below
.Narrative structures in social explanation 103

Acquisition of the data
A major purpose of the study was to explore the social constructionist

view of self-narratives, the use of narrative forms among late
adolescents, and in addition the utility of a protocol developed around
the story line method of constructing a self-narrative. To these ends, a
group of 29 undergraduate students, ages 18-21, was asked to respond
to a questionnaire entitled Constructing the past. The introduction to
the questionnaire stated that the study was concerned with the
construction of personal history - how we look back (and ahead) on
our lives and make sense of events in relationship to each other. In
section I of the protocol respondents were asked to recall the happiest
times in their lives, and their most unhappy times. This section of the
questionnaire was included so that the students would be put in a frame
of mind to think over their pasts, and to recall the high and low points
of this time. It was preparatory to section II of the survey, in which
subjects drew a story line to describe their feelings of generalized
well-being from birth to age 20+, followed by a brief description of
the highest and lowest points. An example of this appears in figure 7.4.
In section III respondents were asked to create additional story lines

reflecting their relationships with their mothers, fathers and with their
academic work, along with descriptions of the highest and lowest
points in each of these domains. The latter areas were selected because
they seemed to be aspects of students lives that would be central to
their overall feelings of well-being. Thus, comparisons and contrasts
could be made between the various story lines for each of the
respondents. In section IV of the questionnaire, respondents plotted
their generalized feelings of well-being on two graphs, one for the day
of the study and the other for their future life, from age 20 to beyond
80. An example of an anticipated future graph appears in figure 7.5.
The protocols designed with the story line method proved to be very
useful in summarizing the feelings of the sample over time. Previous
research using free-form responses to open-ended questions had
proved difficult to transform into narrative forms because of the
subjective nature of the data, and the complexity of trying to graph fhe
narratives on to a two-dimensional space (Gergen, 1980). In this study
the respondents had accomplished this task for the researchers.
Interpreting the data
There are several possibilities for analysing these data. Given that the number of
respondents was small, the major attempt was to assess what types of narrative form
seemed most common to
II. Trajectory of the past. Below you will find a grid on which you can begin with the earliest period you can remember and
draw a continuous line to the present. This line would indicate, for each period, how you now remember your feelings
of generalized well-being. The more positive the feelings the more upward the displacement of the 'life-line'; the more
negative the feelings the more downward the displacement.
Brief description of highest point:

I didn't really begin to like myself until I was about 15 years old. I was very shy until I entered a boarding school and
became more active with other people.
Brief description of lowest point:

I was very shy. Much anger because I didn't have a 'live-in' father. I blamed father for my being over-weight., I was very
withdrawn and I had few friends my age with whom I was close.
Figure 7.4 Trajectory of the past (protocol section II), and descriptions of high
and low parts
302 Mary M. Gergen

IV. Anticipated future. Can you now indicate briefly what you think future years may hold in terms of feelings of generalized
well-being. Can you project your 'life-line' into the future?
Figure 7.5 Future trajectory of life graph
people in the study. It was also possible to explore multiplicities of narrative forms. In addition,
attention was given to questions of how well the protocol was designed for the study and where
there were problems in its construction.
Many possible hypotheses were feasible when contemplating the narrative form of these story
lines. One might expect that these young adults would portray themselves as part of a happily-everafter story, having reached a happy plateau from which they will live out their lives. Or one might
conjecture that the participants would characterize themselves as living out a romantic saga,
overcoming one obstacle after another. Perhaps, more pessimistically, these students might see
themselves as actors in a tragic drama, on the edge of a chasm, heading into a nuclear calamity. To
explore this question an attempt was made to derive the average self-narrative from the story lines
of their past lives. To do this, each point at
Age (years)
Figure 7.6 Composite self-narrative for the sample: generalized well-being, birth to 20+
five-year intervals on the story line labelled generalized feeling of well-being for their lives since
birth was converted to a numerical value for each respondent. The displacement of each story line
from the neutral midpoint was computed at each five-year interval. The resultant story line was the
composite representation of all the story lines for the sample of 29 respondents. In figure 7.6 this
story line is reproduced. As is evident, the general narrative form disclosed by this analysis was
none of those mentioned above. Rather, for the first 20 years, the most common story line
approximates the comedy form of narrative. The basic story as revealed through this assessment is
that life was quite good as a child, but problems occurred during adolescence, which were resolved
during the college years. The overall form averaged out to represent the typical myth of Americans
that childhood is idyllic, adolescence is a time of turmoil, and college days are halcyon.
304 Mary M. Gergen
In this light it is interesting to take account of the highest and lowest points related to each of the
narrative forms. As these accounts revealed, these events were highly diverse. High points included
falling in love, travelling abroad, and finding Christ. Low points included parents divorcing,
moving, and death of a pet. The interesting point here is that while these young adults tended to
agree that adolescence was a miserable period, the content of the misery varied greatly. This
suggests that the participants seemed to be more influenced by the narrative form that prescribes
how life must have been for a young adult than by any specific set of events. Their facts served to
justify the preselected narrative form.
While the average students narrative fitted the comedy prototype, only nine of the 29
participants fully adopted the popular form. Fourteen of the respondents story lines deviated from
the composite in that they had additional hills and valleys in their narratives. The remaining six
students used simpler progressive or regressive narratives, with story lines either rising or falling
steadily from early childhood. In the protocol reproduced in figure 7.4, you will note that the period
from birth to early teen years is described as a low stability narrative followed by a progressive
narrative. As the data indicate, at least two-thirds of the respondents had some variation on the
comedy form. Because there are many versions of the narrative form appropriate to Western
culture, one may hypothesize that young adults have within their narrative repertoire many versions
of their childhoods, which they may be able to produce if the circumstances are appropriate. This
research and similar studies with young adults have indicated that projections of generalized wellbeing into the future show a much more consistent happily-ever-after prototype after age 25, with a
regressive downturn at lifes end.
As was mentioned above, the narrative form which a person uses for telling a story is selected
for specific occasions or functions. Multiple narrative forms characterize an individuals story
telling repertoire. We might conjecture, for example, that a narrative that is shaped as a romance,
with the narrator overcoming obstacles to achieve a goal, is an attempt to present the narrator as a
hero who lives in a world of treachery or danger. The listeners are expected to be enthralled and
admiring of such a protagonist. The tragedy may be designed to elicit sympathy, and the comedy a
companionate spirit of solidarity and harmony.
306 Mary M. Gergen
Different stimuli may bring to the fore different narrative forms. This was discovered in looking
over the story lines for the sample protocol for relations with mother, father and school work (not
shown here). Here we found many divergent slopes in the story lines. The miserable childhood
which is depicted in the feelings of generalized well-being in the past is not found in the story
lines related to mother or school work. The narrative line for mother seems to be a form of comedy
quite unrelated in its temporal pattern to the story line for generalized well-being, except in the
resolution phase after age 15. Life with mother was fairly positive until the teenage years, and then
after a strong negative period became even better than ever. This form echoes the most popular
narrative form used by the sample group. The school work narrative line shows two positive
stability narratives, separated by a dramatic reversal around age 10. This is a different pattern from
the other three story lines.
Only in the relations with father story line is there a parallel with the generalized well-being
line. Yet, when the respondent described his or her most unhappy times in the first section of the
questionnaire, father is not mentioned at all. Later when describing the lowest points on the
generalized well-being story line, the respondent mentions several factors, including being shy and
without friends, in addition to not having a live-in father (figure 7.4). The respondent also
implicates his or her father in the description of low points, as he or she writes I blamed father for
my being overweight. In this section of the protocol, the story of childhood seems strongly focused
on father. The father is made responsible for much childhood misery, including the respondents
weight. The point to be emphasized here, however, is that this rendition of childhood is simply one
version among many possible. At another time and in another place the construction might be
reorganized into some other form.
Within the studies in which I have been involved, the major thrust of the analyses has been to
look at the narrative forms that seem preferred by the group, and to look at the diversity of forms
used. It is clear from this sample that a model form was utilized, and that diversity could be found
not only between group members, but also within each protocol.
308 Mary M. Gergen
It would seem that many other uses might be made of this story line methodology, depending on
the purposes of the researcher. For example, in the case of a clinician looking for means of helping
a client, the protocol might prove a useful starting point for assessing areas of life in which the
person has described himself or herself as unhappy. In addition certain relationships might appear
to be more problematic than others. In the protocol shown in figure 7.4 and 7.5 the narrative points
to problems with father more than mother. The act of construing the narrative might also be looked
upon as a therapeutic exercise. As Donald Spence (1982) has suggested, the central mission of the
therapist and the patient is to construct a narrative of the patients life story. The story line might be
one form of endeavour for the illustrating of this agreement.
Another analysis of the individual protocol might take account of the number of reversals in the
slope of the story lines and the steepness of the slopes. Respondents who deviated greatly from the
average profile (figure 7.6) might be expected to perceive their lives as more dramatic and intense
than the average person. In looking at the protocol presented herein (figures 7.4 and 7.5), it seems
clear that the respondent views his or her past life as changing rather dramatically in generalized
well-being around age 13, for example, and as becoming very stable in the 20s, with only a slight
decrement after age 30.
Further research with different groups of respondents of various ages, backgrounds and interests
should reveal some differences in overall self-narratives. One can envision research in which
various subgroups of youth, for example, might reveal group differences in expectations
concerning life satisfactions. For example, if a group of adolescents consistently perceives that
well-being peaks at 18, and is downhill thereafter, the notion of delayed gratification might be very
repugnant to them.
310 Mary M. Gergen
Within various social groups the cultural clocks - that is, the times at which people are
expected to do certain things such as marry and have children - should help formulate certain
parameters for how self-narratives are shaped (Neugarten and Hagstad, 1976). Research among
pre-menopausal women indicated, for example, that they had begun to formulate expectations of
how their lives would be after menopause, and what they would expect to look like, feel and do for
the next 20 years (Gergen, in press). These attitudes are based to a great degree on what the popular
culture portrays about the older woman. With an elderly population the narrative form also seems
to be influenced by the cultures beliefs about the aged. The tendency to accept a declining level of
involvement and activity seemed to echo the notion of disengagement from society that often is
expected of elderly people (Gergen, 1980; Gergen and Gergen, 1983).
The last portion of the protocol explored two other narrative dimensions: the feelings of
generalized well-being for the day, and feelings of generalized well-being for the future life of the
individual. Of course, story lines could also be created for other periods or for other aspects as well.
The feelings of daily well-being could easily be analysed as to the type of narrative form it
portrayed. The analysis could be useful in giving one a sense of the emotional context in which the
other narratives were being constructed. The daily story line also gives a baseline for how dramatic
the person may be as a story teller. A very changeable line for one day might indicate instability in
other narrative lines. The story line that projects into the distant future can be analysed in terms of
preferred cultural forms, of individual variation and of mental health and well-being. One protocol
contained a story line that ended abruptly at 30, with the word dead scrawled next to it. When
asked about this story line later, the narrator said that he had had a premonition when he was
younger that he would die in a car crash at that time, and so he was planning his life around this
belief. Of course we cannot be certain from this story line or his explanation just what
interpretation to give to this persons expectations for the future. People have multiple narrative
forms about what happens in adulthood, and they are capable of drawing more than one story line,
depending on the situation. For example, college students rarely include marital discord or divorce
in their future life story lines, but if they were to create two or three different versions of their
future story lines, some would be likely to include these rather normal but unpleasant episodes.
312 Mary M. Gergen

Within traditional paradigms of psychology, the study of personality and the self stands in stark
contrast to the narrative approach. Within what I will call the psychometric approach to personality,
the assumption is made that people have fairly stable and predictable personalities and behavioural
patterns. It is also assumed that through quantitative methods the dimensions of the personality and
the self-concept can be obtained. The research is usually focused within a narrow time span, and
results in data that are ahistorical and decontextualized numerical representations of the subjects.
The resultant scores are then often correlated with other variables to predict behavioural tendencies.
This psychometric approach relies on a natural science model of psychology, in which the basic
goals of the science are the creation of universal laws and the prediction and control of people.
In contrast, the social constructionist approach assumes that peoples self-narratives are social
products. They are not seen as enduring, stable fixtures of a pre-existing nature, but as temporary
constructions that are shaped by such important factors as literary conventions, social norms, the
context of the narration and self-determined social goals. Self-narratives are not viewed in the same
light as a traditional psychologist might view a personality profile. Narratives are not seen as
representations of the real self underlying the story. The narrativist gives up the certainty of the
psychometrician by taking this perspective. While this seems to be a disadvantage, the failure of the
psychometrics approach to find generalizable laws regarding the prediction of behaviour from
personality traits makes the loss less significant. At base, the change from a natural science
foundation to a human sciences approach is a matter of intellectual preference and conviction. But
once inside the latter paradigm, the use of a narratives methodology becomes more appropriate for
the study of lives.
There are potentially many ways in which to explore the narratives of lives. One of the most
common approaches is to ask people to give written or oral biographies of their lives, or parts of
them. A derivation of this approach is to ask for narratives that cover certain topics or periods. For
314 Mary M. Gergen
example, in a study of the social explanations of the elderly (Gergen and Gergen, in press)
respondents were asked to state why they were more or less involved in some activity than they
used to be. In terms of analysis, written or oral narratives have as advantages being rich in detail,
individualized, flexible, revealing of personal judgements of importance, targeted to sensitive areas
that might be avoided otherwise, and potentially complex in structure. As disadvantages,
biographies can vary greatly in their elaboration. The level of motivation, of comprehension and of
verbal skills in the respondent makes a great deal of difference in how the biographies are formed.
In oral biographies, especially, respondents can lose their train of thought. It is also somewhat
problematic for the analyst to know how to assess the biography. It is difficult, for example, to
know whether an unhappy childhood story is seen by the respondent as unhappier than an unhappy
period in adolescence. The researcher must make many judgements concerning the relationships
among various episodes in a biographic protocol. The biography form does not easily produce such
comparisons.
This chapter has emphasized the use of the graphic story line to reveal the evaluative dimension
of ones life and relationships with others. A cardinal advantage of this approach is that subjective
appraisals of life events can be converted to dimensions that are quantifiable. Thus, as was
demonstrated with the lives of this college sample, narratives can be compared and combined.
While not designed to be a precise mathematical instrument, story lines can also reveal perceived
peaks and valleys in peoples lives, and their relative impact on the narrator. For example, one can
tell at a glance whether parents divorcing is now viewed as more or less painful than a bad love
affair. Story lines are also useful for conducting dialogues with narrators about their lives. For
therapists and counsellors, for example, the interpretation of the story lines by their authors would
be potentially a very useful guide. The inclusion of questions within the protocol that ask for
specific events that marked the high and low points helps to draw attention to certain critical events
in the persons life, as they are now perceived.
The story line is also relatively quick and easy to fill out, and may be perceived by the
respondent as an interesting and creative mode of self-expression. It is not as dependent on
excellent verbal skills as biographies are. Also, it is not as threatening to the participant as filling
out a questionnaire that asks about intimate details of ones sexual life, for example. Narrators have
a sense of control over how they will be perceived by others. Of course, some psychologists may
316 Mary M. Gergen
perceive this as a flaw in the instrument.
As for limitations, the story line may encourage too much glossing over of events and the details
of life. A year could be filled with extreme highs and lows, and these might be averaged out to look
like a mediocre year. Some years may also be much more important than others to the narrator. The
advantages of the story line in terms of administration and utilization may be also at the base of the
disadvantages of brevity and compression. The story line may also be limited because some
respondents might find the task unusual and hard to understand. Also narrators might create story
lines without reflection.
While each type of methodology in studying narratives has limitations as well as

advantages, the challenge of the approach is an interesting and exciting one for
increasing numbers of psychologists. New innovations are forthcoming, and
modifications in existing forms may improve their capacities for eliciting life stories.
Contributions from other sources, such as those in other chapters of this volume,
should increase the means for exploring the narratives of lives
.PART III
ACCOUNTS IN CONTEXT
8
Accounts on trial: oral arguments in

traffic court
Michael J. Cody and Margaret L. McLaughlin
We propose in this chapter to examine the notion of accounting, which

we take to be the way in which failure events are managed in social
interaction. Early treatments of accounts tended to be descriptive and
classificatory, focusing on the development of a taxonomy of forms
which accounts take when allegations of untoward behaviour arise in
everyday social encounters. For example, Scott and Lyman (1968)
offered an extensive treatment of two kinds of accounts, excuses and
justifications, and paid particular attention to their wide variation in
substantive form. Excuses and justifications were distinguished on the
basis of their treatment of two issues: (1) whether or not a putative
offence had indeed occurred, and (2) whether or not the person called
upon to give an account was in fact responsible for its occurrence. An
excuse was held to take the form I committed the offence, but I was
not responsible, and a justification the form I am responsible, but I
deny that an offence was committed. Thus a person accused of failing
to appear for an appointment in a timely fashion might excuse himself
by claiming that he had been the victim of a flat tyre, or justify the
tardiness by claiming that the one who had been kept waiting had not
in fact been specific about the time they were to meet.
Oral arguments in traffic court 321
More recent work in the taxonomic vein, such as that by Schonbach

(1980, 1985, 1986), attempts to extend the catalogue of accounting
types to forms other than excuses and justifications. Schonbach (1980)
added the categories concession and refusal to deal with, respectively,
admissions and denials of guilt (responsibility). Concessions in their
pure form are confessional; that is, the offender admits to having been
late, having been drunk, having failed to take appropriate action, and
so forth. Such admissions of guilt may or may not be accompanied by
apologies, expressions of remorse, offers of restitution and the like.
Refusals as a response to an allegation of untoward action may entail
the flat denial by the one reproached that she committed the act in
question, or may take the form of a refusal to acknowledge the others
right to reproach. Schonbach has recently (1985) revised and expanded
his taxonomy of accounts, although he retains the four cardinal
categories of his earlier work: concessions, excuses, justifications and
322 Michael J. Cody and Margaret L. McLaughlin
refusals. He revises the received definition of excuse and incorporates

the notion of theme. In the 1985 paper, Schonbach defines excuse as
any account which admits the occurrence of a failure event and some
involvement of the actor in it, but pleads for mitigation in judgment on
the basis of various arguments including, but not limited to, claims of
impairment and hence reduced causal responsibility (p.10).
Justification is defined conventionally as a strategy by which the actor
accepts causal responsibility for the event in question, but asserts that it
was legitimate or at least permissible under the given circumstances
(Schonbach, 1985:11). Schonbach also introduces into the ongoing
discussion of accounts the notion of themes, for example appeal to
loyalties, influence of powerful agents or situational constraints,
which can be exploited as a resource for constructing more than one
kind of account. Thus, for example, the theme provocation by the
accuser could be used to fashion either an excuse (I was so upset by
what you said that I wasnt myself when I hit you) or a justification
(You deserved it after what you said to me).
Much of the current work on accounts is concerned with the

functional utility of accounting to the individual; that is, scholars have
begun to examine how (and/or if) excuses and justifications work.
Some researchers have been interested primarily in the role of
accounts, particularly excuses, in the retrospective (and sometimes
anticipatory) redefinition of events with dispreferred outcomes.
Backman (1985) has examined the relationships among self-identity,
self-presentation and moral decision making: specifically, the role of
the anticipatory construction and rehearsal of accounts in defensively
reinterpreting (Rest, 1984) a morally problematic situation so as to
make the circumstances surrounding the actors choice appear to be
mitigating. The actors search is for situational definitions that will
blunt the effects of anticipated censure from the self or others, either
by reducing the strength of the rules operating in the situation or by
providing a basis for accounts or explanations (Backman, 1985: 268).
Snyder and his colleagues have also been interested in the utility
and efficacy of accounts, particularly excuses. Their attention has been
on the process of making excuses; Snyder and Higgins (1986) define
this process as one of shifting causal attributions for negative personal
outcomes from sources that are relatively more central to the persons
sense of self to sources that are relatively less central (p. 2). Snyder
and Higgins focus primarily on the role of excuses in the maintenance
of self-esteem. After an extensive survey of the relevant literature, their
assessment of the evidence is that the availability of excuses (from
whatever source, including ones own resources) can significantly
reduce the effects of negative outcomes on self-esteem. For example,
being able to claim that a task was too difficult or that ones friends
would also perform poorly could serve to restore self-esteem, or
prevent losses resulting from an intellectual failure. Snyder and
Higgins (1986: 9) maintain that it is the so-called EVS pattern of
excuse-making, that is, attributing negative outcomes to external
(extra-individual), unstable (variable) and specific factors rather than
internal, stable and global ones (the ISG pattern), which is facilitative.
The EVS pattern has been found to be associated with improved
outcomes in psychotherapy, better long-term health, and superior
athletic performance. Similarly, Tedeschi and Norman (1985) proposed
that accounts can be viewed as defensive mechanisms which are used
by the individual to maintain an appropriate public image.
In our own research (Cody and McLaughlin, 1985; McLaughlin,

Cody and OHair, 1983; McLaughlin, Cody and Rosenstein, 1983) we
have been interested in how accounts work in an interactional context;
that is, how parties to a failure event work collaboratively to effect
repair. We have now conducted four studies on accounting episodes.
Our first study (McLaughlin, Cody and OHair, 1983) was a survey in
which 278 undergraduate subjects provided retrospective reports on a
failure event. Our second study involved the analysis of accounting
episodes in 50 half-hour recorded conversations between strangers
(McLaughlin, Cody and Rosenstein, 1983). Our third study was again
a survey of 156 drivers in Texas, New Mexico and Oklahoma, who
were asked to recall instances in which they had been stopped by a
police officer for a traffic offence. In our fourth study, which we report
here, we observed 375 cases of account giving in four traffic courts in
the greater Los Angeles area. Our work has led us to two major
conclusions about the way in which accounts unfold in social
interaction.
Our first conclusion is that accounting sequences have a canonical

form (Cody and McLaughlin, 1985; Rentier, 1978; Schonbach, 1980,
1985). At a minimum, the sequences consist of three slots which may
or may not all be explicitly filled. Firstly, there is a slot for a reproach,
in which a party taking offence communicates to an actor by some
means (possibly non-verbal), or the actor perceives through his or her
own guilt, anxiety or social astuteness, that something is amiss. The
second slot in the canonical sequence is for an account. The right or
obligation to talk in this slot is assigned to the party reproached, whose
job it is to produce a reply or remedy conditionally relevant (Schegloff,
1972) to the form of the reproach employed in the particular context.
(One may of course provide an account for another just as one may
answer her or his own question, and thus effect closure as far as the
requirements of functional coherence are concerned, but this is clearly
not the normative case.) The third slot in the accounting sequence calls
for an evaluation, in which the initiator of the reproach reports his or
her assessment of the offending partys account. This canonical
sequence, variations of which have been reported elsewhere (the
misapprehension sequence by Jefferson, 1972; the corrective
interchange by Goffman, 1971; the account episode by Schonbach,
1980, 1985, 1986; and the remedial episode by Morris, 1985), may
incorporate additional steps such as a precipitating failure event or an
acknowledgement of the evaluation.
Our second conclusion is that the steps or slots in the canonical

form of the accounting sequence relate to one another in some
predictable ways, such that certain types of reproaches are more likely
to induce certain types of accounts, and similarly with accounts and
evaluations. A construct we have invoked in describing this is the
aggravation-mitigation continuum (Labov and Fanshel, 1977).
Reproaches, accounts and evaluations can be arrayed along a
continuum such that at one end are clustered forms which will tend to
exacerbate the tension already generated by the failure event, and at the
other are forms which will tend to defuse or ameliorate the situation.
For example, a reproach in which a truculent attitude on the part of the
offending party is projected (Just try to deny that you were flirting
with that woman!) we would fully expect to provoke an account in
kind (I dont have to deny it; were not married and I can flirt with
whomever I please!). We have found in general that aggravating
antecedents lead to aggravating replies; such a relationship could be
attributed to reciprocity norms, topical relevance constraints, or the
interlocking preconditions (Cody and McLaughlin, 1985; Searle,
1975) of the illocutionary acts (challenge/reply) embedded in the
utterances. Similarly, we have found that more mitigating forms of
reproaches and accounts are likely to be met in kind. For example, in
our study of drivers in south-west USA (Cody and McLaughlin, 1985),
more mitigating accounts were likely to lead to honouring (being let
off without a ticket). In our study of conversations between strangers
(McLaughlin, Cody and Rosenstein, 1983), reproachers honoured
accounts or withdrew their reproaches more frequently when the
accounts took the more mitigating forms of concession or excuse. This
evidence in favour of matching response levels has been stronger for
aggravating reproaches and accounts than it has been for the more
mitigating ones (see in particular Cody and McLaughlin, 1983).
The various forms in which accounts appear may be ordered with

some consistency with respect to the aggravation-mitigation continuum
(and thus with respect to the probability of being honoured). However,
what is mitigating in an interpersonal context may be aggravating in a
legal one. Our studies indicate that in dealing with failure events in
interpersonal contexts, concessions (admissions of guilt, apologies,
offers of restitution) are the most mitigating, followed by excuses,
justifications and refusals. However, in our study of drivers stopped for
traffic offences we found that admissions of guilt were likely to have
an aggravating effect (lead to ticketing). The critical issue here is that
the patrolman does not enter into the accounting episode as an
individual but rather as a representative of the state (city, county), and
his obligation toward a guilt admission is qualitatively different from a
friends or lovers (Cody and McLaughlin, 1985).
Our fourth study, which we use as our empirical example, is (like
our third) an examination of account giving in a situation in which the

reproacher (or recipient of the account) is acting as such in an official
capacity. Our expectation was that in the courtroom context, excuses
would be the most mitigating (least likely to result in penalties),
followed by justifications, refusals and concessions (guilt admissions).
However, Coleman (1976), in his study of courtroom control and
grievance accounts, brought up two points which might argue against
the predicted order. Firstly, the refusal strategy of some defendants is to
bring documents or develop logical proofs that they could not have
committed the act with which they had been charged; for example, a
defendant might present to the court a bill showing that his car had
been in a repair shop all day on the day he had been ticketed for a
parking violation. Such an account qualifies as a refusal under our
category scheme, but would probably not be perceived as aggravating
and might well result in honouring. A second issue raised by Coleman
is the need of the presiding judge to keep things moving along; the
requirements for the rapid processing of cases might dispose the judge
to look with more favour on the defendant who utters a quick no
contest than one who insists on delivering a lengthy excuse.
Special considerations aside, however, we anticipated that the

proposed ordering of account forms with respect to the likelihood of
penalty would be robust. In as much as the mitigation level of an
utterance or series of utterances is carried by non-verbal as well as
verbal factors, we incorporated into our study a number of variables
related to the accounters apparent level of remorse, degree of anxiety,
amount of anger and so forth.
Empirical example: courtroom observation
In this example of analysing accounts, we present in some detail a

study of accounts given in traffic court. Fifty-six undergraduates at the
University of Southern California worked in pairs as observers at one
of four traffic courts in the greater Los Angeles area. Observers had
trained to satisfactory reliability levels (median r > 0.80) on taped
versions of the half-hour television programme. The Peoples Court.

Observers recorded for each of 375 cases the location, date and time of
the session, and the name of the presiding judge; the sex and
approximate age of the defendant; a brief description of the statements
made in the courtroom by the defendant, arresting officer and judge;
the severity of the offence (ranging from drunk driving and excessive
speeding to jaywalking and fix-it tickets (to repair light bulbs etc);
and the nature of the outcome, namely no penalty (dismissal), partial
penalty (traffic school, dismissal of some but not all counts,
community service or reduction of the normal fines) or full penalty
(guilty as charged). Observers also rated characteristics of the
defendants self-presentation on nine-step bipolar scales: confidence
(fast speech rate, no or few silent/filled pauses, air of confidence);
concreteness of account (contained details of actions, motives,
decisions and so on); extroversion and dynamism (animation, high
energy level, erect posture, forceful gestures); anxiety (nervous
gesturing, pacing, speech errors, adaptors, random body movement);
apologetic voice quality; non-verbal expression of regret (dejected
body position, lowered head, sad expression and so on); politeness;
and anger.
Coding of defendants accounts
All observer summaries were read by three coders, each of whom had
read Coleman (1976), Cody and McLaughlin (1985) and Scott and
Lyman (1968). Preliminary examination of the observers summaries
of the defendants courtroom statements by two coders (who
conducted the first reading of the essays) indicated that for the most
part the observation categories emerging in previous investigations
(concession, excuse, justification and refusal) were appropriate for the
classification of accounts in traffic court, with one significant
exception. The refusal category obtained in the earlier studies
(McLaughlin, Cody and OHair, 1983; McLaughlin, Cody and
Rosenstein, 1983) was manifested in three very distinct forms in the
defendants courtroom statements. We have called these categories
challenges (n = 21), denies offence (n = 53) and logical proofs (n = 48;
following from Coleman, 1976).
In using a challenge, the defendant rejected the claim that an

offence had taken place on the grounds that the reproacher (arresting
officer) was unfair, unprofessional or inexperienced. In some cases the
defendant was verbally abusive of the police officer and/or aggressive
towards the judge. Examples are as follows:
Defendant: I am a good tax-paying citizen and I should not be getting a
ticket in this speed trap.
Defendant: I might have been going 60, but I was not going anything
like 70. I think the officer added those extra miles on just to make
me more upset.
Defendant: Not guilty, your honour [answering the judges question].
The police officer was going in the other direction. He could not
have known my speed and I was flowing with the traffic. The
officer pulled me over just because I drive a red Porsche and it looks
like its going fast.
Refusals which we classified as denies offence simply involved a flat

assertion that an offence had not occurred. Although the defendant may
have taken issue with the arresting officers characterization of the
action as an offence, there were no assaults on the officers integrity or
motivation. Conversely, there was no true physical proof of the
defendants claim. For example, a man charged with driving on the
sidewalk and soliciting a prostitute made the following response:
Defendant: I was asking directions from a woman at the bus stop and
had no idea she was a prostitute. I was not on the sidewalk [but] had
pulled into the gas station.
In another example:
Defendant: The light may have been yellow, but it was not red when I
made the turn. ... The officers view was obstructed by other cars.
In a very distinctive form of refusal, logical proofs, the defendant

attempted to demonstrate his or her innocence through logical
argument, often bolstered by physical evidence:
Defendant: Well, your honour, the road was mismarked and the car
behind me was motioning me to turn right, across the line [from the
left lane of two lanes heading north]. I brought the pictures of the
area, and I went to the city planner who agreed that the area was
mismarked and under investigation for a change.
Defendant: Your honour, I cant be guilty because they painted the kerb
red after I parked there.
Judge: This sounds like a cartoon!
Defendant: Here are papers from the city showing that the kerb was
painted the day I got ticketed.
Defendant: The cop said the moped was unfit for two people, but I
have the specific manual here that states the moped is designed for
one or two passengers.
The concessions category (n = 74) subsumed those cases in which the

actor admitted guilt or stated that she or he would not contest the
ticket. Concessions could include apologies, offers of restitution and
promises that the behaviour would not recur, and pleading for leniency,
or might simply consist of a response such as no contest or guilty,
your honour to the judges question How do you plead? The
defendants first two utterances in the exchange below are typical of
the category:
Judge: Reckless driving. What is your plea?
Defendant: Guilty as charged.
Judge: Were you under the influence of alcohol at the time of the
arrest?
Defendant: I had two beers at dinner.
Judge: Oh, the old just two beers story.
Defendant: Yes, your honour.
Judge: Well, the charge will stand, but I want you to attend 12 AA
meetings, and [take] a $406 fine [and] 36 months summary
probation.
Defendant: Isnt that a little bit harsh?
Judge: Not for someone who is a threat to other drivers on the road.
Excuses, as was the case in our earlier studies, were the most popular
form of account (n = 134). Defendants attributed their offences to
(among other factors) accident, misinformation, lack of knowledge,
lack of ability, illness and the behaviour of others:
Defendant: There was a car to the left of me, and the pedestrian was on
the west side; the car on my left was ahead of me, therefore
obstructing my view.
Defendant: The car was parked in the no-parking zone because [it] had
a flat tyre.
Defendant: I was driving a new Mercedes that was shipped from
Europe. It doesnt show you speed in m.p.h., it shows it in
kilometres [per hour].
Defendant: I didnt see the sign because it was dull and I didnt have
my glasses on.
Not surprisingly, fewer defendants in court attempted to justify their

behaviour (n = 45). In this variation the accounter admits that he or she
engaged in the behaviour described by the arresting officer, but argues
that the behaviour should be condoned because others were doing it, it
was done for a good reason, and so on:
Defendant: I left my curling iron on, and was rushing to get home to
turn it off before the house burnt (sic) down.
Defendant: I was driving on the 118. I became very tired and almost
fell asleep, so I pulled over and went to sleep. I was aware the ramp
was closed, and parked there for that reason. After my nap, it would
not have been safe for myself and other drivers if I had backed up
on to the freeway, so I felt it was much safer for me to go up and
over to the ramp to re-enter. I was doing what I thought was safest.
Defendant: I was late for work, and everyone else seemed to be going
just as fast as I was, so I didnt think I was speeding.
Defendant: [accused of drunk driving ]: It was my thirtieth wedding
anniversary and we had a party. I only live around the corner.
One of the two coders who had conducted the preliminary reading of
the 375 essays, as well as the first author, coded the essays into the
respective six categories while blind to the assignments made by the
other coder. The agreement was highest for concessions (98 per cent)
and lowest for justifications (83 per cent), and averaged 89 per cent.
The two coders met and reached mutual agreement for the essays on
which they had initially disagreed.
Analysis
The quantitative approach allowed us to compare the relative efficacy

of the different account types. Logical proofs were by far the most
effective means of avoiding the imposition of a penalty, whereas a
concession was virtually certain to result in a penaltys being assessed
(see table 8.1).
Interestingly, there were no significant univariate linear relationships between defendant characteristics and the imposition of a
partial penalty. Twenty-three per cent of the sample received no
penalty, 24% partial penalty and 52% received full penalty (guilty
as charged). The fact that there were no significant correlations with
the partial penalty category reflected the catch-all nature of the
category: poor defendants unable to pay a fine were assigned
community service; defendants (even those with excessive speeding
charges) were granted traffic court (as an alternative to payment) if
they had clean driving records for over a year and had not been to
traffic court in the last three years; and some judges routinely reduced
the fine on excessive parking tickets (one defendant had over $2000 in
unpaid parking tickets) and some speeding tickets. For the rest of the
analyses, we compare no penalty (23 per cent) defendants with
penalized ones (76 per cent). However, the correlations indicate that
full penalties were associated with the severity of the offence, the
vagueness of the defendants account, and a plea of guilty/no contest.
Escaping penalty altogether was positively associated with lesser
severity of the offence, concreteness of the account, defendants
extroversion and the use of logical proofs, and negatively associated
with anger and concession of guilt.
To tease out more subtle aspects of the data, a multiple discriminant

analysis (MDA) was conducted with the six account types as levels of
the dependent variable (groups) and the outcome and defendant
characteristic variates as predictors. A non-stepwise procedure was
carried out over the entire predictor set. The first two of five
discriminant functions were significant. Function 1, on which the only
substantial loading was the dichotomously coded variable no penalty,
served primarily to contrast defendants using logical proofs (group
mean = 1.62) and defendants using concession (group mean = -0.87).
Function 1, with an eigenvalue of 0.50 and a Wilkss lambda of 0.53,
accounted for 69.59 per cent of the between groups variance (chisquare (d.f. = 60) = 225.55, p = 0.00). Function 2, on which the
variable confidence had a large negative loading, and the variable
apologetic a large positive one, served chiefly to contrast users of
challenge (group mean = - 0.75) and denies offence (group mean = 0.49) with the other categories of message type. Function 2, with an
eigenvalue of 0.10 and a Wilkss lambda of 0.81, accounted for 14.79
per cent of the variance (chi-square (d.f. = 44) = 77.09, p = 0.001),
bringing the cumulative variance accounted for to 84.39 per cent. One
may conclude from the MDA that defendants using logical proofs are
Table 8.1 Rank order of account types with respect to imposition of
penalty
* Proportions with common superscripts are significantly different from

each other, p less than 0.05.
likely to escape penalty and those using concession are not, and that
defendants using challenge or denying the offence are likely to be
perceived as confident and unapologetic.

Our earlier work (Cody and McLaughlin, 1985; McLaughlin, Cody

and OHair, 1983) relied on survey data. We have shifted emphasis in
recent years away from survey data in favour of observing behaviour
in natural settings (whenever possible, ethical and legal). Two of many
important reasons will be discussed here. Firstly, observational data are
not likely to be as biased as recalled data. If we have social actors
recall episodes of account giving, some episodes are likely to be
perceived (and/or reinterpreted) in such a way as to enhance or
increase self-esteem (or, at least, the actor may avoid threats to his or
her self-esteem) (see Snyder and Higgins, 1986 for sample research).
One estimate of the type of biased reporting that one can expect was
provided in a recent summary of research on compliance-gaining goals
(Cody, Canary and Smith, in press). A typology of goals was proposed,
based on Rule and Bisanzs work (1987; Rule, Bisanz and Kohn,
1985), including goals such as obtain permission, gain assistance

(which included both favours and selfish requests - those in which the
actor argued that his or her goal should be pursued at the cost of the
targets goal), give advice, change relationships, enforce
obligation etc. To assess frequency of success of different goals, Cody,
Canary and Smith conducted a survey in which respondents were
asked to recall from communication episodes all types of goals they
could remember pursuing, and a diary study (whose respondents kept
diaries over a 12 week period). Some of the differences in methods
were attributable to qualitatively different types of environments (that
is, actors made fewer attempts to obtain permission from parents once
in college; instead, gain assistance goals increased in frequency).
However, two substantive differences occurred, when comparing
methods, that could be attributable to biased reporting: (1) in the recall
data, actors reported extremely high success rates in change
relationships goals (83 per cent), but not in the diary method (62 per
cent successful); (2) in the diary data, actors made selfish requests
routinely of others, but such requests were rarely reported in the recall
data. In essence, actors do not want to admit that they were not in
control of relational changes (initiations, escalation and, perhaps most
importantly, de-escalations: consult Hill, Rubin and Peplau, 1976), and
do not like to admit how frequently they make selfish requests of
others. With respect to the present data, most (76 per cent) of the
defendants failed to avoid penalty. One would expect that they would
have to attribute the failure to some type of cause external to their own
limited or inappropriate way of influencing representatives of the
court. Hence, data in this goal situation lend themselves easily to
biased reporting (the judge was prejudiced, there wasnt time for a full
explanation, mine was the first case heard and he was trying to set an
example, and so on).
A second issue relevant to recalled versus observational data deals

with the reliance on likelihood-of-use rating scales. Although none of
our own work and, fortunately, few studies in accounting have relied
on likelihood-of-use data, this is not the case in research in compliance
gaining generally or in related work on relational disengagement.
Likelihood-of-use scales are seven-step Likert-type measures that
assess the actors self-reported probability of using a particular type of
tactic or approach in a particular influence situation. This approach
suffers from at least two limitations. Firstly, we gain little new insight
about communication phenomena by limiting the range of responses
that can be made (consider, for example, the different types of logical
proofs we recorded; see above). Secondly, rating scales provide biased
estimates of what people actually say and do because of question
threat (see discussion in Cody, 1982). That is, actors do not want to
endorse an extreme position that might make them look bad to the
experimenter. Such biased reporting has appeared in studies of
relational disengagement (Cody, 1982), in which actors who used
blatantly rude tactics according to their essays on How I disengaged
rated the same type of tactics as only moderately high in their own
likelihood of use (see also Burleson et al., 1986; Seibold et al., 1986
concerning the issue of likelihood-of-use scales).
One problem with the study reported here has to do with

generalizability. While similar (if not exactly the same) results are to
be expected at other traffic courts in the USA, the goal and the
environment are so specific and unique that these findings are not
likely to be generalizable across many types of goals or contexts. In
most interpersonal environments (as in McLaughlin, Cody and OHair,
1983; McLaughlin, Cody and Rosenstein, 1983), we anticipate that
concessions, apologies and excuses would lead to more honouring than
logical proofs (except, perhaps, if ones friends and family were all
judges or lawyers). There are situations which could aptly be described
as hybrids. They are similar to the traffic court and patrolman studies
in that they involve explaining oneself to a representative of an
institution, but similar to the interpersonal studies in that they require
the accounter to explain himself to someone who may know him well
and with whom he can anticipate further encounters. We think in this
context of situations such as: explaining to the local librarian why
juniors book is three weeks overdue; accounting to the principal for
ones tardiness; explaining ones absence from the faculty meeting to a
department chair; explaining to an editor why ones reviews are late;
and so on. We would anticipate that neither our courtroom findings nor
the findings from our earlier studies on accounting in interpersonal
relationships would precisely suffice to predict patterns in these
contexts.
Other limitations to this study are primarily methodological in

nature. If we focus attention on how we coded the defendants oral
arguments to explain their alleged offences, more worthwhile information would have been obtained if we also interviewed defendants
about their plans to influence the judge and their thoughts on what
evidence and argument they believed to be persuasive in this context,
and solicited their accounts for why they failed or succeeded in their
courtroom experience. If we focus attention on the outcomes of the
trials, we would (if we repeated the study) probably sample more cases
from fewer judges and make a comparison of the judges evaluations.
Some of our observers, for example, felt that one judge was
particularly lenient on the unemployed defendant who claimed that he
or she was unable to pay the fine: the judge frequently and
automatically assigned such defendants a relatively small number of
hours of community service. Another judge seemed to be particularly
impressed when a professional (doctor, lawyer, etc.) appeared in court
with records and documents and used the blackboard (obviously
showing considerable planning) to contest a $40 or $50 fine. A fully
developed model of courtroom outcomes of the kind studied here
should take such biases into consideration. Finally, when replicating
this study we propose that some attention be given to the apparent
strength of evidence. Based on the results of this study, we recommend
that a defendant use logical proofs (with supporting physical evidence)
to avoid penalties. However, some pieces of evidence may lack
persuasiveness. For example, a woman was ticketed for parking in a
no-parking zone, and she claimed that she could not tell whether the
kerb was painted red or not:
Defendant: Also, 1 took photos, [to prove the red paint was faded]
Judge: Well, lets have them here. [The judge studied

these photos for a long while]. Nope! These sure look red
enough to me. Guilty as charged.Explanations as
accounts:
a conversation analytic perspective
John Heritage
In this chapter I will discuss some central aspects of explanations or

accounts as treated using conversation analysis. My concern will be
with accounts which are naturally occurring in conversation rather
than elicited by an investigator. I will focus on the use of accounts in
ordinary conversation rather than some more specific or specialized
location in social space such as a hospital, a school or a courtroom.
And I will largely concentrate on accounts which are addressed to, or
explain, some current action going on in the conversation, rather than
accounts that focus on events that are wholly external to the
conversation in which they occur (though see Cody and McLaughlin
in chapter 8 for an approach to the latter type of data). The central
conceptual focus of this discussion will fall on the role of accounts in
relation to conversational organization and social solidarity.
Conversation analytic perspective 349
Conversation
analysisframework
has developed
over the past 15 years
as a subdiscipline
within
the
wider intellectual
of
ethnomethodology
(Garfinkel,
1967).
Primarily
concerned
with
social
action
and
its
underlying
reasoning,
its
focus
has
been
descriptive
and
naturalistic
rather
than explanatory
or experimental. Practitioners have
aimed,
in the
first
instance,
toexplanations
describe
how common-sense reasoning and social action
work,
rather
than
to
develop
in
advance
of detailed have
description.
As
reference
to
reasoningrevolution
suggests, Garfinkel
andsciences
his collaborators
formed
partthe
of
the
recent
cognitive
in
the
social
but,
as
a
sociologist,
Garfinkels
preoccupation
was
with
how
social
actors
can
achieve
a
shared
or
common
apprehension
of the depends
social world.a multiplicity
He proposedofthat
every
aspect
of a cognitively
shared
social
world
tacit
(taken
for granted)
methods of
reasoning.
These
methods, heonargued,
are both socially
organized
and socially shared and they are ceaselessly used during every waking moment to recognize
ordinary social objects and events.
350 John Heritage

In order to demonstrate his claims, Garfinkel devised numerous procedures which were
designed to engineer drastic departures from ordinary expectations and understandings about social
behaviour. The results clearly showed the extent to which the shared nature of ordinary
understandings is dependent on the joint application of shared methods of reasoning. His work also
demonstrated that the application of these methods by social actors has a normative background:
those who departed from the use of these methods were met with anger and demands that they
explain themselves and their actions. His investigations thus displayed the inherent morality of
social cognition and that the sense making procedures we all use are themselves part of the moral
order (Garfinkel, 1963, 1967; Heritage, 1984).
Garfinkel concluded that shared methods of reasoning generate continuously updated implicit
understandings of what is happening in social contexts - a running index, as it were, of what is
happening in a social event. It is through the creation of this running index that social activity is
rendered intelligible or, as Garfinkel puts it, account-able. To make sense, the overt descriptions
and explanations (or accounts) which actors provide for their actions must articulate with these
already established implicit understandings.

Both ethnomethodology and conversation analysis are thus concerned with two levels of
accountability. On the one hand, there is the taken-for-granted level of reasoning through which a
running index of action and interaction is created and sustained. On the other, there is the level of
overt explanation in which social actors give accounts of what they are doing in terms of reasons,
motives or causes. In this chapter I shall mostly be concerned with the latter level. But I shall later
show that an understanding of the properties of accounts at this second level cannot be fully
achieved without reference to how the running index works.
Methods of analysing sequencing in conversational interaction
Conversation analysis (henceforth CA) has developed primarily as an approach to investigating the
normative structures of reasoning which are involved in understanding and producing courses of
intelligible interaction. The objective is to describe the procedures by which speakers produce their
own behaviour and understand and deal with the behaviour of others. The central resource for
analysis is interaction itself. Interaction forms such a resource because during its course the parties,
whether intentionally or not, implicitly display their understanding and analysis of what is
happening as it happens. These implicit displays are embedded in the participants own actions. CA
represents the development of an analytic technology that capitalizes on this fact. Its central focus
is on the analysis of sequences of interaction and of turns within sequences. Thus CA is centrally
352 John Heritage

concerned with the study of the sequential organization of interaction and of the reasoning that is
inherently embedded within it.
A central assumption of interactional sequencing is that, unless otherwise signalled, each turn is
addressed to the matters raised by the turn preceding it. The most powerful expression of this
assumption arises in the form of adjacency pairs (Schegloff, 1968; Schegloff and Sacks, 1973), in
which the production of a first conversational action or first pair part (for example, a greeting, a
question, an invitation, a request, an offer and so on) both projects and requires the production of a
second (for example, a return greeting, an answer and so on). This projection and requirement is
normative in character. Thus a questioner, for example, whose question has not been answered will
usually have the right to repeat it or to request that the intended respondent answers it, to sanction
the non-respondent, or to draw sanctionable inferences from the respondents lack of response. It is
in terms of adjacency pair rules, which relate a first to a second action, that speakers can influence
or even constrain the conduct of their coparticipants.

However, rules that relate first to second actions are not just important resources by which
interactants can shape the trajectory of sequences of action. They are also important resources
through which interactants can grasp how others understand their actions. Consider the following
example. (Note that in each example the convention is to indicate where the talk comes from. In
this case it is SBL, which is the code for a large number of telephone calls collected in Santa
Barbara, California. Fuller details of these conventions can be found in Atkinson and Heritage,
1984). In this example a speaker B responds to a question:
(1) (SBL: 10:12)
A: Why dont you come and see me somertimes.
B:
*! would like to.
Here Bs response exhibits an analysis of As question. Specifically, in being shaped as an

acceptance, it treats As initial question as an invitation geared to the future, and this treatment is
plainly available both to A and to us (the analysts). However, suppose that B had responded as
follows:
(2) (Invented variation on (1))
354 John Heritage

A: Why dont you come and see me sometimes. B: Im sorry. Ive been
terribly tied up lately.
Given that this response takes the form of an apology, it would be clear to A (and to analysts) that
B had understood As question as a complaint directed at Bs conduct in the past. The relationship
between actions in sequence thus provides an interpretative resource both for participants and for
those who are concerned with the scientific analysis of interaction, because each action in a
sequence inherently displays its producers interpretation of the prior actions in the sequence.

Using the sequential organization of action as a data resource in this way has allowed CA to
unravel many complexities in the organization of actions and sequences and in the reasoning that
informs their construction and interpretation. In developing these analyses, CA has made a number
of assumptions. These are:
Interaction is structurally organized.
356 John Heritage

The significance of each turn at talk is doubly contextual in that (a) each turn is shaped by the
context of prior talk and (b) each turn establishes a context to which the next turn will be
oriented.
No order of detail in interaction can be dismissed a priori as irrelevant to the parties

understandings of what is occurring (see Heritage, 1985, forthcoming for further details).

These assumptions have strongly shaped CAs approach to data and its analysis, to which we now
turn.
Data collection and analysis
CA is primarily concerned with analysis of the organization of mundane social action. Researchers
have proceeded by collecting materials from ordinary conversational interaction which is as
uncontaminated as possible by social scientific intervention. These materials are invariably
collected using audiotape or videotape. Audiotape is particularly suitable for the telephone medium
where, because there is no visual channel available to the participants, there is no need for
researchers to have access to a video recording. In collecting data from face-to-face interaction,
video recording is by far the most appropriate means.
358 John Heritage

This emphasis on the use of tape as a means of collecting data from natural settings is based on
the following considerations. First and foremost, the analysis of interaction in detail cannot proceed
without repeated reviewing of materials. The use of note taking or on-the-spot coding of behaviour
cannot allow the researcher to recover the details of original materials, and is indeed already
analysis in itself. Naturally occurring data are strongly preferred to the development of data
generated through the use of role plays or experimental techniques because, although these
methods are undoubtedly valuable, they tend to restrict both the range and the authenticity of the
data that they generate, no matter how cleverly they are contrived.
In place of these methods, CA has adopted the naturalists strategy of building up large
collections of data from as many natural sites as possible. Like a good collection of naturalists
specimens, these growing data bases contain many variations of particular types of interactional
events whose features can be systematically compared. Analysts constantly seek for new variants
and may focus their searches on particular settings in the expectation of finding them.

Once possessed of a corpus of data, CA operates in the first instance using inductive search
procedures. An analyst who is interested, for example, in how invitations are accepted or rejected
will begin by building up a collection of invitations and will attempt to establish regularities in the
organization of positive and negative responses to them. At the core of this task is the
demonstration that these regularities are methodically produced and oriented to by the participants
as normative organizations of action.
To achieve this end, deviant case analysis is commonly used. This means taking cases where
the established pattern is departed from and showing the ways in which the participants, through
their actions, orient to these departures. If both dimensions of the analysis can be adequately
accomplished, then the empirical task of showing that a particular normative organization is
operative in interaction (that is, underlying both the production of and reasoning about a particular
social action or sequence of actions) will have been achieved. Beyond this point, there is the
theoretical task of specifying the role which the organization that has been discovered plays in the
communicative and social matrix of interaction.
360 John Heritage

In what follows a brief demonstration of these tasks will be accomplished using explanations or
accounts in social interaction as illustrative material. It is emphasized, however that this is an
illustrative sketch and is not to be treated as a full-blown analysis.
Empirical examples
We begin by delimiting the domain of interest. Our focus will be on the use of explanations or
accounts that are provided in the immediate context of the activities they account for. We thus
ignore narrative explanations of events that are external to the conversation in which the account
occurs, and explanations which are external to the particular sequence of conversational interaction
in which the account occurs. Those kinds of explanations would have to be treated in different
ways.
Establishing a pattern

Our subject matter has already been the object of considerable philosophical and social scientific
exegesis (for example Antaki, 1981; Gilbert and Abell, 1983; Mills, 1940; Peters, 1958; Scott and
Lyman, 1968). This literature converges in the judgement that explanations and accounts are
routinely provided or demanded in contexts where projected or required behaviour does not occur.
A comparison of our case (1) (repeated here) with another case will begin to illustrate this
phenomenon:
(1) (SBL:10:12)
A: Why dont you come and see me somertimes.
B:
4 would like to.
In this case the invitation, which projects an acceptance, gets a simple unvarnished acceptance. No
account is provided for the acceptance. In effect, the acceptance is treated as projected or provided
for by the invitation. By contrast in (3), which is taken from the same conversation, a subsequent
and more specific invitation is rejected in an elaborated fashion.
(3) (SBL:10:14)
A:
362 John Heritage

Uh if youd care to come over and visit a little while this
morning Ill give you a cup of coffee.
B: 1 > hehh Well thats awfully sweet of you,

2 > I dont think I can make it this morning
3 > .hh uhm Im running an ad in the paper and-and uh I
have to stay near the phone.
It can be readily observed that in (3) the rejection of the invitation is accounted for
with an explanation (arrow 3). Moreover, the account is the final component of a turn
that also includes an appreciation of the invitation (arrow 1) and a mitigated or
cushioned rejection component (arrow 2). All three features are highly characteristic
of rejections of invitations and related actions and have been extensively documented
elsewhere (Davidson, 1984; Wootton, 1981). These same features can be readily
illustrated in relation to other domains of action, for example in responses that reject
offers
:

(4) (Her:OII:2:4:ST) (Bs wife has slipped a disc)
A:
rWell
atsi
mean * can we do any shopping for her or something like tha:t?
4
B:
And we were wondering if theres anything we can
do to help
(0.7)
> Well thats most kind Anthony
2 > .hhh At the moment no::.

3 > Because weve still got two bo:ys at home.
And a similar, though less elaborate, pattern is observable in the context of responses to
questions:
(5) (Trio:2:II:l)
M: What happened at (.) wo:rk. At Bullocks this evening. =
364 John Heritage

P: =.hhhh Well I donkno:::w::.
Here the question relevances the production of an answer which, in this case, is not forthcoming.
In its place, the intended answerer provides an account (ignorance) for the absent answer. A more
complex version is to be found in the following:
(6) (W:PC:1:MJ(1):18) (Concerning how boat-trains work)
J:
But the trai:n goes. Does thtrain go o:n thboa:t?
M: .h .h Ooh Ive no idea:. She ha:snt sai:d.

Here the intended answerer not only accounts for her inability to respond by asserting ignorance,
but also then accounts for her ignorance by reference to a third party, whose forthcoming journey
is the occasion for the sequence in which this question arises.
Here then are a range of instances in which a second speakers failure to accomplish a
projected, or looked for, action is accompanied by an explanation or account of some kind.
Deviant case analysis
366 John Heritage

The second stage of the analysis involved establishing that the provision of accounts in such
contexts is treated as a normative requirement and this, it will be recalled, can be done through
the analysis of deviant cases. The latter are inevitably less common than examples of the normal
pattern and are often more complex to explicate.
A relatively straightforward case is the following, in which S has announced a

disastrous examination result which will prevent her from entering law school. Her
coparticipant asks her if she will take the examination again using a turn design that
clearly presupposes that she will
(7) (Frankel:TC:l:l:4)
G:
So yih gnna take it agai:n?=
S:
=nNo.
(0.5)
G: 1 - Noi?
S:
No.
(0.3)
G: 2 Whyno:t.=
S:
= .t.hhhh I dont rilly wanto.
This sequence shows a number of significant features. Most notably,

after Ss initial and unexplained reply, G waits for fully half a second (a
long time in a conversational context) for some explanatory elaboration
before prompting it with a turn (arrow 1) that queries Ss announced
decision. Ss next response is also unelaborate and G waits for a further
period before initiating an overt request for an explanation. Thus Gs
conduct throughout the sequence evidences her belief that an
explanation is due. And Ss apparent reluctance to volunteer one, in
effect, manoeuvres G into the overt pursuit of one.
A more complicated case involving a similar form of covert

manoeuvring around an absent account is the following. This example is
taken from a situation in which a group of men are engaged in sharing
out goods and fittings from a store which has closed down. From time to
time their talk turns to who will take possession of particular goods. Ms
question, which opens the example, is clearly heard as a request for a
fish tank which V subsequently rejects flatly and without explanation
(arrow 1):
(8) (US:simplified)
M:
What are yuh doing with that big bowl- uh tank.

Nothing?
(0.5)
V:
((Cough))
V:
Uh::
(1.0)
V: 1 Im not intuh selling it or giving it. Thats it.

M:
Okay.
(1.0)
M: 2 That was simple. Khhhh huh huh heh=

V: 3 - =Yeh.
(0.7)
V: 4 > Yeh.
(1.0)
V: 5 * Becuz selling it or giving it I::, thats (all there is)
Heres Ms ironic remark (That was simple) and slight laugh (arrow 2)
audibly sanctions Vs brusquely unaccounted for rejection. V meets it
with a further reassertion of his decision (arrow 3), another sotto voce
repetition of Yeh (arrow 4) and, subsequently, the initiation of an
explanation (arrow 5) that ultimately trails away into nothing. Here,
subsequent to the rejection, the conduct of both parties evidences their
orientation to the absent explanation.
Similarly, in the following sequence Bs unexplained failure to phone

his vicar (arrow 1) is waited out over a number of turns by A, who
thereby gets B to offer the looked-for explanation (arrow 2) without
having to request it overtly:
(9) (Campbell 4:1)
A:
Well lis:ten, (.) tiz you tidyu phone yer vicar ye:t,
(.3) "
B: 1 * No I aint.
(A): 1 -> (.hhh)

A:
Oh:.
(0.3)
(A):
A:
B: 2
.hhhhhAh::-::r::
Mwzgonna wait ntil you found out about....
Once again both parties exhibit an orientation to the moral requiredness

of an account for the reported failure. In As case, this is exhibited by not
advancing the talk beyond the place where an account could be
provided. In Bs it is exhibited by the subsequent offering of an account
in response to As conduct.
In sum, through deviant case analysis we can determine that account

giving is not merely an empirically common feature that is associated
with unexpected or unlooked for actions, but is a normatively required
feature of such actions. Since failures to provide accounts attract either
overt or covert pursuits of them or sanctions, we can conclude that the
giving of accounts in such contexts is itself a morally accountable
matter.
Embedding the analysis of account giving in a
wider analytical context
So far, we have sketched some
aspects of a demonstration that
account giving is a normatively
organized feature of sequences of
interaction in which one of the
parties acts in an unlooked-for
way. Now we can proceed to the
final analytical task mentioned
earlier, namely the attempt to
locate our normative organization
within a larger framework. What I
want to do here is to show that the
short extracts we have been
looking at so far can be placed in a
broader context. There are more
general aspects of social
interaction that only emerge when
one stands back and looks at a
more inclusive range of issues.
Here I shall sketch two broad

themes: what accounts tell us
about the management of self-other
relationships, and their role in
sustaining the underlying
normative structure of social
interaction.136 John Heritage
Accounts and self-other relations
A number of the details of the organization of account giving suggest

that they are strongly sensitive to issues of face (Goffman, 1955;
Brown and Levinson, 1978). Among a number of features, we will
briefly deal with the content of accounts, the internal organization of the
components of the turn containing the account, and the temporal
placement of such turns.
Content. In many of the cases shown in this chapter (for example (2),
(3), (5) and (6)), second speakers account for their failure to carry out
the proposed or required conversational action by reference to their
inability to do it (cf. Drew, 1984). Patently, the speakers could have
accounted for their actions in other ways. They could have asserted an
unwillingness to carry out the proposed action, or denied either the right
of the first speaker to propose it or their own obligation to respond. It is
significant, however, that the latter accounts would all, in one way or
another, have threatened the face of their co-interactants. By contrast,
the accounts which invoke inability (and in case (4) a lack of need for
the help offered) all have a no-fault quality (Heritage, 1984). None of
them implicates a lack of willingness to respond in the proposed way, or
challenges the others rights in the situation. All of the responses avoid
any threat to the other speaker, and they also avoid any threat to the
social relationship between the parties.
Internal organization of turn components. The turns that do not

accomplish a projected activity often exhibit a careful balance between
self- and other-attentiveness. Thus it can be noted that in the rejections
of the invitations and offers (examples (4) and (5)), the speakers
appreciate what is offered before they produce their refusals. Through
this temporal ordering of the components of their responses, the refusing
speakers attend to the others viewpoint before attending to what is
relevant from their own perspective. That this other-attentiveness is
significant can be illustrated by considering what can emerge when
speakers dont show it. In the following sequence, which is analysed in
detail in Schegloff (1988), a student returning to a group of friends has
evidently failed to come back with a promised ice-cream sandwich.
(10) (SN 4:7)
S:
You didn get an ice-cream sanwich,
C:
S:
I kno:w, hh I decided that my body didnt need it,

Yes but ours di:d
Here Cs conduct fails to display any element of other-attentiveness.

She does not apologise for her lapse, and her account is framed
exclusively in terms of her own desires and concerns. Indeed, inthe
way that she presents the lapse as the result of a deliberate decision
rather than, for example, a failure of memory, she in fact aggravates the
original failure. Her self-attentive account is immediately sanctioned
by a matching self-attentive riposte by S, which is purportedly
produced on behalf of the group, and C subsequently remains at odds
with the rest of the group until she leaves a little later with the issue
still unresolved.
378 John Heritage
Temporal placement of account turns. A further dimension of otherattentiveness that is built into the structure of rejections also centres on
temporal ordering. In (3) and (4) the rejection is delayed by being
placed after a variety of other turn components: an appreciation
component, standard pre-rejection objects like uh and well, and, in
(4), by a delay of 0.7 seconds before the turn is initiated. These delay
features, which are systematically present in rejections, have been
shown to create opportunities for first speakers to revise their prior
actions so as to make them more acceptable (Davidson, 1984;
Pomerantz, 1984; Wootton, 1981). Where such revision is successful
and second speakers find that they can then go on to carry out the
proposed action, they have enabled first speakers to save face by
being accepted after all, and they have also enabled second speakers to
avoid threatening the others face by a rejecting action. The building of
delay features into the design of rejecting actions is thus also
intrinsically other-attentive.
All three of these features of accounts for rejecting actions - their

content, internal organization and temporal placement - converge in
extracts (11) and (12) below. In (11) the invitation recipient produces
an initial appreciation of the invitation (arrow 1) which is sufficient to
indicate to the inviter that a rejection is in the course of being
produced. In response, the inviter specifically enquires (arrow 2) about
circumstances that might make the invitee unable to accept:
(11) (NB:II:2:14: slightly simplified)
E:
Wanna come downn av a bite alu:nch with me:?=

=1 got sm bee:r en stu:ff,
(0.2)
N: 1 * Wul yer ril sweet hon:, uh::m

()
N:
rlet-I:ha(v) -i
E: 2 > Lor dyuhav sum-pn else (f)
380 John Heritage
Notice here that Es last utterance (arrow 2) works to save the face of
both speakers. It anticipates and invites a rejection account from N
rather than just leaving it to her to produce it on her own behalf. And
the kind of account that it invites (that is, one based on ability) is, as
we have seen, least face threatening for E herself.
A similar process is apparent in (12), with one important difference:

(12) (JG:II(b):8)
G:
Whenih you uh: what nightsr you available.
1 - (0.4)
M:
G: 2
M:
.k.hhhhh.rhh
*-Are you workin nights et allr anything?
I do: I work, hhh a number o:f nights Gene......
Here the inviting party G anticipates that there will be a difficulty with
the invitation simply from the delay (arrow 1) in the response from his
co-interactant. In this instance, as in (11), the inviter proposes a
difficulty about the invitees ability to accept and his action again
serves to save face all round.
382 John Heritage
All of the features discussed in this section are socially standard, or

institutionalized, features of dispreferred action sequences. The
motivation for their use ultimately derives from this fact of social
standardization. Speakers are treated as having the capacity to produce
or avoid producing these features. They may thus find themselves
being accountable for and evaluated in terms of how they design and
package their utterances. Thus, given that accounts are
institutionalized, the speaker who fails to provide an account may be
sanctioned as wilful and self-attentive, as one who would not, or could
not, be bothered to provide an account. One who fails to provide a nofault account can be construed as hostile or insinuating or, at the least,
careless of the face of a co-interactant. One who disorders the relative
placement of an account and an appreciation (so as to produce, for
example, a rejection followed by an account followed by an
appreciation) may be heard as self-centred, if not downright selfish.
One who rejects something too quickly may be heard as rude.
The social logic underlying these judgements is socially institutionalized along with these institutionalized features of the rejections
themselves. Both are embodied in a web of moral accountability whose
binding character rests on its seamlessness. Within this web of
accountability, the pressures are consistently towards otherattentiveness and towards the maintenance of face, of social
relationships and of social solidarity. The personal sensibilities of face
are intelligible through this logic, and in turn these sensibilities
motivate its maintenance as an institutional form.
Accounts and. the structure of social action
Earlier in the chapter, I proposed that ethnomethodology and CA are

concerned with accountability in two senses of the term. The first is the
issue of intelligibility - the running index through which interaction is
given constantly updated interpretation - while the second concerns the
384 John Heritage
occasions and the ways in which participants explain their actions. I

also proposed that the two dimensions are deeply interwoven. In this
section, I want to comment on the connections between the two levels
and to suggest a social role for ordinary accounts which cuts still
deeper than the maintenance of social solidarity and self-other
relations. This role has to do with the maintenance of the methods or
procedures through which the organization of social action is sustained.
As we have seen, CA focuses on shared methods or procedures as

the central resources through which actions are both produced and
understood. The maintenance of these methods is a crucial feature of
maintaining both the social organization of action and the social
intelligibility of action. Both dimensions stand or fall together. As we
have seen, for example, the fact that answers to questions normally and
properly follow the questions to which they respond is a facet of the
social organization of action, but it is also a resource in the social
organization of interpretation - a resource which supplies the

presumptive basis on which to interpret utterances which follow
questions as answers. For speakers, then, this presumptive linkage
between questions and answers is both primary and presuppositional
on the one hand and normative on the other.
Take the following rule: after the production of a first utterance

recognizable as a question, a second speaker (the addressee) should
produce an utterance that is hearable as an answer to the question. This,
as we have seen, is both a rule of conduct and a rule of interpretation.
The role of accounts in relation to the maintenance of this rule can be
approached by considering the following problem. Our rule is one of a
multiplicity of methodic procedures through which social action and
social interpretation proceed. But, like all rules, it is not always
complied with. The fact that social rules are often not complied with
could conceivably lead to a situation in which, at the social level, the
386 John Heritage
rule ceased to be respected and, at the psychological level, the rules

cognitive salience was eroded. In short, the incidence of non-compliant
actions might imaginably create a process of attrition in which the
social reality of the rule progressively decayed. How then do accounts
function to prevent this? We can begin to address this issue through the
following consideration.
When a speaker remains silent in the face of a question, two major

types of interpretative option are available. The first is that the rule that
questions should be answered no longer makes sense of, or applies to,
the current situation and that some other rule (yet to be determined)
applies instead. We scarcely ever contemplate this option, let alone
implement it. If we did, the social organization of action (and with it
the social intelligibility of action) would have to be treated as
contingent and haphazard. Rather than adopt this option, we cling to
the presupposition that questions relevance answers and, in this sense,
should be answered.
The second interpretative option is that the silence requires

explanation. It may be explained by the speakers deafness, failure to
recognize the question, rudeness, lack of willingness to answer,
inability to answer without self-incrimination, or whatever. At all
events, the failure is treated as requiring explanation and, indeed, it is
a positive signal for us to initiate a search for an explanation that is
appropriate to the circumstances. The explanations which may be
arrived at under such circumstances are almost always negative in their
implications for non-responding parties, and this factor may be a major
motivation for them to produce either compliant actions or to produce
their own accounts for non-compliance in order to forestall the
negative conclusions which might otherwise be drawn.
388 John Heritage
It is crucial to recognize that this second interpretative option

embodies the presupposition that questions relevance answers. For
speakers there are only two options: either the rule linking questions
and answers is complied with and the question is answered, or the lack
of an answer is an exception to the rule which requires some kind of
secondarily elaborative explanation. In either case, the cognitive
centrality of the rule is maintained by treating the rule, together with
the explanations for non-compliance with the rule, as exhaustive of the
full range of possible contingencies. The exceptions with their
explanations thus become the exceptions that prove the rule because
the provision of such explanations maintains the rules presuppositional
status both as a rule of conduct and as a rule of interpretation. Once
again, we encounter a closed circle of interpretation. Presuppositional
rules of action and interpretation interlock with the organization of
account giving to form a seamless web - a self-motivating, selfsustaining and self-reproducing normative organization of action.
Generalizing this analysis, we can suggest that rules of action and

interpretation are sustained in the first instance by an accumulation of
instances that exhibit compliance with them. But they are also
maintained by the provision of accounts which explain away
instances of non-compliance in ways that, by their very provision,
sustain the presuppositional status of the rules. Thus at the deepest
level of cognitive and social organization, the role of accounts is
intrinsically bound up with the maintenance of the methods informing
social cognition and the social organization of action. Accounts
function to repair the ubiquitous relevance of rules of conduct by
protecting them from the entropic process of attrition that could
otherwise arise from the incidence of non-compliant actions. Ordinary
explanations of action, no matter how trivial and apparently
inconsequential, thus play a crucial role in maintaining the foundations
of social organization itself.
390 John Heritage

While it is not easy to evaluate the usefulness of a methodology as

complex and multifaceted as CA in a few words, some basic comments
can be made. As already noted, CA has been developed as a
methodology which is specifically geared to the analysis of the
structural organization of social action. During the past 15 years or so,
many hundreds of detailed studies of interactional organization have
been developed using this methodology in a number of countries. This
widespread use of CA techniques testifies to their transportability
both within and across national boundaries.
The effectiveness of CA techniques comes from their exploitation of

fundamental facts about how interaction is organized. These techniques
are therefore specific to the analysis of interaction. It is unlikely, for
example, that CA techniques would be as effective in the analysis of
explanations provided in a pencil-and- paper procedure that was
relatively disengaged from interactional organization and constraints.
CA analyses are also highly focused. For example, in the analysis of
the provision of accounts for rejected invitations, empirical analysis
might well show that, where accounts of the rejection are given to third
parties as opposed to the inviting individual, many of the features of
accounts sketched above would be absent or considerably modified.
In terms of its specific empirical techniques, the strengths of CA are

balanced by corresponding limitations. The discussion of both can be
grouped under two headings: data collection and data analysis.
392 John Heritage
Data collection
The concern for naturally occurring data has resulted in data collection
techniques which, like the naturalists specimen hunting expeditions,
have so far been largely unfocused. These techniques are geared to the
specific tasks of conversation analysis and, as such, have important
justifications. First and foremost, CA is at present a fundamentally
descriptive exercise which is concerned with the natural history of
social interaction. Its aim is to describe the structural organization of
social interaction and its associated reasoning in as many social
settings (and as many languages) as possible. The focus is on
fundamental structures and the ways in which they both create and
influence choices among courses of action. It is important to recognize
that this task has only been seriously addressed in the past few years
and that we have no reason to suppose that it will be completed in a
short period. It is also important to recognize that an understanding of
how the structural organization of interaction works is not a descriptive

or theoretical luxury, but an essential basis on which the interpretation
of a wide range of social, social psychological and communicational
data may depend.
Much of CA research is focused on data derived from mundane

conversational interaction, and this approach is the product of a
considered analytical strategy. It is based on the recognition that
ordinary conversation is the predominant medium of interaction in the
social world and the primary form to which children are first exposed
and through which socialization proceeds. There is thus every reason to
suppose that the structures of mundane talk form a kind of benchmark
against which the more specialized forms of communication,
characteristic of for example the courts, the classroom, the news
interview, doctor-patient interaction and so on, are recognized and
experienced. This supposition has been confirmed by an increasing
394 John Heritage
body of CA studies focused on these forms of institutional interaction.

It is clear, therefore, that the study of ordinary conversation offers a
principled approach to describing the distinctive character of these
more specialized forms of interaction and also interactions involving
the asymmetries of status, gender, ethnicity and so on.
There is an undeniable temptation to accelerate the investigative

process by using role plays to focus the collection of data or to use
experimental techniques to gain control over variables which may
influence the character of interaction. However, these procedures may
seriously limit the range and kind of social interaction that is produced
and, more generally, may compromise the naturalness of interaction in
a variety of ways.
Thus, although it can be slow and cumbersome, the CA style of data

collection does hold out the hope that a full range of interactive
procedures will eventually be captured. Moreover, there are other
advantages to be gained from the CA approach. In particular, the taping
of natural interaction creates enduring data bases of interaction which,
because they have not been shaped by the constraints of a particular
research design, can be reused in a variety of investigations. Any strip
of interaction will contain exhibits of a very large range of different
interactional phenomena which can be compared with other exhibits
from other data corpora. The accumulation of data corpora into banks
of data will greatly facilitate the process of comparison.
These are some of the positive features of the CA approach to data

collection. These features are intrinsically associated with the fact that
CA is primarily geared to the analysis of the structural organization of
interaction (see Duncan and Fiske, 1977 for an account of structural
396 John Heritage
versus variable analysis approaches to the analysis of interaction).

They are also associated with significant limitations in the use of CA
data. In particular, the data are collected without the use of
experimental controls on the content of interaction or controls on the
sociological or psychological characteristics of the participants. The
absence of these controls may contribute to the roster of acknowledged
difficulties in assessing the impact of sociological (for example,
gender, ethnicity, social class and so on) and psychological (for
example, personality) variables on the specific outcomes of
interactions.
In sum, the data collection techniques used by CA reflect a specific

set of research objectives and strategies. By their very nature, they
occlude other research objectives. This is not to say that the research
techniques of CA are ultimately incompatible with other approaches.
On the contrary, there is every reason to anticipate the development of
complementary research programmes that integrate the findings of

structural and variable analysis approaches. The adoption of its chosen
approach to data collection reflects CA decisions about the priority of
structural analysis in the research process at present.
Data analysis
In this chapter a basic three-phase framework for data analysis has

been sketched, comprising (1) the inductive search for regularity, (2)
deviant case analysis and (3) theoretical integration with other findings.
Once again, this framework is specifically appropriate for the analysis
of normative structural organization in interaction. It cannot be applied
to patterns of interaction which are not normatively organized because
in such data deviant case analysis, in particular, is inappropriate.
398 John Heritage
Although this methodology can be simply stated, it is complicated to

apply. The complexity and specificity of the structural organization of
interaction is such as to make the mechanical application of research
technologies inappropriate. Much therefore depends on the ability of
the analyst to isolate regularities and pattern in data cleanly and
without overgeneralization. CA data analysis has developed as an
operation in which empirical advances in particular domains have
increased descriptive purchase in other, related areas of investigation.
For example, the discovery that response tokens (such as mm hm,
yes, oh and so on) are not a set of undifferentiated back channel
utterances, but are systematically differentiated in terms of their
placements, valences and tasks, has enabled researchers to gain access
to new understandings of how topics in conversation are organized
and shaped. Effectiveness in CA research is not guaranteed by the
application of a fixed set of methodological canons, but rather by
resourceful use of the corpus of current knowledge.
CA research thus involves careful attention to the contextual and

sequential details of interaction. It requires extensive listening or
viewing of tapes, careful transcription of interactional detail, and
exhaustive analysis. These tasks are time consuming but their results
have proved rewarding and influential.
Conclusion
10
In this chapter, I have attempted to explore some

of the interests which ethnomethodologists and
conversation analysts have in lay accounts and in
the more generic phenomenon of interactional
accountability. I have also tried to illustrate some
aspects of CA methodology by examining some
cases in which conversationalists account (or fail
to account) for their conversational actions. And I
have examined aspects of the fundamental role
played in social interaction and social life by
interactional accountability and the giving of
accounts. From an ethnomethodological and
conversation analytic point of view, the last
aspects lie at the very heart of the normative
organization
of
social
life.
Accounts
and
accountability are fundamental features of social
10
organization
itself
and
their
foundational
significance is not restricted to the domain of

social interaction (Pollner, 1974, 1975, 1987;
Mulkay and Gilbert, 1982; Gilbert and Mulkay,
1984; Heritage, 1984). Accordingly the topics of
this book have a particular relevance for those
social scientists who are concerned with the
phenomena of social organization. They are also
of concern to anyone who has an interest in the
fact that, from the point of view of social
participants, the members of the social world are
agents
who treat one another as morally and
socially accountable for their actions.Completion

and dynamics in explanation seeking
10
Ivan Leudar and Charles Antaki
In this chapter we want to talk about a framework for analysing what

happens when speakers ask each other for explanations. This kind of
10
dialogue is causing a degree of puzzlement in artificial intelligence

and human-computer interaction circles, as some articles in, for
example, Norman and Draper (1986) show. The workers in the field
know that, without an explanation, the information given by a socalled advice system can be sterile; but they have had severe
problems in designing computers better able to volunteer explanations,
or to understand the explanatory requests that users sometimes make
and to give them satisfactory replies (Alvey KBS Club, 1987).
10
It seems that what is missing is a theory to guide the design of

systems which could engage in explanatory dialogues (or, at worst,
which would explain why such systems are not feasible), despite the
fact that conversations have been studied intensively in psychology,
sociology and linguistic pragmatics. The problem is that not enough
attention is paid to the variety of dialogue types, to dynamics of conversation or to the interplay between the individual and the social, and
all of these are essential in understanding explanations.
10
We shall outline some concepts we have been developing which

would allow the study of such aspects of dialogues. We shall not say
much about explanations themselves, but shall try to persuade the
reader that some of the ideas of turn-of-the-century pragmatic
philosophers are useful tools, especially if used in conjunction with
the methods and findings of modern pragmatics.
The focus here is on the work of G.H. Mead and, in particular, on his
notion of completion. This is developed and used in analysing how
10
participants in dialogues construct each others (and their own) moves

as explanations, conceiving of conversations in terms of asocially
distributed cognitive system. The notion of completion is also used to
develop a framework for conceptualizing dialogues as dynamic
processes, which is necessary if one is to understand how it is possible
that a dialogue game between participants may metamorphose as they
talk to each other, and if one is to understand the function of
explanations in that metamorphosis.
Completion and conversation
10
Activities are social in that the acts begun within the organism require
their completion in the action of others (Mead, 1973: 446, our italics).
(We shall refer to the completion of one persons act by another person
as other-completion; and the completion of ones own act as selfcompletion.) Mead, working in a tradition which had still not lost faith
in grand theories, intended his notion of completion to cover the whole
ground of human behaviour. It seems to us that Mead was basically
right to postulate completion, even if he left it general and unspecified.
Most intentional behaviours are not just aimed to evoke social
responses, but require them in order to become themselves social
actions. For something to qualify as a purchase, for example, the
money tendered by the purchaser has to be accepted by the
shopkeeper. The act may have begun with my wanting a chocolate bar,
but it would fail to be a purchase until the shopkeeper took my money.
The behaviour is not constituted into a purchase by the customers
intent only. For something to count as a reception at court, there has to
be a monarch to receive ones homage and graciously to return ones
bow. The participants have to agree on the situation and have to show
evidence of that agreement in their ready completion of actions
initiated by others. The homage is defined both by the intent of the
supplicant and by the features of the objective context including the
actions of others.
10
Although faith has been lost in the tradition of grand theories such
as Meads, echoes of the notion of completion are still to be found
here and there in social science thinking on communication. It is at
least implied in such concepts as intersubjectivity, joint action, mutual
knowledge, framing and so on. One can see a clear (but
unacknowledged) Meadian heritage in modern pragmatics.
Conversational analysis in particular (e.g. Sacks, Schegloff and
Jefferson, 1974; Schenkein, 1985; Atkinson and Heritage, 1984)
seems to us a framework which makes use of the idea of external
completion. This is obvious from the fact that its starting position
about the analysis of conversation is premised on taking exchanges,
rather than individual utterances, as units of communication:
Utterances are in the first instance contextually understood by
reference to their placement and participation in a sequence of actions.
For conversation analysts, therefore, it is sequence and turns within
sequences, rather than isolated sentences or utterances, that have
become primary units of analysis (Atkinson and Heritage, 1984:5).
Even this, Levinson (1983: 304-5) points out, is a first approximation,
and examples of much more complex units of conversational
organization can be found: our discussion and examples, however,
focus on two-move exchanges.
10
The point is that in the conversational analytic framework, individual utterances achieve their significance as moves in exchanges. In
this, conversational analysis is radically opposed to speech act theory
(Searle, 1969; Searle, Kiefer and Bierwish, 1980) in which the
illocutionary force of speech acts is defined by their form and the
propositional attitudes of their authors. The controversy is made clear
in an argument put forward by Levinson (1979). He points out that a
promise (one of the classical illocutionary acts) is not a promise
unless it has been taken up by the addressee. If I promise to pay you
back for the meal you just bought, does what I said still stand as a
promise if you tell me not to be silly, it was nothing? Well, without the
uptake, I certainly would not enter into the commitment that is the
point of promising.
10
A systematic analysis of the extent to which different acts of

communication can be incomplete is lacking, although incompleteness
has been argued for requests (Habermas, 1975) and the indeterminacy
of communicative intent has been argued for indirect speech acts
(Edmonson, 1981). Indirectness and the use of various illocutionary
force indicators make it obvious that speakers in fact exercise some
control over the extent to which their utterances are open (to
interpretation).
10
In general, the argument is that at least some of the conditions

necessary for an individuals action to count as a speech act cannot be
satisfied without the uptake by an audience or are only satisfied on the
uptake. This is one sense in which individual acts can be underdefined,
open to interpretation, and requiring a completion. The conditions
which a behaviour must satisfy to be an action need not be coextensive
with the mentality of an individual but may be distributed over the
individual and the context of her action. This argument has a close
parallel in philosophical debate on whether meanings are in the head
(e.g. Putnam, 1975; McGinn, 1983).
Conversational analysis thus admits openness and consequently allows

the significance of an utterance to be negotiated. The initial move of
an adjacency pair constrains the audience to a set of completions
(partially ordered for preference), and thus partly defines the structural
identity of those completions. In fact, the deviation from the set of
follow-ups is normally signalled bythe audience. The adjacency pair
initial move is, however, itself basically open in that some of its
properties remain to be defined by the addressees move which affirms
or (reinterprets) its intended identity. What follows an offer can be an
acceptance, an evasion or a refusal; what preceded an acceptance was
an offer! The completion, in its own turn, interprets or reinterprets the
preceding initial move. Thus one way to specify Meads notion of
completion is to use conversation analysis and say that completions
can be understood in terms of exchange structure.
Empirical example
Most of the data we use come from dialogues that we arranged

ourselves. The aim was to set up situations which were similar to
human-machine dialogues. They are rather contrived, but the
purpose was to simulate the situation of someone asking a machine for
an explanation, which is itself a rather peculiar situation.
In each dialogue there were two participants. One was always our
confederate, who was working from the same knowledge base - in this
case, unemployment and its causes. We gave the confederate no
instruction about unemployment, and the knowledge base was simply
his own beliefs and theories about it. His goal in conversation was
given as informing the other participants (one at a time!) of his views
on unemployment. The other participants were naive to the purposes
of the study. They were asked to try and find out the confederates
explanation of unemployment. In some dyads, respondents were of the
same knowledge status as the confederate; in the others they were
junior to the confederate, who was introduced as an expert on
unemployment.
Our data indicate that not all other-completions reflect and are
determined by the exchange structure. Another level at which
addressees complete messages is that of content, and the cohesion of
moves is a matter of logic. Examples of structural completion would

be a greeting being reciprocated (as in (1) below) and a question being
answered (as in (2)). In both of these the second move completes the
exchange structure. In (3), in addition, the pragmatic identity of the
initial move is fixed: Ks Thanks establishes Ls previous move as an
(indirect) offer.
P: Hi, Im Pete.
416 Ivan Leudar and Charles Antaki

J:
Hi, Im John.(2)
J:
Do you think this government is to blame for the unem

ployment?
P: To some extent.
(3)
K: Im starving.
L: I have a pound left.
K: Thanks.
L: Right.
Completion and dynamics in explanation seeking 417
Cases (4) and (5), on the other hand, are examples of cognitive
completion:
(4)
P: I would have to ask them whether they mean that it [new
industrial revolution] is purely a technological change.
J:
Or a complete reorganization of society, the same way as

the industrial revolution.
(5)
CA:(i) If you educate people in a certain, say, vocational way,
then you produce a certain kind of skill and a certain resource
to be used by the economy. If you dont, if you have a liberal
education, for example, you have a certain range of other
skills for the economy.
BF: (ii) So do you then believe that the education system is at
fault, that it should be changed, and if it were changed, there
would be less unemployment?
CA:(iii) I dont believe anything so direct, no. I think its just one
aspect of demographic variables, which are themselves an
aspect of economic trends, which are themselves an aspect of
unemployment.
These are different from the completions in (1), (2) and (3), and it is
not clear how they are to be explained in terms of exchange structure.

In (4), Ps proposition (that the new industrial revolution is a purely
technological change) was completed by J by a further proposition
(that the new industrial revolution is a complete reorganization of
society) and forming a disjunction (new industrial revolution is either a
purely technological change or a complete reorganization of society).
In (5) BF treats the argument put forward by CA as a premise and
provides a conclusion, thus completing her move into an implication
which CA subsequently rejects. In (1), (2) and (3), what was at issue
was how a response completed the initial move into an adjacency pair,
and the pragmatic identity of the moves was fixed. The completions in
(4) and (5) are different. In (4), for example, Js move is not itself a
sentence, but it completes the sentence just uttered by P into another
sentence. The cohesion of the completion is not provided by exchange
structure but is in this case a matter of grammar and logic. One can
certainly say that in (4) there is now one new and complete statement
which is distributed over two participants. In (5) the implication is at
first distributed between BF and CA, although CA subsequently denies
his individual commitment to it.
We seem to have moved into controversial territory by suggesting

that meaning and cognition can be socially distributed and not
coextensive with individuals; this is at odds with social cognition as
currently understood in most experimental social psychology (as in for
example Fiske and Taylor, 1984). The notions on relationships between
cognition and individual/social divide have changed over time. The
distinction between subject and environment and in particular their
radical modularization was certainly not always as sharp and as taken
for granted as it is nowadays (Novack, 1975). Dewey put it this way:
The distinction [between an individual and the environment] is really
a distinction due to our interest; it is an intellectual distinction. ...
There is a functional unity, a unity of the life process which maintains
itself in, and by means of, a continuous series of diverse changes
(Dewey, 1975: 275). We may not agree with Dewey in detail, but had
the winds of social change blown in another direction we should all
now be contextualists (or, according to radical critics, we should have
given up the distinction altogether; see Henriques et al., 1986).
Self-completion and internal completion
Mead was obviously sensitive to the problem of individual/environment dualism and offered a solution, some aspects of which are
relevant here. He claimed that individuals act as their own audiences,
completing internally their own utterances:
In the human organism the pattern of the whole social act is in some
sense initiated in the individual as the pattern of his act. The
mechanism of this is the effect, which the gesture of the organism has
upon itself, that is analogous to the effect which it has upon the
other. ... When this gesture, as is the case in the vocal gesture, tends to
arise in the individual who makes it the response or responses which
it calls out in the other or others, there may appear in his organism
the initiatory stages of the act of the other or others. (Mead, 1973: 446,
447, our italics)
So according to Mead, not only the addressee but also the author
completes her (individual) acts.
It may simply be a fact that an individual has no option but to react

to herself as another, and to this extent objectively (even though the
actual perspective on herself may vary and not be accurate). The idea
of internal completion may seem, nevertheless,
on the face of it, rather bizarre; and if the parts we italicized are
misread appropriately, does away altogether with an individuals need
to communicate. And of course it would do so if the internal selfcompletion was exactly the same as the prospective external othercompletion: but it is not. The internal self-completion is not simple;
often, and probably as Mead intended it, it establishes only the type of
a response, not a specific response. The question is, to what extent is
internal self-completion like other-completion?
Our strategic hypothesis is that internal completions and the overt

moves tie together according to the same principles of discourse
structure, logic and syntax as moves tie in ordinary conversations. Our
speech samples contain two kinds of relevant evidence - extended
turns and false starts. These have been noted previously (for example
Labov and Fanshel, 1977) and analysed under the heading repair.
(Levinson, 1983, chapter 6). In neither of the conversations from
which examples (6) and (7) come did the audience indicate (and
presumably notice) anything odd. In fact, the turns make very good
sense if we assume that the speaker was completing her own overt
moves internally and subsequently reacted overtly to the tensions
between the moves and their completions.
Extended turns such as (6) are like ordinary conversations in that

they consist of a series of recognizable moves, all of which are,
however, produced (overtly) by one person without being interrupted
by another person:
(6)
P: What did you think of Jills daughter?
I:
(i) She is quite interesting, isnt she? (ii) Whats her name?
(iii) Isobel. (iv) Right.
In this example, (i) presupposes that the identity of the person referred
to by she is established. Let us assume that the internal completion of
(i) queries this presupposition and the speaker I discovers that he
cannot specify the identity of she by naming her, although her name
is on the tip of his tongue. Question (ii) is supposed to resolve the
tension by finding the name from the other participant P, but the fact of
putting the question seems to make the name available and I provides
the answer for himself in (iii). In fact, the answer is acknowledged in

(iv).
The second type of evidence is sequences of false starts, as in the

following:
(?)
A: (i) So what has the government - (ii) if you blame the
government - (iii) you blame the government - (iv) because
they havent done the things that you think they should do,
like reflation?
This text makes sense if we make a couple of assumptions. Firstly, (i)

was going to be a question: So, what has the government done to be
blamed for the unemployment? This presupposes that the addressee
actually believes that the government is to blame. Secondly, the author
has completed the (intended) question (i) intrasubjectively by, for
example, but I do not believe that the government is to blame. In
other words, the speaker is uncertain that the presupposition of (i) is
shared. The tension between the presupposition of the question (i) and
the intrasubjective completion is expressed in (ii), which is supposed
to solve the tension. In fact the speaker answers (ii) himself and
thereby resolves the uncertainty in (iii); and proceeds to answer his
own original question in (iv).
The process of internal self-completion makes it possible for the

speakers intention to change in mid-sentence as she becomes aware of
ways she might be misinterpreted or misunderstood; or perhaps, as she
starts talking, she foresees likely rebuttals or objections. In any case,

the original aim of the utterance needs to be realized in a form different
from the original.
In a sense we are improvising on the ancient claim that one must

treat thought processes as inner dialogue. The conception goes back to
Plato, and has in psychology been particularly elaborated by Vygotsky
and his followers (Wertsch, 1985). In fact, even the notion of the
language of thought (for example Fodor, 1975) is also broadly
consistent, even if the conception of language is often rather restricted.
The advance is that it seems possible to apply the findings from
pragmatics to thought processes and, in fact, the position has been
cogently expressed by Carlson (1983).
Meads concept of internal completion seems to be useful but

requires a careful analysis and tightening up. The problem is that he
has never given a systematic account of which completions are
possible and what are their consequences. Once this is attempted, the
notion becomes less straightforward. The first complication is that an
addressee may respond to an authors move by a variety of moves,
including dispreferred ones (such as evasion or querying
presuppositions). This poses a problem for extrapolating the response
back into the initiators mind. Is it possible for an initiator to ask a
question and at the same time to make preparations for doing
something as dispreferred as answering it untactfully? Can one issue a
request and also stubbornly refuse to carry it out? In other words, are
there constraints on internal completions absent in ordinary dialogues?
The problem is that the evidence is likely to be selective, and only the
internal completions by dispreferred moves are externally exhibited in
false starts, as seems to be the case in our data. The only candidate for
self-completions by a preferred move in our data is (iv) in example (6),
and even then it is external rather than internal self-completion. One
possible mechanism which would produce such biased external selfcompletions is the following. The speaker internally completes her
own move and a tension may arise between the move and the
completion; the function of the external self-completion is to resolve
the tension. That is what we have in fact assumed in the analysis of
examples (6) and (7). It may be possible to formulate a relatively
general method to guide decisions on the identity of internal selfcompletions. We start with two assumptions.
Assumption 1 The external self-completions resolve tensions between

moves and their internal self-completions.
Assumption 2 The set of possible internal self-completions (of a move)

consists of the conversational moves which could follow that move
in ordinary, well-formed conversations (and this is where
conversation analysis helps).
On these assumptions, it is possible to determine or at least to narrow

down the identity of internal self-completions. The move being
completed is known and so is the external self-completion. One has to
choose from the set of possible internal self-completions the one which
would produce the tension resolvable by the external self-completion.
It seems to us that an analogue of the postulated process of intrasubjective completion operates observably in dialogues and was
manifested in the conversations between the participants in our study.
The analysis of extended turns can be complemented by analysis of
dialogues, and both contribute towards a classification of kinds of
tensions which produce dynamics and of their resolutions. In the next
section we give some relevant examples.
Completion and dynamics in conversations
Our text suggests that inconsistencies between on the one hand a

moves significance for the author, and on the other its interpretation as
implicit or expressed in the completion by the addressee, are one major
source of dynamics in conversation. An example of this was already
given in (5). There BFs interpretation (ii) of CAs argument (i)
produces a tension which is supposed to be resolved by the explanation
in CAs final move (iii). Another example is as follows:
(8 )
BF: (i) So what else affects unemployment?

CA:(ii) Demographics.
BF: (iii) What does that mean?
CA:(iv) Age, sex, immigration, emigration, schooling, education.
BF: (v) In what way has education led to increased
unemployment?
CA:(vi) The way you put the question sounds as if you mean: Is
it not paradoxical that you are claiming that education leads
to increased unemployment?
BF: (vii) I did not say anything about education causing more
unemployment.
Here, in move (v) BF draws a conclusion that education has led to

increased unemployment. CAs subsequent move (vi) challenges this
inference. What happens is that one participant treats another
participants move as a premise and provides a conclusion, this
resulting in a distributed implication. The process is again obvious in
(9): (ii) is interpreted in (iii) as a premise, and both constitute a
distributed implication which is challenged in (iv).
(9)
A: (i) So how do you think one should go about reducing
unemployment?
B: (ii) Well, its ... perhaps by reducing the working hours and
overtime.
A: (iii) So they would be getting less money.

B: (iv) Well, yes, but not necessarily because you could ...
Exchanges (5), (8) and (9) all contained examples of cognitive

completion which helped us to understand some of the dynamic
characteristics of the dialogues. Similarly, examples can be found of
structural completions which play a role in dynamics, mainly where
the intended pragmatic identity of a move is not affirmed in uptake.
The examples suggest that some moves in conversations might be
generally understood as resolving tensions between one participants
moves and their interpretations in another participants moves. One
way in which explanations operate in dialogues is to establish the
communality of beliefs where completions indicate its absence, as
happened in move (iii) and text (5) and move (iv) in text (9).
At the functional level of analysis the process of self-completion

does not seem to be too different from the dynamics characteristic of
ordinary dialogues. The characteristics of tension giving rise to
dynamics are, however, not exhausted by their formal descriptions.
We have attempted to extend Meads notion of completion and use it to

analyse conversations. The advantage seems to be that the framework
acknowledges links between individual cognition and communication
and allows the treatment of some cognitive processes as socially
distributed. The approach also enabled us to consider the dynamic
aspects of conversations and to analyse how tensions between
individual moves and their interpretations seem to motivate much of
conversation dynamics.
The disadvantages are questions, as always, of level of analysis.

Does the approach pay too little attention to the content of the
conversation, the participants roles, and the macro-structure of the
interaction in general? The worry is that by emphasising its formal
characteristics, we overlook the conversations idiosyncratic flavour.
We might also be losing sight of how the episode functions as a
discourse, that is to say as a rhetorical display that has an implied or
explicit social force.
11
Our way of thinking is not incompatible with

other levels of analysis. Whatever view one takes
of conversations the fact is that they work
extremely well, and peoples success in bringing
off such a difficult exercise needs explanation.
The extension helps show that at least part of the
explanation lies in treating cognition as a social
process. This pulls us away from a more
mentalistic account of talk, and into the sort of
social account where contacts with other
disciplines can be made.Accountability within a
social order:
the role of pronouns
Rom Harre
11
One of the most important webs within the fabric that makes up a
social order is the network of accounts. Accountability may be
differently conceived in different societies and rights and duties to
explain oneself may be differently distributed, but the giving and
receivingofexplanationsisprobablyubiquitous.Rightsandobligations to
perform accounting acts are non-uniformly distributed among the
various ranks and categories of societies, and are in continuous
historical flux: witness the changing rights of women and children in
our own society within the last century. There are the experts, the
credible and untrustworthy, the penitent and the brazen, and so on. The
variety of contributions to this book is in part occasioned by the
variety of ways in which position in a grid of roles and status locations
appears in and influences the procedures of accounting.
438 Rom Harre

In this chapter I want to direct attention to the fine structure of the
languages in which accounting is generally expressed. Accounts must
be tied to the array of relevant persons, on pain of unintelligibility.
Among the most effective devices by which speech is linked to speaker
and listener is the system of pronouns of its equivalent. These ties are
effected through the linguistic property of indexicality. Pronouns are
indexical expressions since they mark the relation between speech acts
and person, place and time of their utterance. The role of indexical
devices has been underemphasized in Anglophone studies, since
English, for a variety of historical reasons, is amongst the most
indexically impoverished languages in current use. The expressions
available for denoting persons in many other languages encode, to a
greater or lesser degree, the social relations between speakers and
listeners. And this may be so to such an extent that the very grammar
of the language determines what can be said to whom. This adds an
important dimension to the understanding of accounts.Theoretical
background
A court [in Germany] has ruled that familiarity means contempt and
has ordered a woman fruit pedlar to pay a fine of 2250 marks for
insulting a policeman during a confrontation in the market square.
Testimony brought out that the woman continuously addressed the
policeman as du\ the familiar form of you. The policeman used the
formal and respectful Sie and thereupon filed a complaint of insult
and misdemeanour. The woman is appealing against the verdict.
(Herald Tribune, quoted in Brown and Gillman, 1970)
Accountability: the role of pronouns 439
This news story nicely illustrates the power that grammar has both to
express and to offend social status. The womans explanations,
whatever they were, were not at fault. We are not told whether or not
she actually had a vendors licence. The unacceptability of her account
turned on a choice of pronoun.
That languages encode social orders in the rules of their grammar

is one of the central tenets of the social constructionist approach to the
study of human psychology. But they do so in varying degrees. The
Japanese language encodes most of the Japanese social system,
whereas at the opposite pole English encodes very little. No competent
speaker of Japanese could produce an account which did not also
express the requisite condescension and deference called for by the
social relations between the persons involved. It is advisable to begin
our approach to the encoding of social relations with an analysis of
some of the Western European languages, on grounds of both
440 Rom Harre
familiarity and simplicity.
A powerful influence on psycholinguistic thinking was the work of

Brown and Gillman (1970) on the grammar of pronouns. Their theory,
which I shall take as my point of departure, can be summed up in two
principles. The first is status: a person will use the T form to refer to or
address an inferior, and will receive the V form in return. The second is
solidarity: status equals who are not members of the same solidarity
(that is, roughly speaking, intimate group) will use the V form to each
other, while members of a solidarity (say an Army platoon) will use T.
The generic V/T distinction is realized in different ways in the different
European languages: in French tu and vous; in German du and Sie; and
so on. The variety of ways of grammatically marking the V form is a
matter of some importance. English now has no honorific forms, and
has lost the distinction between singular and plural in the second
person. The honorific second person in Italian and Spanish is
grammatically third person (as it is in American military speech). The

explanation is simple enough. Lei is a contraction of la vostra Signora,
just as Usted is short for vuestra Merced. But why the plural form
should have taken on the role of the honorific is another matter. Brown
and Gillman offer three possible explanations, not necessarily mutually
exclusive.
The most fanciful - though I suspect it is the one they secretly

prefer - is that it derives from Diocletians reforms. In these, though
there remained two Roman Emperors, there was only one system of
administration. Necessarily then when addressing the ultimate source
of power one used the plural vas. Eventually, they say, the Latin
plural was extended to other power figures, but they do not suggest
why! Their second suggestion is that royalty and heads of state and
family are plural in another sense: that is, they are the summation of
the people. However, if the Royal we is used in addressing the
442 Rom Harre
subjects, should not a plural You be offered by them in return?

Finally they suggest that anyway plurality is a metaphor for power,
since no doubt two horses can pull a heavier load than one. In the
absence of any hard evidence one can take ones pick.
The stability of the total system is threatened when an asymmetrical

status relation exists between people who inhabit a common solidarity,
such as horseman and groom, professional and amateur cricketers, and
so on.
Finally it was Brown and Gillmans contention, even in the revised

version of 1970, that solidarity semantics has tended to displace status
semantics as the basis for the grammar of pronouns and other person
deictic devices, at least in Europe. Thus there would be two coacting
tendencies. There would be a tendency towards a perceived symmetry
in relationships favouring VV and TT over the asymmetrical forms,
and a further tendency as solidarity attains dominance to favour TT
over VV. Linguistic changes involved in the disturbances of 1968 seem
to support this thesis, as indeed to the attempts by the ideologues of the
French Revolution, such as Robespierre, to impose an egalitarian
grammar based on tu. The spirit of 1789 is still abroad. I was present at
a meeting of the faculty committee, a kind of soviet, at Aarhus
University in 1972, at which a middle aged professor, hostile to the
student revolution, threw in the towel by agreeing to use the intimate
pronoun and to accept it when addressed to himself. The fate of these
managed changes is instructive. As in the France of Napoleon, so in
the post-1968 era the simple shift to TT has not occurred. Danish has
drifted back to a condition near to its pre-1968 style.
444 Rom Harre
What of Brown and Gillmans theory? I believe it is a reflection of

a folk theory of pronominal and other forms of address in accounts and
indeed in speech in general. People seem to believe that by saying
something it can be made to be so. And here we come across one of the
dimensions in which the priority of linguistic relations to other social
relations must be qualified. The ancient changes in English do not
seem to be susceptible to so neat a sociolinguistic explanation as
Brown and Gillman offer for current changes in European languages,
yet it seems that their dimensions of status and solidarity can
nevertheless usefully be evoked. In English up to and including the
Elizabethan period, the second-person pronoun was ye in the plural
and thou in the singular. You was the accusative of ye, while the
accusative of thou was thee. So one would say I want you to do
this as a polite alternative to the brusque Ye shall do it. But you
drifted across the linguistic landscape to become the honorific singular,
and is so used by Shakespeare. I owe a quite tidy explanation of this
drift from accusative plural to nominative singular and finally to a
ubiquitous second-person reference to Michael Silverstein (1985). He
believes that the Norman French practice of using the plural V form
was copied by the speakers of English, to create a grammatical
analogue of the status/solidarity semantics. So thou/thee would be
the pronoun of solidarity and of deference, while ye/you would be
the pronoun of distance and of condescension. Drawing on Brown and
Gillman, Silverstein cites Sir Edward Cokes contemptuous remark to
Raleigh: All ... was at thy instigation, thou viper; for I thou thee,
thou traitor. In the mouth of such a person as the Attorney-General,
thou is a pronoun not of solidarity but of contempt.
The Quakers wished to repudiate both kinds of human relations

expressed in the ye/you forms, since they wanted to use that part of
the language resource which would emphasize relations of intimacy
(closeness on the solidarity dimension) and to delete those usages
which expressed the asymmetries of deference and condescension.
Curiously the Quaker authors, such as George Fox, argued the case for
thou and thee on the basis of the impropriety of the use of a plural
form of address for a single person. In linguistic terms Fox and others
took the plural to be the marked value of ye/you. Now we can see
why ye/you became the general forms for second-person address.
Unless one adopted the distant alternative one would be heard, when
using the old singular solidary form, to be talking like a Quaker. But
why the accusative should have displaced the nominative requires a
further explanation. I have already hinted at one above. It is also
characteristic of English to use indirect forms for the giving of orders,
rather than drawing on specific politeness terms. In modern English
Would you mind doing ... is routinely preferred to Please do ... .
Since the indirect forms usually have the pronoun in the accusative, the
tendency to generalize to you seems unsurprising. Like all honorifics,
you can also be used as an ironic belittling device, a usage found in
Elizabethan drama. By the end of the seventeenth century thou and
thee had disappeared outside Quaker circles, to be replaced by the
ubiquitous you, lacking both a marked plural and an honorific
counterpart. Two pseudo-plurals youse and you-all have drifted in
and out of use, but I know of no pseudo-honorifics. Without the plural
model which seems to have animated the first shift, perhaps a
contracted honorific title, in the Spanish manner, could have come on
the scene. Yug for your grace hardly has the right deferential ring! It
may be, however, that the rapidly changing English social order with a
rising bourgeoisie needed what Brown and Gillman calls a linguistic
waiting room, a device which leaves social relations unspecified.
Grades can then be marked with official titles.
446 Rom Harre

Empirical: social relations and accountability in
Japanese
The analysis of ways of encoding social relations in European

languages has focused our attention on pronouns and the two social
dimensions of solidarity and status. But even the more elaborate
European languages provide their native speakers with a fairly limited
resource for pre-empting the best positions from which to give
accounts and to hear and comment on the accounts of others. The
Japanese language is tied into and encodes a much more complex
system of social relations than almost any other. Indeed the Japanese
social order is so fully encoded in grammar that accounting must
always take place within an effectively non-negotiable grid of
condescension and deference. This has a profound effect on whose
stories have the greater weight in the creation and re-creation of
society. Since this book presents a repertoire of methods, a word is in
order about the way grammatical analysis is to be undertaken.

Someone like myself who neither speaks Japanese nor lives within a
Japanese social world is dependent on the work of linguists. But
grammar is not enough, since the burden of the argument of this
chapter is the role of grammar as laying the groundwork of a social
framework within which all linguistic interchanges must occur. The
analyses of experts must then be examined against a kind of folk
ethnography of speaking that one gleans from native speakers and
explores through shrewd (one hopes) probing and eliciting of their
intuitions. I have had help from a number of native speakers of
Japanese, but above all from Shinya Kimura.
Harada (quoted
personal pronouns
European personal
essentially correct,
in Bachnik, 1982) has suggested that no real

exist [in Japanese] that correspond to the Indopronouns. This judgement seems to me to be
though for expository convenience I shall occa-
448 Rom Harre
sionally refer to some Japanese deictic expressions as pronouns. The

Japanese language provides the user with a wide range of very subtle
devices for interpersonal address and reference to which all account
giving must conform, but neither functionally nor etymologically are
they pronouns. We have already seen that deixis, and particularly
indexicality, is the dominant function of Indo-European pronouns in
use, with important sociolinguistic functions facilitated by this
underlying grammar. Japanese seems to occupy the opposite pole, in
that the social uses are etymologically prior and deixis and indexicality
are built upon them.
Three technical notions are needed to bring out the nature of the
Japanese systems of person reference and address: deixis, register and
cline. In an excellent discussion of these matters, Bachnik (1982)
draws attention to the fact that, concerning deixis, we may perhaps
safely assume that the deictic anchor point [indexical locus] is Ego in
societies where the individual is the basic unit of social organization.

But in societies where the basic unit is not the individual, the question
of the deictic anchor point and its relation to social organization must
be raised. In Indo-European social orders it is enough to know who is
the speaker and who is the person addressed. But in the Japanese social
order we need to know what is the relevant speakers group and the
group of the person or persons addressed. I hope the terms my-group
and your-group, though not ideal, are sufficiently unfamiliar to bear
the differentiation I want to mark.
The term register is used in linguistics to refer to a system of lexical

items by the use of which a number of non-linguistic distinctions are
represented or expressed. J apanese includes a number of registers for
expressing person distinctions. Since I am arguing that social relations
do not exist independently of sociolinguistic practices, I shall be taking
the use of the Japanese registers to be expressive rather than
450 Rom Harre
representative of social relations.
Finally the term cline will be necessary to refer to graded sets of

distinctions. Adownwardclineexpressessuccessivelymorecondescension or more self-deprecation, while an upward cline expresses more
respect or self-aggrandizement. The J apanese person registers turn out
to be complex systems of reciprocal and interwoven dines.
Now let us turn to some observations of word forms that express

deixis, register and cline. The role of second-person address is played
by a noun register created by the use of suffixes. The system involves
five terminations, related in a complex pattern. The pair -sama and
-san are used to express respect for the interlocutor, the former being
addressed to persons of a higher-status group than the group of the
speaker, and the latter for equals. Both are polite forms of address. The
termination -sensei, which can be used alone almost like a pronoun,
comes from a word for teacher. However, it is widely used in
addressing those to whom the speaker stands in other kinds of
dependent relation than pupil. At the other end of the cline the suffix
-tyan is an affectionate and condescending diminutive. Incidentally, it
is never used reflexively of speaker - a grammatical problem for small
children, who are frequently addressed as -tyan. Indeed, none of these
suffixes can be used reflexively. Womens grammar differs in many
ways from that of men in Japan, and there is a suffix -kun in use only
amongst men.
452 Rom Harre
Complementary to this register is another system used indexically

by speaker. But again one must be wary of treating this as some kind
of elaboration on the pronoun I. Not only is the indexicality directed
to the me-group, but the register includes a complex of dines of
formality and informality. Combining these first-person indexicals
with the second-person register of honorifics above permits an
amazing variety of speech acts. Harada gives an extensive list of such
words, from which I extract four. As far as I can grasp the system, the
principle seems to be that the more formal the first-person term, the
more respect is being shown to the addressee. I have a sense that in
some way the more formal the term the more self-status is being
expressed too. There is an androgynous cline with watakushi and
watashi in descending order of formality. When used by women each
term is one level down in formality, reflecting and expressing the
overall social status of women as a group relative to men. Then at the
lower end of the cline of terms usable by men are two informal words,
boku and ore. Boku is a term of speech rather than writing. It is not so
much informal as formally self-denigrating. As my Japanese informant
explained, you can use it when you know inside yourself that you are
quite important but wish to display public deference. Formerly
reserved to the speech of men, it has now come into use among liberal
and radical women. Ore is the true term of intimacy. It is never used on
first meeting. Moreover, it would sound very odd if used by a highstatus person in ordinary conversation, and indeed is never so used.
However, such a person could and would use it during a drinking
party, underlining the asocial nature of the temporary grouping, a kind
of life out of real social time and space. The most general informal
term in use by women is atasi. (If it is used by a man he would be
identified immediately as a homosexual.)
By choice amongst these terms the speaker expresses relative status

from a pole reciprocal to that expressed by the choice of second-person
term. Since any speech act must be analysed as a two-term relation
linking speaker and addressee, a language sensitive to relative social
position of the two termini, as they are members of this or that group,
must find expression. A better sense of the force of these first-person
terms and a clearer grasp of their difference from say English I ,
which always leaves the social status of the speaker vis-a-vis the
addressee undetermined, comes from observing their range of usage.
Bachnik (1982) quotes Wolfe to the effect that watakusi is used in lots
of ways that are not indexical of speaker. In these uses the term seems
to mean something like belonging or pertaining to this or that person.
However, there are two English first-person indexicals which do
express status in the Japanese way. These are the use of we in the
singular (appearing at the high end of a cline in the mouth of Victorian
royalty and at the low end in the mouths of the excessively selfconscious) and the use of one by the junior royals. I am informed
by Fred Vollmer that the Norwegian Je, which normally appears only
in one of the written forms of that language, can be used for spoken
acts of self-reference by persons of high status. If used by the lowly it
sounds absurd.
454 Rom Harre
Two other groups of terms serve pronominal deictic functions.

Kinship terms are differentiated to express the distinction between megroup and you-group, as well as to express respect. Bachnik points out
that the respectful otoosan (father) can be used to refer to your father,
to address my father and to refer to our father (when talking with
siblings). But another term titi is used when referring to my father in
talking to a member of the you-group.
The group indexicality of Japanese person terms is clearly shown by

the use of the word uchi (etymologically house) as a way of making
speaker reference. We must not call this self-reference because it is
not clear that the English concept carried by the word self can be
applied to the Japanese person at all. The second-person term
reciprocal to uchi is otaku, etymologically your house.
Summing up the principles involved in this fragment of grammar,

Bachnik remarks that the reference/address distinction is closely
related to the boundary of the speakers group.
The noun registers I have sketched illustrate the general principle

that the Japanese language permits the expression of social distance by
either advancing the status of the addressee or reducing the status of
the speaker (both kinds of speech acts taken as indexical of me- or
you-group) or both. Verb registers work in the same way. Again I owe
to Bachnik a very clear exposition of the complexities involved. Just as
in the noun registers there are always two dines at work, one from
formal to informal style and the other (best called for the verb register)
from exalted to humble. So the choice of verb form lies in a twodimensional space defined by two non-orthogonal axes, one of
formality and the other of deference. The axes are not orthogonal
because, as we have already notice, formal style of address is adopted
456 Rom Harre
when speaking to one of higher-status you-group. Sliding down both

dines expresses, so to say, double deference. Thus degree of closeness
expressed through greater informality is also an indirect mark of
degree of deference. (One notes in passing the inversion in English
forms of address in which a formal style is adopted for addressing
those lowly in an institution.)
I cite Bachniks layout of the variants of the verb to go in

illustration. We start with iku in the centre of the space. Choosing
instead the verb form irassyaimasu exalts the addressee, and in a
reciprocal response the addressee can choose mairimasu, humbling
himself. Again starting at the centre with iku a speaker can choose
iki-mas-u as a formal stylistic variant, leaving ik-u as the informal
contrast. Likewise in exalting address the particle can be inserted to
give irassyai-mas-u as a formal stylistic variant, leaving irassyar-u as
the informal. Since mairi-mas-u is humbling and so already formal the
-mas- cannot be omitted; you cannot be both humble and informal in

Japan. In addressing David Copperfield Uriah Heep managed this
sociolinguistic feat in English. But then we are a linguistic community
famous for our hypocrisy.
In analysing first-person deixis in terms of a collective deictic

anchor I have so far shelved the problem of plural self-reference. What
devices in Japanese correspond to we? Correspondence cannot be
exact since singular terms of self-reference take collective deictic
anchors. So the Japanese singular and plural self-referential devices
cannot stand in the same kind of contrast as do I and we. In the
following I am greatly indebted to Shinya Kimura. There are three
pluralizing devices in use in Japan. One can repeat the word, as for
example hito-hito (= person-person, that is people) or yama-yama (=
mountain-mountain, that is mountains). But these are not simple
plurals. This device accomplishes roughly what each does in English.
458 Rom Harre
Not all nouns can be iterated in this way. For instance neko-neko (=
cat-cat) is never used. The second device is the attachment of a suffix
from among -ra, -tachi, -domo or -gata. These can be added, with
minor exceptions, only to words denoting human beings or creatures
personified as such. It will come as no surprise to learn that choice
amongst these suffixes is motivated by considerations of status. The
suffix -domo is pejorative, for instance in the Marxist expression
shihonka-domo for capitalists. Sensei-gata is honorific, meaning
respected teachers. The other two suffixes are neutral, perhaps a little
casual or rude. For words of Chinese origin, pluralizing is done by
prefixing sho-. For instance, sho-zan means mountains; however,
again it is not simple plural, since it has the sense of every or a
considerable number of mountains. Professor Kimura has remarked
that the first and third devices are resorted to with growing frequency
to render the simple plurals in European texts. All three devices are
used to pluralize pronouns. The deictic term ware is no longer used in
the first-person singular and is currently used as a very vulgar and
pejorative second-person pronoun. The plurals ware-ware and ware-ra
are in use in literary contexts to express manliness, hard style speech.
Watashi-tachi is reasonably polite and commonly used, with watashira slightly more polite and watashi-domo at the humble pole. The last
is used in formal expressions of humility and by shopkeepers when
addressing customers. Boku-ra and boku-tachi sound like student
words, as Kimura puts it, though they are in common use even among
the elderly. They can also be used in contemporary J apanese speech to
refer to a we-group which includes women. It is not clear to me
whether these devices can be used to include the addressee. Some
difficulty must surely arise in using the humble watashi-domo when
the group referred to includes the addressee. Perhaps the growing
popularity of boku-tachi resolves the obvious difficulties.
Between the impoverishment of English and richness of Japanese

lies a vast range of possibilities of intermediate degrees of social
encoding. Tamil, a language of South India, is an instructive
intermediate example. I owe the following account of Tamil pronouns
to Vimula Mani. The system is of interest because of both its
elaboration of the TT/VV semantics, and its refinement of the firstperson plural. There are four we forms, sensitive to whether the
addressee is or is not a member of an intimacy or solidary group. Nam
is used among intimates, while nangal is used when the addressees of
the immediate collective do not belong to the solidary group. Similarly
the first-person dual is marked for the same distinction. Thus
namiruvar is the unmarked dual, when the pair are members of a
solidary, while the marked form nangaliruvar is used when the one of
the pair does not belong to the solidary of the speaker. Again in
second-person address three grades of social relationship are
expressible. M is impolite, ningal is intimate and friendly, while
nangal is polite. Interestingly Tamil has a pair of affixes which can be
used to create pejorative personal exclamations, corresponding to one
of the uses of the English you, as for instance in Take that, you
swine! Chi- is derogatory as in chi-nay (= you dog!), while the suffix
e is derisive as in kurangel (= you monkey!).
460 Rom Harre
Throughout the languages I have discussed, these kinds of

distinctions are operative in some degree. But just as in IndoEuropean the indexical system for place and moment, though not
carried by pronouns and not picking out speaker as such, nevertheless
in most non-Indo-European languages is person based. The Japanese
language operates in a person space independent of the demonstratives
by which indexical reference to physical locations in space and time
are achieved. In English the pronoun system and the demonstratives,
directionals like left and right and other devices for expressing
spatiotemporal indexicality, are independent of one another, just as
they are in Japanese.
According to social constructionism the role of pronouns in

conversation is to locate acts of speaking at locations in a social world.
Social worlds are structured by the fields we have called moral
orders. We would expect, all else being equal, that these structures
should be encoded in some degree in deictic systems, such as the

pronouns. But people are not only moral agents but also things in the
physical world. So we might expect the deictic system to ramify into
various devices for locating speech acts in physical time and space.
And this is just what the demonstratives enable speakers to do. Again
the question of how much of the structure of the relevant world is
encoded in the grammar of these devices is the matter of interest.
In English spatiotemporal indexicals are an independent system

from the person indexicals amongst which are the pronouns. Not only
is it an independent system, but it is markedly impoverished relative to
the range of devices available in other languages. Kami (old Javanese)
illustrates a system in which people space and physical space
constitute one manifold, but I shall not lay out the details of its
fascinating grammar here. As A.L. Becker (1974: 241) puts it, in
contrasting European languages with Javanese, person is an
462 Rom Harre
obligatory inflection in deictic usage. ... It is primarily person, the

marked shifts of point of view ... that gives coherence to Kami texts,
somewhat as tense and the sequence of tenses ... give cohesion to
English texts.... Notions like plurality, temporality, sequence and tense
are, in a sense, secondary metaphorical developments of the interplay
of person, location and definiteness. All space-time is person spacetime. It is impossible to find any corresponding European model, for
instance for inflecting pronouns for tense.
Taking all this grammatical material together, the message is clear. The
intuitions of how the social basis of accounting is created that we
deploy as English speakers are a poor guide for a more general social
psychology of accounts. Because of the social impoverishment of

English grammar as a code, most of the work of creating and marking
social relationships perforce is brought off by more or less explicit
devices, such as facial expressions, intonation and the use of titles and
other marked forms of address. Yet if we are aiming at a
comprehensive social psychology of account giving and receiving we
can hardly ignore the grammar of those languages in which social
relations are very richly precoded. Research into their structure and
conditions of use becomes a central target of concern.
I have advanced the analysis of grammar without, so far, doing

more than occasionally hint at problems that may beset the unwary
researcher. What are these problems? The thesis of the grammatical
determination of thought, inaccurately ascribed to Sapir and Whorf, is
surely false. Grammar constrains but does not predetermine all the
uses to which language can be put. Speech act theory makes as much
464 Rom Harre
use of social as of linguistic categories. Case studies, such as that of

Japanese, can lure one into forgetting that language is a system in use.
Limits in lexical resources can be turned by the power of context, and
often are. Furthermore, studies of grammar present snapshots of a
static language/society relation and so are radically incomplete.
Complementary studies of current conventions of use, and of their
location in the historical development of conversational implicatures,
are needed to fill out the dynamics of accounting. For instance it is just
not true that the grammatical similarity between masculine and generic
forms in English works to conceal the role of women in society.
Recourse to grammar by feminists may be a proper part of the rhetoric
of reform, but it is not a contribution to sociolinguistics, though the use
of grammatical observations as rhetoric is a proper subject for
investigation. Context and usage almost always disambiguate formal
similarities.
But, in so far as any act of accounting is the presentation not only of

reasons but of oneself, the representation of relative status, power and
solidarity vis-a-vis ones audience of that moment must be a central
matter of concern to the speaker. And, as such, it must remain at the
focus of any programme developed to research into the social practices
of accounting and the conditions that make them possible.PART IV
RHETORIC AND IDEOLOGY

12
Discourse analysis and the identification of

interpretative repertoires
Margaret Wetherell and Jonathan Potter
In this chapter we intend to illustrate the approach to lay explanations

taken by discourse analysts. We will attempt, in particular, to
demonstrate the value of interpretative repertoires as a basic analytic
unit. Firstly, however, let us situate this methodological approach
within broad developments in both social psychology and modem
theories of discourse.
Many social psychologies - even new ones - have a rather oldfashioned view of language. They assume that language acts as a
neutral, transparent medium between the social actor and the world, so
that normally discourse can be taken at face value as a simple
description of a mental state or an event. Peoples utterances might
occasionally be distorted by the desire for social desirability but these
cases of distorted discourse are unusual. More often, accounts are

taken to be simple, unintrusive, neutral reflectors of real processes
located elsewhere.
>The implausibility of this standard assumption in traditional social

psychology and in its alternatives can be seen when we look at the
burgeoning work in sociology, philosophy and literary theory on
language function. Analysts in these fields have demonstrated the
essential and inescapable action orientation of discourse (Heritage,
1984). Discourse has become seen as a social practice in itself, as
opposed to a neutral transmitter, with its own characteristic features
and practical consequences.
Discourse analysis and interpretative repertoires 469
There is no space here to trace the threads of our own preferred

theoretical framework - discourse analysis - in the intellectual
traditions of speech act theory, ethnomethodology, conversation
analysis, semiology and post-structuralism. The contributions by
Billig, (chapter 14), Parker (chapter 13), Heritage (chapter 9) and Cody
and McLaughlin (chapter 8) are complementary to our own, and
contain a great deal it would be wasteful to duplicate here. Let us start
from the position that utterances are acts, and that language is
functional all the time - not simply on atypical, special occasions. The
meaning of an utterance is not a straightforward matter of external
reference but depends on the local and broader discursive systems in
which the utterance is embedded.
470 Margaret Wetherell and Jonathan Potter
Clearly, we are not suggesting that disparate and contradictory perspectives on discourse such as speech act theory and post-structuralism
can simply be blended together to form a new analytic perspective.
These are separate traditions with their own disputes and difficulties.
But they indicate some issues traditional social psychologists neglect
when they continue to take participants talk as simple referential
statements indicating a more or less trouble-free path to actions,
attitudes and events. And from this basis we can begin to think about
the form a discourse approach to social psychology might take.
Discourse analysis
Discourse analysis can best be understood by introducing the

interconnected concepts of function, construction, variation and the
analytic unit: the interpretative repertoire. We will start with the notion
of function.
Both speech act theory and ethnomethodology stress the action

orientation of language use. People do things with their discourse; they
make accusations, ask questions, justify their conduct and so on. At the
same time, post-structuralist work suggests that we need to modulate
and supplement a study of the performative dimension of language use
with work on wider unintended consequences; when people deploy a
particular form of discourse, it has repercussions of its own which may
not have been formulated or even understood by the speaker or writer.
The discourse analyst incorporates both of these aspects under the

general term function. We can think of a continuum from more
interpersonal functions such as explaining, justifying, excusing,
blaming and so on, which define the local discursive context, to the
wider purposes discourse might serve - where, for instance, a social
analyst might wish to describe an account, very broadly, as having a
particular kind of ideological effect in the sense of legitimating the

power of one group in a society.
In some cases functions are very easy to identify. Some utterances,

for example, emerge clearly as explicitly meant speech acts. The
utterance, for instance, I name this ship the Titanic is an obvious
example. In the appropriate circumstances (Austin, 1962) it will be
perfectly clear to the assembled dignitaries and spectators that this
utterance has the function of ship naming. Similarly other specific
language functions such as requests, excuses and accusations may
frequently be quite unambiguous.
Functions of this kind are not always clear-cut, however. Much of

the time participants have good reasons for keeping the exact nature of
their utterances inexplicit. For example, Drew (1984) has demonstrated
how speakers often make requests indirectly, allowing the recipient of
the request to reject them without making this rejection obvious.
It can be equally difficult to specify the broader consequences of

discourse. Choice of terminology can have subtle effects which may be
overlooked by speakers. For instance, it is commonplace to describe
the current practice of moving the handicapped and the mentally ill out
of institutions and into smaller hostels as community care. One of the
functions of this particular description is to draw on the positive evaluations tied to community discourse and to develop a characterization
focused around the organic and agency metaphors which distinguish
community talk (Potter and Collie, 1987; Potter and Reicher, 1987).
Critics of the policy may find it necessary to deconstruct or negate the

community emphasis to effectively argue against it. But neither users
nor receivers of this discourse need be intentionally aware of these
consequences when formulating their description: it seems right. In
many cases talk which is simply packaged as describing the situation,
as the speaker sees it, can be analysed in terms of discursive
functions and effects which go beyond mere description.
For these reasons discourse analysis cannot be, in a straightforward

way, an analysis of function, because functions are not in general
directly available for study. One of the difficulties in any analysis of
language function and the actions contained in utterances is that we are
immediately involved in interpretation. Essentially, discourse analysis
involves developing hypotheses about the purposes and consequences
of language. As we have seen, discourse does not usually come ready
labelled with the functions neatly displayed on show, so that one kind
of form is always an accusation, or always marks out a rationalization,
or always suggests consequences which we could describe as
ideological in their effect.
The elucidation of function is one of the endpoints of discourse

analysis. That is, functions are the findings rather than the raw data.
Flow, then, are functions to be revealed from a study of discourse?
There is no single answer to this question; however, one important
response is that functions are revealed through a study of variation.
The fact that discourse is oriented to different functions means that

it will be highly variable: what people say and write will be different
according to what they are doing. An event, a social group, a policy or
a personality may be described in many different ways as function
changes from excusing, for example, to blaming or from formulating a
positive evaluation to constructing a negative one. Speakers give
shifting, inconsistent and varied pictures of their social worlds.
In many ways this is a highly counter-intuitive claim. Psychologists

are accustomed to regard the individual as a coherent, consistent unit the starting point of their investigations. Only one description is
possible of a state of mind, and once that description is achieved the
quest is complete. Variability within and between the accounts that
people give is not part of this image.
Variability is best demonstrated with data, and when we move on to

a concrete research example we will return to our argument. For the
moment our point is that variation has a crucial analytic role. As
variation is a consequence of function it can be used as an analytic clue
to what function is being performed in a particular stretch of discourse.
That is, by identifying variation, which is a comparatively
straightforward analytic task, we can work towards an understanding
of function. We can predict that certain kinds of function will lead to
certain kinds of variation and we can look for those variations (Gilbert
and Mulkay, 1984; Mulkay and Gilbert, 1982; Potter and Mulkay,
1985).
Again, we should emphasize that variability need not be a

consequence of deliberate or intentional processes. Much of the time
people in their lay explanations will not be strategically planning, or
self-consciously adjusting their discourse in a machiavellian fashion,

but just doing what comes naturally or saying what seems right for
the situation.
The fact that discourse is oriented, consciously or not, to particular

functions, which in turn throw up a mass of linguistic variation, tells us
that discourse is being used constructively. The realist model, outlined
earlier, assumes that discourse is organized in a way which reflects the
nature of the entities it describes. For discourse analysis, however,
language is put together, constructed, for purposes and to achieve particular consequences. So variation is both an index of function and an
index of the different ways in which accounts can be manufactured.
The term construction is appropriate for three reasons. Firstly, it

cues the analyst to the point that discourse is manufactured out of preexisting linguistic resources with properties of their own, much as a
bridge is put together using girders, concrete and cable, some of which
are flexible, some hard and so on. Secondly, it reminds us that active
selection is going on: out of the many available linguistic resources,
some will be used and some not. Thirdly, the notion of construction
emphasizes, once more, that discourse has an action orientation: it has
practical consequences. Much of our social lives depends on dealing
with events and people which are experienced only in terms of specific
linguistic versions. In a profound sense, then, discourse can be said to
construct our lived reality (Potter, Stringer and Wetherell, 1984).
Several specific analytic practices have emerged from our

discussion so far. Firstly, and fairly obviously, we are suggesting that
close attention is paid to the detail of language use, working from

transcripts or documents rather than from (in the social psychological
tradition) some numerical transformation of these things or (in microsociology) the remembered gist of interaction. Secondly, this approach
suggests that discourse itself becomes the primary research focus. It is
not a subsidiary path to the true nature of events, beliefs and cognitive
processes. Thirdly, we propose that discourse analysis depends on the
study of variation, working from the way discourse is constructed to
some conclusions about the functions discourse might be serving.
Moving on from these points we come to our final analytic tool: the
interpretative repertoire. We suggested earlier that discourse is
variable in the sense that any one speaker will construct events and
persons in different ways according to function. This is not to imply
that there is no regularity at all in discourse - simply that regularity
cannot be pinned at the level of the individual speaker. There is

regularity in the variation. Inconsistencies and differences in discourse
are differences between relatively internally consistent, bounded
language units which we have called, following Gilbert and Mulkay
(1984), interpretative repertoires (Potter and Mulkay, 1982; Potter and
Reicher, 1987; Potter and Wetherell, 1987; Wetherell, 1986; Yearley,
1985).
Once again, repertoires are best described in relation to a particular

empirical example. However, some points should be noted here in
abstract. In dealing with lay explanations the analyst often wishes to
describe the explanatory resources to which speakers have access and
to make interpretations about patterns in the content of the material.
The interpretative repertoire is a summary unit at this level.
Repertoires can be seen as the building blocks speakers use for
constructing versions of actions, cognitive processes and other

phenomena. Any particular repertoire is constituted out of a restricted
range of terms used in a specific stylistic and grammatical fashion.
Commonly these terms are derived from one or more key metaphors
and the presence of a repertoire will often be signalled by certain
tropes or figures of speech.
The community talk mentioned earlier is, in fact, a classic

example of a repertoire in this sense. It is a set of terms used in
explanations which depend crucially on certain metaphors and tropes.
The community repertoire is a resource which the analyst can
identify as a recurring pattern in the content of certain materials. Our
empirical example examines repertoires in a quite different domain the construction of models for race relations.

Empirical example: the construction of race
relations
Background
The study which we will focus on is taken from an extended project on

racism in New Zealand. This project was concerned with how the
majority group of white European New Zealanders understand issues
like inequalities of employment opportunities and how they make
sense of their relationship with the indigenous Maori people.
It is not necessary, fortunately, to be familiar with the theoretical

background to this work or with the New Zealand context to appreciate
our study as an example of discourse analysis at work. Suffice it to say
here that our project developed as a way of looking at racial prejudice
that went beyond social cognition, attitude socialization and
personality (Wetherell and Potter, 1986a, 1986b; Wetherell and
Potter, forthcoming; cf. Billig, 1987b; Reeves, 1983; Van Dijk, 1984).
We were interested in how ordinary, mainstream white New
Zealanders would describe their past, present and future relations with
the minority Maori group, and our concern was with the ideological
consequences of their discourse. We wanted to look at how our
samples practical reasoning about race might justify and work to
maintain asymmetrical power relations between the majority and
minority groups, rationalizing and naturalizing a certain kind of status
quo (Giddens, 1979; Thompson, 1984). Needless to say, when
discourse analysis is used with these kinds of goals it must be
combined with a careful analysis of the particular intergroup situation
in question. Discursive patterns must be located within an account of
their wider context.
We conducted interviews with 40 women and 41 males, of varying

political affiliation, relatively evenly spread between National (rightwing) and Labour (more left-wing) voters, and covering most age
ranges from 18 years upwards. Those interviewed could be described
as middle class, by education in the case of the young people, by
socioeconomic circumstance in the case of the non-working women
and men, or by their professional or managerial occupation for the
remainder. This is a relatively large sample by discourse analytic
standards, chosen because we were particularly interested in the
generality of our conclusions across a wide group.
Interviews were designed to elicit extended sequences of talk on a

number of topics. They focused especially on perceived causes of the
disadvantaged position of the Maori people, recent multicultural social
policies, proposed reasons for racism and tension, and special
provision or positive discrimination for ethnic groups. All interviews
were tape recorded and fully transcribed. The general procedural

details of interviewing and transcription are discussed at length in
Potter and Wetherell (1987).
Variation
Let us first use some of our data to illustrate variation in discourse

before presenting our method and findings with regard to interpretative
repertoires. We argued that variation in individuals accounts was an
essential feature of natural language use, and an important route to
understanding function. In terms of this project, a contrast could be
made between the expectation of attitude theory, that peoples
statements will reflect an underlying attitude which is either
consistently prejudiced or tolerant, and the discourse analysis claim
that people use language, including attitude-type statements,
functionally and thus variably as the discursive context changes.
The extracts below have been grouped under two categories: A and
B. The first extract under A is an anecdote about the speakers bible
class where an incident is described and the point drawn that children
internalize racist attitudes from their parents. In the second extract
under A the same speaker gives her view on racist jokes,
commensurate with her point of view about racist remarks in general.
Finally, under A, this speaker indicates the aspects of Maori culture she
admires. Overall this speaker can be heard to feel strongly about
racism and positive about Maori culture. If so inclined, we might
characterize her as a liberal non-prejudiced person.
Extract A
(1) I do this bible class at the moment, not highly religious, I just

think children ought to know about religion ... and last night we
were just discussing one of the commandments, love your
neighbour, and I had this child who said What would happen if
you got a whole load of Maoris living next door to you? and I
said to him Thats a very racist remark and I dont like it, and he
shut up in about five seconds and went quite red in the face, and I
realized afterwards that obviously it wasnt his fault he was,
turned out to be thinking like that, it came directly from his
parents.
(2) [Racist jokes] I dont like them I dont find them amusing.
(3) [What can we learn from Maori culture?] The extended family
situations brilliant, theyve got this lovely idea that a child born
out of wedlock would have to be the best sort of child because it
was obviously born in love ... I think their way with children is
wonderful. ... Theyve got a lot to show us I think.
The three extracts under B present a different picture. The speaker here
seems more willing to attribute negative characteristics to groups and
unwilling to accommodate Maori culture. In the first extract, it is
suggested that Australians are handling their intergroup relations in an
unfortunate way - letting other groups get on top of them because of
their inherent laziness. The second extract puts forward the view that
because white British settlers conquered the Maoris who, in their turn,
had conquered the Maorioris, Europeans have licence to define their
own terms in New Zealand. And, then, the speaker suggests in the third
extract that Maoris must accommodate to European society, although it
is recognized that being the indigenous group they cant be repatriated
in any sense unlike other Polynesian groups who are immigrants to
New Zealand.
Extract B
(4) The Greeks live in one part of Sydney, all the such and such, and
theyre all growing up and speaking their own language and doing
everything, theyre going to have all these groups, and the
Australians are basically a lazy people, and that other cultures are
getting on top of them, theres going to be big problems there one
day.
(5) The ridiculous thing is that, if you really want to be nasty about it,
and go back, um, the Europeans really did take over New Zealand
shore, and I mean that Maoris killed off the Maorioris beforehand,
I mean it wasnt exactly their land to start with, I mean its a bit
ridiculous. I think we bend over backwards a bit too much.
(6) And this is the part that I think is wrong with (.) a bit wrong with
the Maoris as well there, the problems they have, theyre not
willing, I mean its a European society here and theyve got to
learn to mix and get in and work, otherwise its, I mean you cant
tell them to go back to where they came from.
In fact, both A and B extracts come from an interview with the same
speaker, whom we will call Benton. The pattern one finds in these
extracts was common in our interviews in the sense that, as we have
argued, people construct different versions depending on the functional
context. Over an entire interview it is exceedingly difficult to
summarize their views. Indeed, from the perspective of attitude theory,
how would one describe Bentons underlying attitude? Is this a
prejudiced person? A tolerant person? A person likely to be in favour
of multiculturalism and/or anti-racism, and against attributing negative
qualities to people on the grounds of race or nationality? Or is this a
person who supports the maintenance of white cultural hegemony?
We see Benton, then, drawing on different, often inconsistent

resources, as they seem appropriate. She is not an isolated example.
The same complexity is apparent in the next extract, taken from
another speaker whom we will call Anna James.

(7) I think it [multicultural social policy] is a good thing, cos I said
before, I think people need to know their roots and their culture
and that sort of thing. But I think weve got to be really careful
that we dont go overboard in that at this stage. Thereve been a lo
... a few injustices against the Maoris over the years, thats going
to happen with any sort of racial minority in a society, urn but I
think at this stage weve got to be careful not to go overboard um,
I think weve got to be careful not to force the white children to
learn Maori if they dont want to, its not the white childrens
heritage. Now, okay, you give the opportunities, I mean they have
the Maori playschool, Maori language playschools, and I think
thats really good for Maori kids. But I dont want my kid to learn
Maori, I dont want my kid to learn about grass skirts and ... I
want my kid to learn about the history of the British Isles and the
history of Europe which is incredibly important to the formation
of their ideas about, you know, society. Okay, um, I think we
should understand how Maori (.) uh, culture works, I mean I
know very little about it, I know the basics, and I would like to
see my kids do a daytrip to a marae, and see how they... but I
really think weve got to be incredibly careful about pushing it
down peoples throats, particularly the white peoples throats.
This extract can be seen to be organized around a particular kind of

functional feature - namely a disclaimer. A disclaimer is a verbal
device designed to ward off potentially obnoxious attributions (Hewitt
and Stokes, 1975). Thus if someone prefaces their remarks with Im
no sexist but... , the listener can be relatively confident that obnoxious
remarks about women will follow but the speaker is wishing to head
off, or disclaim, the possible implication that they are sexist.
Disclaimers create a form of variation which presents particular

problems for the realist model of discourse and the assumption that the
individual actor can be assumed to be a coherent consistent starting
point for analysis. In the case of this extract Anna James first
establishes that she is not one of those people opposed to Maori
culture, but then goes on to give arguments explaining why her child
should not be exposed to multicultural education. Her views on
multicultural social policy are clearly qualified. Now, of course,

peoples views are frequently qualified in this way. Something is a
good thing provided certain conditions are met, and then it becomes
unclear whether it is a good thing or not. But how could this complex
view be summarized and put into a response category? Simplistically,
does this persons utterance support multicultural education policies or
not? She is clearly not indifferent to the issue, or without a view, and
thus in the middle of some response scale.
These extracts, therefore, indicate the problem of variability. They

demonstrate very clearly some of the difficulties involved in working
with natural language. How are we going to proceed with our analysis
of these data? As we noted earlier, discourse analysis attempts to study
variations in content to work towards an understanding of function. By
studying the resources from which an account is constructed, we can
also investigate what it might achieve. We noted earlier that

differences in content reflect differences between interpretative
repertoires. If we look, therefore, for regularities at this level, in
language, and abandon the individual as our principal unit of analysis,
then progress might be made in understanding the complex
inconsistencies in the discourse centred around race relations.
Interpretative repertoires
The first goal in a study of this kind is to perform some preliminary

coding and thus sift out a manageable subset of data from hundreds of
pages of transcript. We selected out from the interviews all passages of
talk relating to our topic of models of race relations. This included all
material relating to the respective places of Maori and European
culture in New Zealand - issues such as integration versus separate
development or assimilation, the teaching of Maori language, reactions
to attempts at a renaissance in Maori culture and so on. The topics

developed in the interview questions were set by the structure of
contemporary debates and the general agendas of the main political
parties.
As an aside, it is important to stress here that the analysis which

follows is in no way intended as a complete presentation of our
conclusions for this part of our data. The aim is simply to use some
material as an example, to illustrate our method, and not to give a
definite picture of the repertoires articulated or of the ideological
implications of white New Zealanders analysis of the path they
suggest relations with the Maori people should follow.
Our coding policy at this stage of discourse analysis is usually an

inclusive one, accepting all borderline and anomalous cases, and the
end product is a file of photocopies of the original transcript. It is at
this point that analysis proper begins, with careful repeated readings of
the materials in a search for patterns and recurring organizations. This
process is not a matter of following rules and recipes; it often involves
following up hunches and the development of tentative interpretative
schemes which may need to be abandoned and revised over and over
again. Discourse analysis is a time consuming and laborious business,
with the search for regular pattern giving way to the formation of
hypotheses about that pattern of repertoire use.
From this process a number of different repertoires were identified

for our general topic - too many to be discussed here. However, three
particular dominant repertoires can be picked out: we have labelled
these culture fostering, pragmatic realism, and togetherness (Wetherell

and Potter, 1986b). It is worth spending a bit of time clarifying the
overall pattern of use of these repertoires and the grounds for
identifying them as different, before going on to the details of their
content.
Culture fostering was used by over 90 per cent of the respondents,

while pragmatic realism and togetherness were drawn on by about half
the respondents. So the most common pattern of accounting was
culture fostering combined with either pragmatic realism or
togetherness, although an appreciable number of respondents - perhaps
10 per cent - combined all three. The immediate point, then, is that
there is no sense in which we could have divided our respondents into
three classes, each distinguished by a different pattern of belief. Each
respondent selectively combined different repertoires. Indeed, it may
be that if we had asked a few more questions all respondents would

have drawn on all three repertoires. The combination of repertoires
produces the kind of complex and varied versions we noted in the
extracts considered earlier.
What grounds have we for defining three different interpretative

repertoires here? In this analysis we used three central kinds of
evidence for this. Firstly, as we will show, there are inconsistencies noticeable to both analysts and participants - between the different
forms of account. Secondly, these forms of account are generally
separated into different passages of talk so that inconsistencies do not
become a problem for participants to deal with. Thirdly, on these
occasions when the different repertoires are deployed together,
participants display in their talk an orientation to the potential
inconsistencies, or the variation is organized for different functions -
one repertoire presented for disclaiming, for example.
Let us now look in a bit more detail at the makeup of these

repertoires. Culture fostering presents arguments for the development
of Maori culture. It appears to advocate multiculturalist social policy
and the importance of Maori culture for New Zealand society. There
are two major facets to it. On the one hand, it presents the view that
Maori culture should be encouraged, fostered, protected and conserved
because it uniquely and distinctively identifies New Zealand, and is a
worthwhile culture in itself. For example:
(8) Im certainly in favour of a bit of Maoritanga it is something
uniquely New Zealand. I guess Im very conservation minded and
in the same way as I dont like seeing a species go out of
existence I dont like seeing a culture and a language and
everything else fade out. (Shell)
On the other hand, culture fostering presents the view that it is

important for the Maori people to have a sense of identity and history
or roots, in the way it is considered important for every person to have
a sense of identity or place. For example:
(9) I think the sort of Maori renaissance, the Maoritanga, is important
because like I was explaining about being at that party on
Saturday night, I suddenly didnt know where I was, I had lost my
identity.... I think it is necessary for people to get it [Maori
identity ] back because its something deep rooted inside you.
(Reed)
Culture fostering here is seen as positively compensating for what is

viewed as a deficit or a weakness within Maoris. Formerly Maoris
tended to be seen as deficient in relation to European culture, in need
of European enculturation and civilizing influence. In modern liberal
New Zealand such an obviously white supremacist view is less
acceptable, but this form of accounting retains the notion of deficit;

specifically Maoris are seen as deficient as Maoris, and therefore now
in need of Maori enculturation (Nash, 1983).
What function in the broader sense might this deficit notion achieve
for those who use it? Firstly, it seems to make sense of another
commonplace understanding that Maoris have a deprived social
position and are discontented, through using the idea of rootlessness
and loss of identity. In this lay sociology, people without roots - those
who have lost their identity in some way - do not perform well and
are likely to agitate. Secondly, in using the notion of cultural deficit
speakers can effectively place Maori problems elsewhere, removed
from their own responsibilities and actions. In this way speakers can
convey that they themselves are in no way to blame for these
problems.
One of the consequences of this form of talk is that multiculturalist

social policy is advocated without requiring reciprocal change in the
majority European group. The development of Maori culture becomes
a matter for Maoris only, assisted by Europeans who merely clear a
space, as it were, for this development, as a mother might clear a space
for a child to play, knowing that play is good for the mental health of
the child. Respondents virtually never characterized the inclusion of
Maori culture in a way which involved active effort or change on their
own part; effort and change was depicted as a Maori problem and duty.
We hypothesize that this is a repertoire used to positive liberal effect
but which locates the problem outside the responsibility of the white
majority group.
The second repertoire used in discussions of race relations

strategies is pragmatic realism. Pragmatic realism, used by roughly
half the sample at some point in the interview, and thus including
many of those who also draw on the culture fostering repertoire,

stresses the promotion of those things which are useful, modern and
relevant today. It combines with this an emphasis and appreciation of
the practical constraints on action. For example:
(10) I actually object to um them bringing um massive Maori culture
curricula into schools etcetera ... because I do feel that this doesnt
equip them for the modern world at all. Because whats the use of
being able to speak Maori if you can only speak it to a limited
number of people in a limited area and it has no use at all in the
actual, you know, in the real world as it were, if youll pardon the
expression. (Bradman)
The modern world referred to in accounts of this kind is predominantly

defined in Western terms. In the next extract, the same point of view is
applied to learning Maori language:

(11) You know this is a side line, you cant go back. I mean you can go
back but you are not going to live in this day and age. Id far
sooner see them all learning Japanese so they can say Ah so
when all the visitors come. (Sedge)
When using this repertoire, it becomes incomprehensible to turn the

clock backwards. The idea is that on sheer pragmatic grounds there is
little to be gained from Maoritanga, for Maori cultural practices are
unrealistic or impractical.
This is a very different, even contradictory, repertoire from that of

culture fostering. Its principal idea is that much of Maori culture is
antiquated and should be quickly abandoned. This apparent
contradiction, as we noted, does not stop the interviewees drawing
repeatedly on both repertoires.
As part of this kind of talk, speakers often suggest that introducing

Maori language is impractical because most Maoris no longer speak
Maori. There is a depressing irony about the members of a group who
have done their best in the past to suppress an indigenous culture,
arguing at this stage that the success of their suppression is good
grounds for continuing to suppress it. Pragmatic realism is a particularly flexible repertoire. We suggest that its effect is to enable speakers
to present constraints which are beyond their control, and which they
can thus regret, but at the same time to construct themselves as people
who are at least realistic and practical.
The third repertoire, which we have called togetherness, is the most

interesting in many ways. It articulates a familiar position and is the
notion that there should be no divisions or barriers between people: we
should all be one together. People should be treated as people, not in
terms of their colour or cultural background.
(12) I wish that we could stop thinking about Maori and European and
think about New Zealanders (Int: Mmhm.) and to hell with what
colour people are. (Dixon)
Or:
(13) I think its important that we recognize that we are in fact all New
Zealanders and we should be tending to become more one rather
than separately developing. Were tending to... pull that part of
our culture to almost a sense of importance that I dont think it
really has. We are one people, despite history. (Barr)
On the face of it this seems a highly positive and caring approach, and
indeed in psychological terms it is probably well intentioned.
However, the implications of applying it in this context (as over half
the respondents did) are not so positive. European culture is the
dominant culture in New Zealand; it thus sets the normative
framework for what it means to be one people together. Put another
way, for these people New Zealanders are often depicted as basically
white Europeans and the divisions between people or barriers objected
to are those created by the legitimate claims of the Maori people. The
upshot of this form of talk is that Maoris should stop encouraging rifts
and conflict and accommodate to dominant European values; however,
this consequence is hidden behind the innocuous moral formula of
togetherness.
Overall, then, respondents talk around this general topic is made
up from a combination of repertoires (and we have only discussed

three of these) which produce a complex and potentially inconsistent
or, at least, variable response. We have concentrated in this analysis on
briefly speculating about the broader ideological consequences of these
repertoires, although the functions they achieve at the level of the
localized discursive context could also be considered through the study
of variability.
In general these speakers could not easily be described as

oppressive or racist; indeed, on a superficial hearing they may come
over as well-meaning and sympathetic. However, we would want to
make a strong distinction between the psychological motivations for
using these discursive forms and their social psychological
consequences.
Well-intentioned talk can have reactionary consequences, as we have

tried to demonstrate. For this reason also, we reject psychologically
reductive theories of racism (Potter and Wetherell, 1987; Wetherell and
Potter, 1986a, 1986b; forthcoming).
Each repertoire is relatively innocuous in abstract and possesses its

own morally virtuous self-presentation, yet the potentially disruptive
force of Maori protest and anti-racism is safely contained. Put at its
simplest, the critic can be silenced by culture fostering, undermined by
pragmatic realism and in addition accused of creating barriers between
people. It is only when looking at the organization of explanations in
the discourse as a whole that the fragmented and inconsistent nature of
the talk becomes apparent. We would suggest that this flexibility in
articulation is crucially important to ideological effectiveness
(Wetherell, Stiven and Potter, 1987).

There are two areas of potential disadvantages with discourse analytic

work of this kind. Firstly, there is the sheer effort involved. The time
taken up by conducting a set of interviews and then fully transcribing
them is considerable, and few people find transcription pleasurable although there is nothing better for encouraging close attention to what
people are saying. Analysis is a craft skill which takes time to develop
and is slow to conduct. It is quite possible to follow up an analytic
scheme for some days only to find that it is simply impossible to
validate with the materials available.
Secondly, work of this kind is not suited to the production of the

kind of broad empirical laws which are commonly the goal of social
psychological research. Unlike some social cognition researchers and
personality theorists, we do not claim to have found a universal
psychological process underlying racism. On the contrary, our findings
are specific to a particular class of New Zealanders at a particular point
in New Zealand history, and we would not necessarily expect white
British majority group members, say, to account for racial inequalities
using the same interpretative repertoires. Nevertheless, we see this not
so much as an inadequacy of the approach but as a consequence of the
fact that explanations are always fitted to specific occasions and
constructed out of the available interpretative resources.
A further disadvantage arises from the novelty of discourse analysis.

This is not yet a tried and tested approach elaborated and honed in
many empirical settings. Clearly further theoretical work is required.

The notion of function, for example, needs further elucidation. This
kind of elaboration, however, is best
carried out in relation to particular empirical tasks and with data rather
than in abstract. Undoubtedly discourse analysis will revise its
operating base as its application grows.
Offset against these disadvantages we see certain crucial advantages

of the discourse analytic approach. Firstly, the approach is intended to
do justice to the subtlety and complexity of lay explanations as they
are deployed in natural contexts. We have not had to constrain the
participants response options to obtain usable data; indeed, we could
have carried out a similar analysis to the one above on purely
naturalistic records of everyday conversations or newspaper articles
(Potter and Reicher, 1987). And we are not treating lay explanations as
a degraded form of some underlying schema or cognitive process; in
discourse analysis the language used is a constitutive part of the
explanation, not a medium that the explanation is translated into.
Secondly, although discourse analysis is sensitive to linguistic

nuances and the kind of contextually sensitive features of talk which
are extremely difficult to recover using traditional content analysis, it
is a systematic approach whose findings are open to evaluation.

Elsewhere we have discussed in detail the techniques through which
the conclusions of discourse analysis can be validated (Potter and
Wetherell, 1987); one of the most important of these is the report of the
research itself. Studies of this kind include a representative set of
extracts along with detailed interpretations which link the analytic
points to specific features of extracts in such a way that the reader is
able to assess the success of interpretations and, if necessary, offer
alternatives. The overall goal is to openly present the entire reasoning
process from data to conclusions.
1
3
Finally, we would argue that the results of
analysis of this type are both interesting and
potentially useful because they focus on both the
specific linguistic content and the organization of
lay explanations. When dealing with racist
explanations, this means that we come to
understand the various interpretative repertoires
through which racist explanations are
constructed and warranted, and can start to
understand the techniques through which these
explanations can be undermined and
transformed.Deconstructing accounts
Ian Parker
1
3 techniques developed in literary
In this chapter I want to show how
theory can be applied to everyday explanation. However, as you will

see when we come to the empirical example, the deconstruction of a
piece of text involves a radical overturning of traditional social
scientific distinctions between what is an everyday, spontaneous
account and what is an extraordinary, manufactured item of script. The
approach I will focus on, that of deconstruction, extends the critiques
of positivism and individualism in traditional social psychology,
reformulates our notions of what constitutes lay explanation and reemphasizes the part power plays in social life.
1
3
The emergence of a deconstructive approach to social psychology

should be understood in the context of attempts to provoke a paradigm
shift in the discipline over a decade ago. At the time, critics of
traditional laboratory-experimental social psychology hoped to
reconstruct the discipline out of the ruins of the crisis-ridden old
paradigm (Armistead, 1974; Harre and Secord, 1972).
518 Ian Parker
The crisis in social psychology developed primarily out of a claim

that most published research was trivial, presented a mechanistic
model of the human being, and failed to engage meaningfully with real
issues and experiences. We have to be aware, though, that other (as yet
unresolved) problems played a role. Most important was the question
of power: power as a pervasive quality of social interaction
(highlighted in the classic obedience and conformity experiments),
power as a profound influence on laboratory results (demonstrated in
the work on experimenter effects and demand characteristics), and
power inherent in the situation of a privileged researcher vis-a-vis his
(and sometimes her) subjects. The phenomenon ofpower is a difficult
one to conceptualize without appreciating the ideological status of
social psychology in our culture. In turn, to understand the work of
ideology we have to appreciate the power of language, which is where
new paradigm social psychology really made its mark.
Deconstructing accounts 519
New social psychology
Language, the proponents of the new paradigm argued, was the key to
understanding social life. Psychology mistakenly viewed itself as a
natural science, and, on top of that, it had a mistaken view of how the
natural sciences obtained knowledge of their objects. Instead,
psychology (and especially social psychology) should recognize that
the pre-eminent role of language, of meaning, in social action
necessitates a turn to the human sciences, and that a more adequate
(realist) model of inquiry could be adopted. There are, of course, two
lines of attack here. The first stresses the difference between natural
science and human science and tries to shift psychology from one to
the other. The second stresses the value of methods actually employed
in the natural sciences and tries to bring psychology up to date. There
is a creative tension between the two positions advanced in the new
paradigm. At the risk of caricaturing them and glossing over their
520 Ian Parker
points of convergence, it is useful to identify hermeneutic and

structuralist strands.
The hermeneutic strand focuses on the ability of the social actor to

produce meaning, to construct new definitions of a situation and to
communicate intentions and the import of action to others. Social life
is constituted by the accounts which we must give one to the other to
maintain the peculiarly human quality of the world. Inquiry into the
life world of social actors requires an empathetic involvement in, and
elaboration of, the hermeneutic circle of meaning. In this view, then,
the formal properties of interaction are no more than interpretations,
constructions of an interested observer, and have no other existence
(Shotter, 1975,1983,1984).
The structuralist strand, in contrast, makes a determined effort to

uncover the patterns of interaction, the structures which inform the
activities of social actors. Accounts are gathered, negotiated, with
those being studied because the accounts and actions have the same
source. Accounts are used, then, to reconstruct a social world of which
each individual actor has an imperfect, fragmentary knowledge. In this
way, the researcher takes note of the explanations of action given by
people and arrives at a deeper understanding, following the methods of
realist science, of the formal properties of the chosen social world
(Harre, 1979, 1983b; Marsh, Rosser and Harre, 1974).
The debate between the two sides of the new paradigm echoes longstanding and unresolved conflicts in social psychology between
agency and structure (Harre, 1983a; Shotter, 1980). What both
tendencies in the new paradigm have neglected, however, is the way
interpretation, conflict and resistance in social life cry out for an
522 Ian Parker
analysis of meaning in terms of power. We are able to put these issues

to the forefront of our research if we adopt a deconstructive approach
to texts, and move beyond the dichotomy between structuralism and
hermeneutics. Deconstruction uncovers the way language, and the
analysis of the texts we weave with language, works.
Deconstruction
While the arguments over the nature of the paradigm shift which
fuelled the crisis inside social psychology were burning, intense
debates were also raging in the other human sciences. These debates,
which filtered through into some of the formulations of our own new
paradigm critics, revolved around the value of structuralism and poststructuralism as ways of understanding language. In literary theory,
where the conflict between the different positions was most intense,
the transition from structuralism to poststructuralism also saw the
emergence of a deconstructive approach to texts (Eagleton, 1983).

There is some value in attending to these debates because we can then
use them to extend the work of the new paradigm writers in social
psychology, arrive at a more productive understanding of the nature of
texts, and gain a better understanding of the relationship between
explanations and power. In order to appreciate the value of
deconstruction in social psychology, we have first to deconstruct
structuralist and hermeneutic approaches.
Language, according to structuralism, should be understood as a

system - a system in which each item, or word, gains its meaning only
by way of its difference from the other items (Saussure, 1974).
Intentions we express are immediately caught up in the web, system,
structure of language and escape our grasp. The structuralist approach
to language, then, sets up an opposition between the system of the
language as a whole (a structure which would be uncovered by an
524 Ian Parker
investigator) and the use of that language by individuals (the day-today occasions when the words are transmitted). The first is studied and
the second studiously ignored by a structuralist.
The value of structuralism, for literary theorists, was that it directed

attention to the underlying pattern of a piece of writing. Whereas the
old interpretative tradition had attempted to trace the meanings in the
text to the intentions of the author, the structuralist approach was able
to produce a rigorous explanation of the way the text worked. At an
extreme, such an approach was supposed to involve the conceptual
death of the author (Barthes, 1977). While the conclusions they drew
about the irrelevance of individual intention and agency were not, of
course, shared by social psychologists interested in structuralism in the
new paradigm, the model of language they adopted was taken up.
However, if you pursue that argument you will eventually arrive at

the breakdown of strictly structuralist notions and their collapse into
post-structuralism. It is, in short, possible to deconstruct the opposition
between the structure of language and its use. Each person constructs
an imperfect, fragmentary, idiosyncratic version of the structure which
affects the relation between the words each time she or he
communicates to another person. In addition, each act of interpretation
requires the reproduction of the system of language and, crucially, the
production of new metaphors which disrupt the system (Derrida,
1981).
The lesson I as a social psychologist take from these developments

is that any text I would want to examine is open to further
interpretation. The meanings in a text shift from one moment to the
next, and according to its wider context. A reading of the text, then, is
provisional, and is profoundly affected by my own and your prior
526 Ian Parker
understanding and expectations (themselves aspects of experience

which are impossible to definitively fix as if they were confounding
variables). The most that can be done is to persuade you that my
interpretation identifies what is going on.
The obvious escape route from the relativism into which these
debates seem to be dragging us has already, of course, been closed off.
We cannot ask the authors what they really meant. But what of the
authors of social texts such as speeches, conversations and
explanations? Here, the development of deconstruction in literary
theory is particularly useful. It places into question the status of
intention in the production of language.
An obvious, and appealing, way of marking the difference between

literary theory and social psychology is to say that there is a crucial
distinction to be drawn between writing and speech. Writing is
constructed in such a way as to meet the expectations of a number of
readers, and it can be interpreted in many ways as time flows on and
as it becomes progressively detached from the experience and memory
of the author. We believe that speech is different; it expresses thought,
and as part of the accountability of conduct it is immediate and more
spontaneous. The experience of speaking is one which leads us to
imagine that speech flows more directly from what we really want to
mean.Speech, though, is constructed so as to be a communication to
others. The language used as the medium of communication both
facilitates and moulds what we say. We cannot invent anew terms to
express our intentions. On the contrary, we find ourselves using words
which have been repeated many times before. Old words in new
contexts obtain their meaning from their identity with the way they
were the last time they were used, and their difference from the way
they were used before. Neither the language nor the context can be
controlled by, or traced to, the intention of an individual. Is not speech,
in fact, a variety of writing? This is the conclusion we might arrive at
when we deconstruct the opposition between writing and speech. In
turn, this warrants our use of the term text to denote both written and
spoken forms of language production (Derrida, 1978).
528 Ian Parker
As far as social psychology is concerned, the focus on texts as our

objects of inquiry should enable us to produce interpretations which
are intrinsically social. Our deconstruction of an item of text is a way
of arriving at interpretations which are locked into the surrounding
cultural context. I take this as an opportunity to raise general issues
about the way texts re-place us in positions of power. Each specific
deconstruction is illustrative of, and gives more leverage to, a
deconstruction of social relations.
I must confess that I am loath to formalize the deconstruction of

theoretical positions into an abstract method which can be applied to
non-theoretical texts. Strictly speaking, deconstruction is not a
method in the sense of being a set of techniques (Derrida, 1981).
However, it is possible to summarize three steps or stages to a
deconstruction (Wood, 1979).
The first step is to identify conceptual oppositions, or polarities, in

the text, and the way one pole is privileged over the other. In the case
of structuralism the key dichotomy was between structure and use. In
the case of hermeneutics I pointed to the opposition between speech
and writing. In the course of this chapter, I want to draw attention to
further conceptual oppositions: between what is spontaneous and what
is scripted; between what is real and what is fictive; and between what
is to be counted as everyday and what as theoretical explanation. In
the empirical example, I will draw attention to further oppositions
which will serve as devices to highlight processes at work in the text:
between metanarrative and narrative; between reflection and
subjectivity; and between recasting and casting.
The second step is to demonstrate that the privileged pole of the

opposition is dependent on, could not operate without, the other. We
bring the hitherto subordinate concept to the fore. In the case of
530 Ian Parker
structuralism, then, the use of language profoundly affects the

structure, thus making capture of the structure of a language
impossible. In the case of hermeneutics, writing is brought forward as
a model of speech, making invalid the traditional linking of speech and
intention. The consequences of a deconstruction involve, I hope to
show as we proceed, a re-emphasis in social psychology on scripted,
fictitious and theoretical items of text over what we currently
understand as being spontaneous, real or everyday forms of
explanation. We will see, when we come to the empirical example, the
importance of narrative, subjectivity and casting in the production of a
text by social actors.
The third step involves a reinterpretation of the opposition, and the

production of new concepts. In the cases of structuralism and
hermeneutics, we arrived (implicitly) at the notions of text and
deconstruction. The third step, however, requires an intervention into
the production of the text (Derrida, 1981). For social psychologists,

this calls for the production of new practices (and new problems for
those wanting to stick to the text, as we shall see).
A deconstruction of a piece of text carries with it assumptions about

the nature of meaning, and it produces new meanings out of the texts it
contacts. It changes what it studies. In addition, the way I read a text is
informed by the moral/political structure of the interpretation (the
scene, which in this case is an academic paper), and the experiences I
bring to it (which result from my participation in the production of
other academic and non-academic texts). These factors give power to
an interpretation, and may boil down to relations of power between
those who are permitted to construct real theories and those who are
deemed to produce everyday explanation.
532 Ian Parker

Empirical example
For our text we have an item of script in which the social actors
contest the nature of the distinction between reality and fiction, and the
power of theory over everyday life. We will see one character attempt
to construct a metanarrative and implicitly reflect and recast the other
within it. We will also address the issue of implicit explanation.
The nature of text
The idea Of text which lies at the heart of deconstruction, and which
emerges from the wreckage of structuralist and hermeneutic styles of
new paradigm social psychology, has a number of useful and exciting
qualities. Any text - be it a speech, a monologue or a treatise - consists
of a number of warring themes. The conflicts between the poles of any
conceptual oppositions that we might uncover need not necessarily be
identified with conflicts between persons. In fact we would expect that
where a number of characters congregate to debate differences, each
person would slip from one position to the next, exploiting ambiguities
and loopholes in the others argument. However, the stuff of everyday
explanation is in the interchange of opinion. Most of the talk that
people engage-in is in the presence of others. We will, then, examine a
text which is wrought through dialogue. Here, the different positions
adopted will be easier to define and explore.
A notion we may draw from the new paradigm is that the utterances
534 Ian Parker
of social actors have an overwhelmingly expressive function (Harre,

1979; Shotter, 1984). When social life is viewed as self-presentation
through ingeniously constructed rhetorics, or it slips into well-worn
rituals, we are right to use the notion of a script to highlight what is
happening. Once we have deconstructed the distinction between
speech and writing, however, we can go further than this. We then
have a licence to turn to scripts themselves. Written scripts need be no
less spontaneous than real speech. The text in this example once had
the status of a script. The objection might be made that such a text is
not real; it is fiction. Some fictions, however, are more real than
others. The explanations we give to others - the texts we pass on - are
sometimes mistaken, sometimes half-truths, occasionally deliberate
lies. Some items of fiction are lived by large numbers of people in a
culture, and the meanings are discussed and contested. They have
effects as real as ostensibly unscripted non-fictitious explanations.
They operate as an ideology.
Furthermore, a problem with much research in social psychology is

that it illuminates a social world we do not implicitly know. Readers of
such reports can only either take what the research says on trust or
retreat into a cynical suspicion of all accounts. A deconstruction
should be open to revision. An interpretation should be reopened as
times and contexts change. Readers should have some space at the
margins of even an academic text.
Sample text
We do know the social worlds of soaps. Sociologists have turned in

recent years to analyse the way audiences receive these worlds and
incorporate the narratives into their own worlds. In the case of
American television series, the preoccupations and antics of the
characters become templates for the understanding of the non-soap
world, metaphors for the lives of politicians, and frames to organize
536 Ian Parker
explanations. The advantage the soap has over discrete narratives is

that actors develop a history and are attributed with personality traits
and intentions.
Although the most popular television soaps have the advantage of

having a large audience in many countries, their status as texts
presents problems for a deconstructive analysis. There are the
meanings of clothes, scenes and locations which could be treated by a
semiological analysis. There are also the non-verbal cues which
supplement, organize and occasionally contradict the spoken message.
For these reasons I have taken a piece of text from an (English) radio
soap.
The Archers, which has been running on Radio 4 since 1950, has
developed a minor cult following in the past few years. Radio 4 is the
serious channel of the BBC. From being a fairly traditional serial,
The Archers has roused controversy with the introduction of social
issues. Letters to Radio 4 news programmes and The Guardian (a
liberal middle-class newspaper) discuss the fate of characters and the
corruption of the series by the unpleasant practices and vices which
feed the storylines of the American soaps. Books about the characters,
faked local newspapers and maps of Ambridge (the fictitious village in
middle England where the action is set) are starting to appear.
The text was transcribed from a 2 minute 20 second scene broadcast

on 22 October, 23 October and again on 26 October 1986. I have
screened out the paraverbal paraphernalia. This, of course, excludes
the new and different meanings put into the original written script by
the actors. This is a limitation because of the non-correspondence
538 Ian Parker
between the transcript and the original. However, the script-writer may
have spoken the original text into a dictaphone with quite different
intonation, and a complicated notation for emphases and pauses would
not solve the problem; it would merely displace it.
The scene was not selected at random. I chose it because it

illustrates some current concerns in the series which connect with
those of its target audience. It also raises issues about social relations,
in particular gender relations, in the non-soap world. Finally, it
highlights a problem to do with interpretation - the interpretation we
are permitted to construct of the real meaning of a persons actions,
and the interpretations social actors give of one anothers acts. The
distinction between a theoretical form of explanation and a lay form of
explanation is one which is bound up with power. The distinction
gives leverage to some social actors in positions who are permitted to
have access to knowledge. For those in other social positions, lack of
knowledge can have the effect of reducing them to silence.
An immediate problem is that the temptation to intervene and flesh

out the context is almost too much to bear. As with all data collection,
the researcher can undermine the supposed objectivity of the process
by supplementing the data with personal knowledge (It was like this, I
was there!). Perhaps I need only allude to the structural class relations
that cast Clarrie as the wife of the son of a poor tenant farmer, and Pat
as a member of the Archer family which runs the estate. Those who
are outside the community of listeners will not know that Clarries
own father works as a labourer for the Archer family. Would it be
helpful to suggest that Pats interest in organic farming also helps
maintain her own marginal position within Ambridge culture? To give
such information outright would destroy the integrity of the text, but
perhaps to let little extra clues slide in unnoticed would be more in
keeping with the spirit of deconstruction.
540 Ian Parker
Pat Archer has picked up Clarrie Grundy, and she is giving her a lift
to do her shopping. The scene opens with the noise of the car.
Pat: Er, look, I can drop you in the middle of the town but the trouble
is Im going almost straight home. How long do you think youll
be?
Clarrie: Oh not to worry Pat thanks. Im dead lucky to get a lift one
way. I can take the bus back easy.
Pat: What, with all your shopping?
Clarrie: Well, Dad said to start walking and hed see if he could come
and pick me up. Oh poor Dad. He was in a terrible state last night.
His friend Dixies dead.
Pat: This is his friend in Canada?
Clarrie: Yeah thats right. He aint seen him in over thirty years.
Pat: Oh well, maybe he can go to Holland for his holiday instead.
Clarrie: Oh he were getting so excited about going to Canada too. Oh
its such a shame.
Pat: Ah, is the supermarket alright for you?
Clarrie: Oh yeah, lovely thanks.
Pat: My friend Jilly lives behind the supermarket, and Im just
dropping by to pick up the scripts for the panto.
Clarrie: What panto?
Pat: The Ambridge Christmas panto. Its instead of a review this year.
Clarrie: Ooh thats nice! Which one is it?
Pat: Cinderella.
Clarrie: Oh! my favourite!
Pat: The scripts written by some friends of mine. The writers group
call it a work in progress.
Clarrie: Do they?
Pat: Thats not to say it isnt finished. Theres a beginning, a middle,
and an end. Well, er, a sort of end anyway.
Clarrie: I thought Cinderella always had the same ending. She lives
happily ever after with Prince Charming dont she?

Pat: Well not in the Ambridge Christmas review she doesnt. Not
exactly. Were doing a modern reworking of the story.
Clarrie: Oh, thats nice.
Pat: What were out to challenge is the standard interpretation of the
story. What the writers group calls the tyranny of the finished text.
Clarrie: Oh yeah.
Pat: I mean. Why should the ugly sisters be ugly? Why cant they just
be sisters? If they were brothers, they wouldnt be the ugly brothers
would they?
Clarrie: Oh I do like pantos though. Theyre somewhere to take the
kids after Christmas. Oh, there were one in Felpersham a couple of
years back.
Pat: Im going to have to do something about Prince Charming too. I
didnt do away with him completely. In real life theres no such
person as Prince Charming.
Clarrie: Well whos Cinders going to fall in love with then?
Pat: She doesnt have to fall in love with anyone. Shes her own
person. She can find fulfilment in her work.
Clarrie: Clearing up for the ugly sisters!
Pat: No. I had something a little more elevated in mind. Erm Clarrie,
can you come to auditions in the village hall next week?
Explanation
542 Ian Parker
What series of explanations is Pat giving to Clarrie, and how may we

understand the role and status of those explanations? As the text
proceeds Pat explains why she cannot wait for Clarrie to finish her
shopping and run her home. She explains what the rewriting of the
panto entails, and why. Note that her first series of explanations is
concerned with warranting an action which has a primarily practical
aspect (dropping Clarrie), while the second set is concerned with
warranting the production of a symbolic representation - an expression
(writing a script).
There are two interconnected points about the nature of the

relationship here between Pat and Clarrie. The first is to do with the
transition from the practical resources. She is driving her car in which
Clarrie is the passenger, and so she is able to determine Clarries
behaviour. This practical leverage, however, operates through the
expressive meanings through which Clarrie can understand the
relationship. Thus Clarrie projects her disadvantages (heavy shopping

and a bus journey) back on to Pat (concern about Clarrie: not to worry
Pat thanks) in a display of gratitude for the lift. Pat sticks with the
practical topics (where Clarries Dad could go for his holiday),
ignoring the emotional focus of Clarries description of her Dads
troubles and her own concern for them. Instead, Pat waits, and then
determines the transition to a later discussion of the panto. Pat handles
the bridging sequence from the one realm to the other by explaining
why the supermarket is convenient as a point to drop Clarrie off, what
the script is about, and finally why the script is a modern reworking
of the story. Pat, then, carries over the control she has over practical
resources into the expressive sphere.
The second point is to do with the relative status of the practical and
the expressive orders. (It would be possible to deconstruct the
opposition. We could argue that ostensibly expressive activities are
544 Ian Parker
really practical in that they determine relationships and, therefore,

what is deemed to be practical.) For the purposes of the present
analysis, we make no assumptions about the supposed relationship
between the two spheres, preferring instead merely to use the
distinction as a device to understand what is going on.
In the text, Pat is implicitly using the distinction to identify and

reproduce her relationship with Clarrie. Pat is able to introduce the
question of the script for the panto. Pat constructs an explanation of
the contents of the script, and lures Clarrie into playing a part that
(expressive) representation of (practical) life in Ambridge. We should
understand the nature of this luring in both its real and metaphorical
sense. In a real sense, as Pats invitation to Clarrie (to come to
auditions in the village hall) makes clear, the upshot of the
conversation is that Clarrie should actually play a part in the panto.
Prior to that, however, Pat draws Clarrie into a discussion of drudgery,
subordination and fulfilment in which Cinderella becomes the

metaphorical frame through which they both might understand the real
world. Pat, then, controls the expressive sphere through which she
hopes to change practical relations.
But what is Clarrie explaining to Pat? In this text Clarries

explanations are responses to Pats topics (practical questions of where
to drop Clarrie, and expressive constructions around the contents of
the panto). Clarrie explains how she might cope with the shopping
(Dad would see if he could come and pick me up), and why Dad is in
a terrible state (His friend Dixies dead and he was so excited about
going to Canada too). There is also a level of implicit explanation
where Clarrie attempts to express her view that the modern
reworking of Cinderella is not actually a real panto (Oh I do like
pantos though), and that fulfilment in her work would not be an
escape from drudgery (Clearing up for the ugly sisters!). Note that,
546 Ian Parker
for Clarrie, explanation is a function of her acquiescence in Pats

explanations. She has little power. With these issues raised, we can
turn to the elaboration of the deconstructive polarities which operate as
a kind of conceptual cartilage for the text.
Deconstruction
Step one. Here there are two characters upon whom the conceptual
polarities can be mapped. You might like to think of the opposing
poles as being the dominant points in a definition of the situation in
which the antonyms are suppressed and to which Clarrie submits. I can
identify three oppositions: metanarrative/narrative, recasting/casting
and reflection/subjectivity. Each pair raises issues about what social
actors may understand about the relationship between what they take
to be practical concerns and what they imagine to be the expressive
gloss through which they represent those practices to others.
The recasting Pat engages in is one which attempts to break Clarrie

out of her everyday role. As the conversation proceeds Clarrie gets
drawn into a symbolic representation of her life which is the script of
Cinderella. The story parallels her own predicaments, but now Pat is
raising the possibility that the story does not always have the same
ending (it is a work in progress). The casting which stands in
opposition to this is Clarries everyday role as dutiful daughter (with
poor Dad) and harassed mother (where the panto is somewhere to
take the kids after Christmas). Clarrie refers to these aspects of the
way she is cast in her life in Ambridge, but Pat resorts to inviting her
to an audition which will radically recast her.
Clarries life, then, is being mirrored. An image is being wrought

which is reflecting the character Clarrie plays in Ambridge. Look
says Pat to Clarrie at the opening of a text in which Clarrie is
progressively reduced to silence. Her assertions about her own
548 Ian Parker
experiences (of fulfilment equated with clearing up) - her own

subjectivity - are discounted as Pat breaks beyond her social world
(the tyranny of the finished text) to write another for her.
Step two. Take the third polarity-reflection/subjectivity - first. How

might Clarries view of her life be restored? More to the point, what
attention might be drawn to the subjective investment Pat has in the
mirror she holds up to Clarrie to recognize herself anew? Pats control
of resources, both practical and expressive, is at issue here, and her
power. Note the way that Pats own subjectivity, her own involvement
and responsibilities, are sidestepped by her when she defers to the
writers group statement about the tyranny of the finished text. Does
not the mirror she holds reflect herself, and does not the relationship
between herself and Clarrie (Pat as holder of the mirror and Clarrie as
the reflected subject) reproduce the relationship between the writers
group and herself (the writers group as interpreters of the text and Pat
as the relay of that interpretation)? The mirror, then, is also an

expression of subjectivity.
By the same token, Pat is engaged in an activity where the recasting

of Clarrie simultaneously enacts the casting of herself in the role of
producer, director and purveyor of knowledge. Could we provoke
recognition of this supposed recasting as a mere reproduction of the
way Pat and Clarrie are cast as actors in Ambridge (with all the power
that attends the unequal distribution of knowledge)? And again, the
metanarrative to which Pat appeals functions as such only because it is
a narrative invested with power - power Pat enjoys as a result of her
knowledge of theory.
550 Ian Parker
The next step. The third step of a deconstruction involves

moral/political choices. These are choices which can only be pursued
by the social actors themselves. There are issues about Pats
employment of theory to shatter Clarries everyday explanations, and
how her use of that theory frustrates her desire that Clarrie should take
control of her own life. These issues, of course, are to do with the
relationship between power and knowledge. To offer prescriptions to
our subjects recast as naive lay folk who might be enlightened by
theory and liberated from the tyranny of the finished text would surely
tangle us in the same contradictions.
I will have to set my discussion of the advantages and disadvantages

of deconstruction in the context of two separate, though related, dis-
courses. One is the discourse of analyses of everyday explanation.

The other is the discourse of politics.
As with the structuralist strands of new paradigm social

psychology, deconstruction locates conceptual categories which
organize the text. Rather than posit these as being underneath what is
said in the deep structure of interaction or in cognitive templates
located in the heads of individual speakers, deconstruction finds these
conceptual categories in the surface of the text itself. It recognizes that
the categories are important because of the way they are used.
Oppositions are the contradictions utilized by speakers in their rhetoric
to interpret and determine the actions of others. We saw this in the way
Pat interpreted Clarries explanations within her own metanarrative as
she tried to reflect Clarries life and recast her in a rewriting of
Cinderella. More than this, a deconstruction also shows how the
oppositions are contradictions which capture all subjects in a text and
552 Ian Parker
undermine the intentions of the authors of statements. We saw this in

the way the power Pat wielded undermined her attempt to empower
Clarrie.
As with the hermeneutic strands of new paradigm social

psychology, deconstruction attends to the fluidity and indeterminacy of
meaning in a text. Instead of attributing this indeterminacy to the free
exercise and expression of individual human intention, however,
deconstruction locates the shifts of meaning in the text itself. Again,
this focuses attention on the powers distributed to social actors in
different positions. Power and language are intimately linked. It
supports social actors attempts to change power relations and to
engage in the overthrow of dominant ideas. So we saw how it is
possible to reinterpret Pats mirroring of Clarries life and her attempt
to recast her accounts in a metanarrative. Pats rhetoric depends on the
very issues of subjectivity, casting and narrative which she sought to
exclude.
Deconstruction goes even further when it puts into question notions

of everyday explanation. Theory itself is derived from ordinary
explanation and is conveyed in a medium which uses the language of
the surrounding culture. Then the theories seep back into everyday
usage. Even complex debates find expression in the apparatus of
everyday explanation and practically deconstruct the distinction
between the two realms.
554 Ian Parker
While the overthrow and transcendence of conceptual oppositions

in the third step of a deconstruction is possible in philosophical and
literary texts, a deconstruction applied to social texts raises questions
of practical political involvement of the researcher in power relations.
The first two steps uncover these relations, but the next step can only
be taken outside the text. Without this step, which involves the
construction of non-oppressive social relations, the hierarchies will
simply be written and rewritten in the same old way. Listeners to The
Archers will know that Pats panto was abandoned, and the
inhabitants of Ambridge returned to the Christmas review.
A lesson which proponents of the new paradigm in social

psychology tried to teach those gripped by the old paradigm was that a
researcher had to go beyond data to ask people what they were doing.
The process of gathering accounts is part of research and is the
foundation of the idea that results are provisional and open to revision.
Whilst the fruits of a deconstruction are always ready for the next
interpreter to get his teeth into, we are limited by the boundaries of the
text. The sequences of signs which are offered to the reader are still
there to be manipulated and pulled apart.
Social psychology in the new paradigm took what people said

seriously, acting on the slogan treat people as if they were human
beings. Deconstruction, on the other hand, carries with it the danger
of reducing persons to ciphers of the text. The relationship between
knowledge and power, the violence which attends the prioritization of
some concepts over others, also deeply affects the relationship
between a researcher and subjects. The theories in and around
deconstruction also find their way into everyday explanation. The
conflicts between Pat and Clarrie illustrate the problems of power
which attend it. So when Pat is given a book by Lacan for her birthday
this thrills the cognoscenti and drives back the inhabitants of the
556 Ian Parker
realms of everyday explanation (people like Clarrie) further beyond

the pale of theory.
The reader should be aware of the privilege that the position

confers. The tyranny of the finished text can also operate when
theoretically aware researchers approach and interpret the accounts
given by the lay public. Perhaps this problem would be mitigated if
proponents of deconstruction could advance a moral/political position.
Both strands of the new paradigm in social psychology do propose a
politics which flows from the value attached to a persons right to
account for themselves and their actions. Unfortunately, in the case of
deconstruction this is not possible without supplementing the
approach. By its very nature it is hostile to any attempt to construct
conceptual or political priorities. Further, I would want to resist the
implication that Clarries resistance to Pat is somehow progressive.
The reflection of social actors lives and the recasting of their acts
within wider political narratives is, I think, a good thing.
I avoided confronting the issue of where the theory leads, and the
power of theory in general, by choosing a text in which the speakers
were no more than constructions. We may learn lessons about social
relationships from such texts, but the pity is that a deconstruction
would not necessarily lead to a progressive type of action research if it
was applied to real explanations given by real people. I recommend
that readers wishing to adopt deconstruction as a method take as their
texts other fictions: discussions extracted from the pages of novels, or
the products of role-play studies, or social psychology textbooks.
Note
14
Thanks to Carolyn Kagan for her comments on

an earlier draft of this chapter.Methodology and
scholarship in
understanding ideological explanation
Michael Billig
This chapter will not outline the whys and wherefores of a particular
14
methodology as such. In fact, it will challenge the use of methodology,

and the importance which social scientists give to methodological
matters. When social scientists advocate the use of a methodology,
whether for understanding explanations or for anything else, they are
prescribing a set of procedures which the analyst is to follow. Social
scientific investigation is frequently presented as being based upon the
following of methodological rules. However, this chapter will
recommend an alternative approach: that of traditional scholarship.
The approach of the traditional scholar can be considered antimethodological, in that hunches and specialist knowledge are more
important than formally defined procedures. This anti-methodological
stance will be illustrated by considering the issue of ideology and, in
particular, the conspiracy theory of politics. The views of one
politician will be examined in detail. It will be suggested that more
understanding is to be gained by using the traditional, ill-defined skills
of scholarship than by following a rigorous, up-to-date methodology.
560 Michael Billig
There
are a number
of waysmean
in which
traditional scholarship
can involves
be distinguished
from
what
modern
investigators
by methodology.
A methodology
presenting
rules
of
procedure
about
matters
such
as thetocollection
of data
and
their analysis.
The
rules
are
impersonal,
in
that
they
are
meant
apply
equally
to
all
researchers.
It
is
assumed
that any two researchers who approach the same problem should arrive at
identical results,
as long as neither infringes the methodological rules. Thereby, it is hoped that individual bias is
excluded from the research process. In this way, methodology attempts to standardize the practice
of the social sciences and to eliminate quirkiness. To the modern methodologist, traditional
scholarship seems a haphazard and biased affair. The traditional scholar does not seem obsessed
with laying bare the methodological procedures, which can be followed by anyone with sufficient
training.
Methodology and scholarship in explanation 561

Individual quirkiness is very much part of traditional scholarship. It was taken for granted by the
traditional scholar that one should read as widely as possible, and in as many languages as possible.
Through wide reading, breadth and depth of knowledge would be gained, as well as the ability to
make connections between seemingly disparate phenomena. The learned scholar would be able to
interpret individual texts with an acuity not available to those of restricted reading. Traditional
scholars are not particularly bothered with the origins of their insights, in the sense that they do not
attempt the impossible task of laying bare all the intellectual experiences which lead up to the
ability to make a scholarly judgement. Nor do they presume that other scholars will read the same
texts in just the same way as they have. In fact, scholars spend a great deal of energy in criticizing
the readings of their fellow scholars.
The major theme of this chapter is that the traditional skills of scholarship have much to offer
the study of ideology. It is certainly not the case that students of ideology have adhered to a
particular methodology. In fact, they have not even adhered to a simple definition of ideology itself:
different theorists offer contrasting definitions (Abercrombie, 1980; Billig, 1982; Larrain, 1979;
1983). According to McLellan (1986:1), Ideology is the most elusive concept in the whole of
social science. Despite all the disputes about definition, there is agreement that an ideology refers
562 Michael Billig

to a pattern of ideas, values, interpretations and so on rather than to a single attitude. For example,
Mannheim (1960: 49-50) expressed this view when he wrote that We refer to the ideology of an
age or of a concrete historico-social group, e.g. of a class, when we are concerned with the
characteristics and composition of the total structure of the mind of this epoch or of this group.
Thus, the analyst of ideology is interested in the whole pattern of a groups beliefs, not just its
constituent parts, and in addition the analyst seeks to place this pattern in its social and historical
context.
The last point can be illustrated by considering the issue of explanation. It is quite clear that
people are prepared to offer explanations for major social events. For instance, attribution theorists
have shown that people will explain phenomena such as unemployment or poverty in terms of
personal or situational attributions, either blaming the victims themselves for their own condition or
attributing the blame to society (Feagin, 1972; Feather, 1974; Furnham, 1982, 1983; Furnham and
Lewis, 1986; Lewis et al., 1987). To probe the ideological significance of these attributions one
needs to go further than documenting their existence. One needs to discover how the explanation of
one sort of social event fits into a wider pattern of explaining social events. To do this, one needs to
look at the way that different ideologies might be characterized by different styles, and traditions,
of explaining social events.
It is possible to study ideology by attempting to situate ordinary discourse within its wider
ideological context. The pattern of an individuals thinking should therefore be interpreted in terms
of broader historical and social patterns. For example, the chapters written by Adorno in The
Authoritarian Personality (Adorno et al., 1950) show a great awareness of the way that bigoted
respondents could be linked to the intellectual and ideological history of racial bigotry. Similarly,
Billig et al. (forthcoming) trace the heritage of Voltaire and the Enlightenment in the comments of
young racist respondents (see also Billig, 1988). A further social psychological analysis of ideology
is provided by Billig (1986), who links the thinking of young members of the Conservative Party
with the ideology of international finance capital and the assumptions of everyday life in capitalist
society. Such studies attempt to uncover the social psychological dynamics of some of the central
ideological themes in contemporary society. However, the topic of ideology can be approached
from an opposing direction, by examining those ideologies which remain just beyond the
mainstream of everyday thinking.
564 Michael Billig

The conspiracy theory of politics represents such an ideology, existing just beyond the
mainstream of everyday thought. Moreover, it represents an ideological pattern especially relevant
to the social psychology of explanation, for the ideology is characterized by its particular style of
explaining social events. In essence the ideology of conspiracy seeks to explain all major political
events in the world in terms of an evil conspiracy, or series of conspiracies. The conspiracy theorist
tells a story of hidden machinations by small groups who are plotting to subvert the natural order of
the world. According to Lipset and Raab (1970), the conspiracy theorist offers a monomanic
explanation for social events, in that all major happenings are explained in precisely the same way:
no matter what happens, the conspiracy theorist sees the malign hand of hidden conspirators. In this
sense conspiracy theory represents an extreme form of personal explanation, in that nothing
happens by chance, since all is to be explained in terms of deliberate plotting (see Billig, 1978:
313ff. for discussion of the conspiracy mentality and the psychological theory of attributions; see
also Kruglanski, 1987; Moscovici, 1987 and Zukier, 1987). Over and above the conspiracy
theorists use of a particular type of explanation, there is the matter of explanatory style. The
conspiracy theorist employs apocalyptic terms, asserting that the masses have been duped: unless
people awake soon to the conspiratorial reality, all will be lost. The conspiracy theorist is like a
prophet, who has glimpsed a higher reality and is berating the masses for their unseeing
complacency in the face of imminent danger.

Historically this tradition of political ideology has been linked to the collapse of the old
aristocratic order at the turn of the eighteenth century (Cohn, 1967; Groh, 1987; Roberts, 1974).
Supporters of the old order, unable to grasp that the mass of the population no longer retained
traditional feudal loyalties, suspected that freemasons had poisoned the minds of the populace. This
general mentality of conspiracy has been identified as the mentality of the dispossessed, who hope
that, once the evil machinations of the conspirators have been rendered harmless by revelation, all
wrongs will be righted (Bell, 1962). The same ideological structure can be found amongst American nativist groups (Lipset and Raab, 1970), German Nazis (Cohn, 1967; Poliakov, 1974) and
contemporary fascist groups (Billig, 1978; Seidel, 1986; Taylor, 1982; Thurlow, 1987), who, above
all, fear that the conspirators are seeking to destroy the independence of nations and to take control
of the world. The arch-conspirators are variously freemasons, communists, international capitalists,
Jews or, better still, small groups who mysteriously share all these attributes.
In one sense, it might be thought that the conspiracy mentality represents a style in which
ideologies can be expressed rather than an ideology itself. For example, the style could be found on
the left wing of politics. Left-wingers might articulate their socialist ideology in such a way as to
suggest that the hidden hand of capitalist plotters is seen in every failure of the masses to turn in
566 Michael Billig

revolutionary upsurge to the teachings of Marx and Lenin. Certainly it would not be difficult to find
parallels between the ideological systems of explanation proposed by some far left- and right-wing
groups (see Feuer, 1970). However, formal similarities are not proof of an ideological unity. It is
easier to argue that the conspiracy theorists of the far right and the far left draw upon differing
ideological traditions than that they share a common intellectual heritage. Even where there is a
resemblance in content, as for example when anti-semitic themes are used, distinctions can still be
made between the ideological heritages of right and left (Billig, 1987a). Therefore, the issue of
exploring ideological heritage is crucial to an ideological analysis of a pattern of beliefs such as
conspiracy theory.
In his study of British contemporary fascism, Billig (1978) showed that quantitative
methodologies are insufficient for investigating the ideological heritage of conspiracy theory. He
showed that a content analysis of UK National Front propaganda only revealed the surface
characteristics of that groups ideology. To probe further the deeper meanings and traditions of the
ideology it was necessary to explore the wider traditions of conspiracy theory. It was necessary to
show how National Front ideologists had absorbed notions from an unbroken ideological tradition,
which includes overtly Nazi theorists, non-Nazi groups such as the John Birch Society and

eighteenth- century writers such as Auguste de Barruel and John Robison. The identification of this
common political culture was not achieved by methodological means as much as by the practices of
traditional scholarship. Texts had to be sought out and read and half-hinted allusions had to be
noticed and then interpreted. The ideologies of different extremist groups had to be compared, to
see whether they were drawing upon common ideological heritages. Only by so doing is it possible
to discover when an ideologist is alluding to a writer or a set of ideas firmly situated within the
conspiracy tradition.
The tradition of conspiracy theory is in the main to be found on the periphery of political life,
although, as the history of the Nazis revealed, there can be no guarantee that it must remain
marginal. At present one would not expect conspiracy theory to have a central place in conservative
thinking at least as an explanation of world governmental behaviour. The political successes of
contemporary conservatism should ensure that there are few failures and dispossessions which need
to be explained away in terms of an overpowering conspiracy. However, there have often been
links and overlaps between conservative and conspiracy thinking (see Griffiths, 1983; Webber,
1986). In investigating the political terrain where the racist right of conservatism shades into fascist
extremism, one should look to see whether conservatives are adopting the ideological style and
568 Michael Billig

content of extremism. In particular, it is necessary to examine conservatism for signs of the
conspiracy mentality.
If such an investigation is to be undertaken, it is not enough merely to see if one or other

conservative politician uses conspiracy as a political explanation for a particular event. More is
required if an individual conservative is to be located within the ideological traditions of conspiracy
theory. The individual must show evidence of possessing an ideological structure which ties
together the untidiness of the social world into a tale of deception and conspiracy. In addition there
should be evidence of a linkage with the ideological tradition itself. The following sections will
consider these problems in relation to the beliefs of a particular UK Conservative Party politician,
Enoch Powell. They will pose the question of whether Powells beliefs should be located within the
ideological traditions of the conspiracy theory.

Empirical example: J. Enoch Powell
The political discourse to be analysed in this chapter is taken from reports of the speeches of a
single politician, J. Enoch Powell. Before proceeding to an ideological analysis of Powells
discourse we need to know something of the life of this most singular of contemporary British
politicians.
570 Michael Billig

Until the last General Election Ulster Unionist Member of Parliament for South Down, Enoch
Powell has had a notable and highly controversial career. He was born in 1912, and studied classics
with great distinction at Cambridge University. At the age of 26 he became Professor of Greek at
the University of Sydney. At this time he was publishing translations of Thucydides and Herodotus,
as well as slim volumes of his own poetry. After two years as an Australian professor, he
abandoned academic life to join the army as a private. He has explained this decision as follows: I
was determined that I would do something to stop the disintegration of the Empire which seemed
imminent (quoted in Berkeley, 1977: 14).
After the war Powell decided upon a career in politics. He entered Parliament in 1950 as the
Conservative member for Wolverhampton South East, in his native West Midlands. His intellectual
capacities, as well as his talents for composing and delivering speeches, marked Powell out for high
office. He became Minister of Health in Harold Macmillans cabinet, yet he resigned when
Macmillan was replaced as Prime Minister by Sir Alec Douglas-Home. He did not remain on the
backbenches long, and became a senior member of Edward Heaths shadow cabinet. Ideologically
Powell and Heath represented different wings of conservatism. Whilst Heath was the pragmatist,
who believed in state intervention in the economy, Powell was elaborating his own political

philosophy based upon the principles of a laissez-faire monetarism, long before the label
monetarist became popular.
However, it was not differences of economic policy which caused the decisive rupture between
Heath and Powell. On 20 April 1968 Powell delivered a speech which was to have significant
effects on the course of race relations in Britain and was to change dramatically his own political
career. Powell spoke about his fears for the future of a Britain, which contained a sizeable New
Commonwealth or non-white population. It was not so much the anti-immigrant themes which
made the speech notorious, as the choice of language. Powell, the intellectual Greek scholar,
repeated stories of negroes pushing excreta through letter-boxes, of wide-grinning piccaninnies
harassing old white ladies, and so on. The speech foretold of racial conflict, as Powell said that he
was filled with foreboding. The insults of the gutter were mixed with the learning of the high
table: Like the Roman, I seem to see the River Tiber foaming with much blood (text in Powell,
1969: 213ff.; for analyses see Foot, 1969; Barker, 1981).
572 Michael Billig

The impact of the speech was immediate. Heath sacked Powell from the shadow cabinet, calling
the speech racialist in content (Schoen, 1977: 34). Immediately Powell had become a national
figure of substance: letters of support for him poured in, dockers marched for him in East London,
and the right-wing popular press depicted him as the man who had voiced the fears of ordinary
(white) people. Studlar (1974) has shown how dramatic was Powells simultaneous public rise and
political fall. In early April 1968 Gallup surveys had shown that 1 per cent of the electorate
believed that Powell should become Conservative leader were Heath to resign. One month later, the
figure had risen to 24 per cent. Powells popularity came from all income groups and was focused
entirely on the issue of race (Schoen, 1977; Studlar, 1974). Although sacked by Heath from the
shadow cabinet, Powell seems paradoxically to have helped Heath win the 1970 General Election.
Studlar (1978) and Miller (1980) have suggested that Powell established the Conservative Party in
the minds of the electorate as the party of strict immigration control. Moreover, Heath himself
adopted some of Powells policy recommendations on immigration control, but without the lurid,
river-foaming rhetoric (Layton-Henry, 1984).
Had Powell bided his time and awaited a change of leadership, he might have eventually
resumed his ministerial career. Certainly his laissez-faire economics would have been no handicap

in the Conservative Party of Margaret Thatcher, for, as Gamble (1986: 49) suggests, all the
arguments adopted by Thatcher and the leadership group she slowly assembled around her had
been anticipated by Enoch Powell in the 1960s. However, Powell continued to tread the maverick
path. He attacked Heaths commitment to the Common Market, on the grounds that the EEC
threatened the national integrity of Britain. His opposition to the Common Market was total. He
declared that he was unable to stand as a candidate for a party whose stance he so profoundly
disagreed with. Powell resigned his Wolverhampton seat and declined to stand in the February
1974 General Election. In addition, he committed the heresy of all Conservative heresies; he
advised the electorate to vote Labour (the effects of Powells appeal are analysed by Steed, 1975:
331-2).
Powell did not remain outside the House of Commons for long. The October 1974 General
Election saw him take a step remarkable for a senior English politician: he was standing as an
Ulster Unionist candidate for a Northern Ireland constituency. Powell was returned as the member
for South Down, strongly committed to the policy of integrating Ulster totally with the rest of the
United Kingdom. He retained his Northern Irish seat until the 1987 General Election but as a
Unionist MP he had no reasonable hopes of entering government again. Despite his contacts with
574 Michael Billig

senior Conservatives to the right of the party (for example, in the Conservative Philosophy Group)
Powell has remained an anomalous figure. He is a national politician but is confined to the one area
above all others in Britain which thinks in local terms. Powells Irish constituency was largely
indifferent to the issue which created his national prominence. And although he might be thought
the apostle of Thatchers laissez-faire conservatism, he is in permanent exile from the Conservative
Party. Yet he continues to disconcert his former colleagues with his unmistakable style of rhetoric,
urgently predicting doom and chaos.
Selection and presentation of text
The question under consideration is whether the views of this unorthodox Conservative, Enoch
Powell, can in any way be located within the conspiracy tradition of ideology. A sound
methodological procedure for attempting to produce an answer would be to opt for a quantitative
content analysis (for an excellent discussion of the relations between quantitative and qualitative
research see Bryman, 1984). The speeches of Enoch Powell could be systematically sampled perhaps every fifth or tenth of his published orations could be selected. Word counts could be
made, with or without computer assistance. Perhaps the word conspiracy could be chosen for
special attention. The analysis might then show how many times, per 10 000 other words, Enoch

Powell uses the key term conspiracy. Powells ratio could then be compared statistically to the
ratios of other speakers, whose texts had been chosen with as much methodological care.
All this might be quite interesting, but, as Beardsworth (1980) has shown in an important
critique, the techniques of such content analysis are essentially limited. This sort of methodology
can count words, but it cannot interpret them. Under some circumstances mere counting can lead to
misleading conclusions. For example, someone who is continually scoffing at the ideas of
conspiracy theories might have just as high a usage of the word conspiracy as the most obsessed
believer. Furthermore, the question about Powell is not whether he talks about conspiracies, but
whether any such talk should be located within the conspiracy ideological tradition. As will be
shown, mere talk about conspiracy, even belief in a conspiracy, is not of itself sufficient evidence
for such a location. Over and above statistical identification, such beliefs need to be interpreted by
the ideological analyst. Interpretation cannot be achieved by handing over the whole business of
scholarship to a programme of computation.
576 Michael Billig

It is one feature of traditional scholarship that attention is not confined exclusively to a single
text. Perhaps the scholar might be perplexed by the meaning of a particular text, but that will often
signal the start of a search which will embrace other texts and wider reading. Part of the scholars
skill is not to follow a preset programme, laid down in advance by a methodologist, but to gather
up clues which can nudge the search one way or another. Scholars have to feel their way around
their library and archival sources, backing hunches as they proceed. The chosen example will
illustrate this process, by selecting a text which is not to be analysed in itself but which sets in
motion a search for new information and new interpretations.
The text is a newspaper report of one of Powells speeches, and therefore is one that is available
to experts and laypeople alike. However, once an ideological question is raised, the analyst must go
beyond the specific text. In fact the analyst has to go beyond this specific speech, or the report of
the speech, and to look directly at other reports and in particular at other speeches. As the search for
ideological significance proceeds, more information is required, and the analyst must widen the
investigation to include conspiracy material from non-Conservative sources. Inone way the original
text is a starting point for a search, rather than being the object of a methodological examination in
itself. In another way, the text is not the starting point: the analyst will already have built up a

knowledge of the topic before starting the search required for understanding the particular text. In
fact, without this background knowledge, the analyst might not even decide that the text was one
which was worth detailed analysis. However, the background knowledge does not guarantee the
outcome of the search.
On 19 January 1986 the British Sunday papers carried stories that Enoch Powell had suggested
that the United States was responsible for the murder of a senior Conservative politician seven
years earlier. The Observer carried the story on its front page with the headline: US had Neave
killed - Powell. The text summarized the main themes of Powells speech:
Mr Enoch Powell yesterday accused the United States, the Foreign Office and British military
intelligence of a conspiracy to create a united Ireland inside the NATO alliance in an extraordinary
speech to the Federation of Conservative Students in Birmingham.
Mr Powell, the Unionist MP for South Down, suggested that Washington had 20 years ago
secured from Britain an undertaking to organise the transfer of Northern Ireland out of the United
Kingdom into an all-Ireland state.
He said the first step in the conspiracy was to eliminate the stumbling block of the Northern
578 Michael Billig

Ireland Unionist Government. ... MI6 and their friends proved equal to the job, and in 1969 active
committal of the army in Ulster subordinated Stormont to Whitehall.
He suggested that none of the Prime Ministers of the period - Edward Heath, Harold Wilson and
James Callaghan - knew what was happening as the conspiracy unfolded.
Mr Powell went on to suggest that the Americans were responsible for the murder of Mrs
Thatchers close friend and shadow Northern Ireland spokesman, Mr Airey Neave, in 1979.
Evidence, or what was thought to be evidence, that the new leader, Mrs Thatcher, and her aide,
Airey Neave, had no intention of playing ball alarmed them.
The road block was cleared by eliminating Airey Neave on the verge of his taking office; and from
then onwards events were moved ahead again along the time tabled path.
Mr Powell described Mrs Thatcher as an unhappy traveller who was taken along the road
signposted towards the next stage - signing the Anglo-Irish Agreement.
Mr Powell said he was convinced Mrs Thatcher thought it was in Britains vital interests that
Ulster should be sacrificed for the good of all.
One need not translate the newspaper report into a computer input to realize that the
word conspiracy appears three times. It is not merely a matter of the word being
repeated; Powell was obviously telling a dire story of devious conspiracy and murder.
If the tale of hidden conspiracy was being told to a Conservative audience, does this
imply that Powell was engaged in something more general? Was he undermining a
whole series of conventional explanations, in order to spread a wider tale of
conspiracy? The single report cannot answer this question. Instead, one must turn to
other speeches and other reports and, in so doing, examine various possible
interpretations
.Methodology and scholarship in explanation 209

Interpreting the themes of conspiracy
From a rhetorical perspective, one of the first tasks in analysing a piece

of discourse is to discover the argumentative meaning (Billig, 1987;
Billig in press a; Billig in press b). This involves establishing who, or
what, is being criticized in a piece of argumentative discourse. In an
obvious sense, one might say that Powell is attacking the Americans
for their conspiratorial deeds. However, Powell is not addressing his
remarks to the Americans, for they seem beyond the argumentative
pale. His remarks constitute arguments against those who trust the
American government. He is overtly criticizing Thatcher for not seeing
the truth about the Americans. In consequence, Powell is following
traditional conspiracy theorists in directing his arguments against
conventional, non-conspiratorial explanations of political events.
Powells speech was not an isolated outburst, but is of a piece with

others. If the analyst wishes to probe the ideological significance of the
one speech, attention should not be confined exclusively to that single
text. Instead, the single text must be located within a wider pattern.
The following week, after the Birmingham speech, Powell was
expanding upon his theme. He suggested that the British Prime
Minister had been obliged to enter into a pre-existing compact
concerned with the strategic purposes of the United States (The
Guardian, 27 October 1986). It was not only Airey Neave whom the
Americans had silenced, but also Lord Mountbatten. Since 1969 there
had been avowed compliance with the dictates of American strategy.
The following month, there was more of the conspiratorial same. In
Ballyculter, Powell declared that British policy must be subservient to
that of the United States, who wanted a United Ireland within NATO:
the hunt for a traitor who would deliver goes feverishly on (The
Guardian, 1 December 1986). Like many conspiracy theorists, Powell
evokes the image of public opinion being fooled but about to grasp the
truth. According to Powell, The sleeping dog of British opinion ...
presently will get up, yawn, stretch and set about wondering whether
after all, and why, it is so essential for the United Kingdom to comply
with every fantasy of the United States that it must perjure itself and
betray its own people in the process. Using a metaphor familiar in the
writings of conspiracy theorists, Powell declared that the British score
is written out for the second fiddle, and anyone looking over the
shoulder of the violinist can observe how obediently the score has been
played.
Nor are all these new themes, for past texts and speeches show Powells
opposition to a United Ireland. He had previously accused Mrs Thatcher
of treachery over the issue (The Times, 8 July 1985), and similarly he
had claimed that the British andAmerican governments aided terrorists
in Northern Ireland (The Times, 12 August 1985). The claim that there
existed an American plot to establish a United Ireland was an old theme
(The Times, 27 September 1980; The Times, 7 December 1981). Nor
were the American conspiracies confined to Ireland. For example,
Powell had spoken about the EEC suffering the admission of Spain and
Portugal in conformity with American strategic interests (The Times, 19
October 1985). In all these speeches it is possible to detect, in greater or
lesser form, the themes of conspiracy and treachery, as powerful
manipulators dupe smaller nations into abandoning their national
independence.
582 Michael Billig
The ideological significance of these conspiracy themes still needs to

be assessed. It might be asserted that the themes possess more
significance in terms of the psychology of Mr Powell than anything else.
For instance, a case might be made that the disappointments in Mr
Powells own political career have led him to suspect the machinations
of more successful politicians. This would imply that the images of
conspiracy stem from individual motiviations and are likely to be found
only in his later speeches. Even if the themes only occur since Powells
translation to the suspicious world of Ulster politics, it is still possible to
find ideological significance: away from the politics of England, he may
have acquired a conspiracy mentality not uncommon within the circles
of Loyalist politics.
However, there is no need to entertain the hypothesis that Powells

conspiracy themes date from his period in Northern Irish politics if
earlier texts also contain the theme of conspiracy. His speeches against
non-white immigrations were not confined to innuendoes about
grinning piccaninnies and dire warnings about the future: dark
conspiratorial deeds were afoot to fool and then destroy the (white)
nation. He claimed that the government and public were being
deliberately misled about the numbers of non-whites living in Britain
(for example The Times, 16 February 1971). There was an enemy
within, which was more sinister than the Nazis or the communists (The
Times, 15 June 1973). So dreadful was the conspiracy to conceal the
facts of non-white residency in Britain that One begins to wonder if the
Foreign Office was the only Department of State into which enemies of
this country were infiltrated (11 June 1973; quoted in Schoen, 1977:
53).
The theme of conspiracy is not confined to the topics of race or
584 Michael Billig
Northern Ireland, but over the years it has been a constant theme in
Powells political rhetoric. When he opposed British entry into the
Common Market, he did not see himself as merely opposing a political
decision, but he was standing against conspiratorial forces. He declared
that there was a conspiracy to force the public into
accepting British entry. He warned that there was a conspiracy of

the best people in parliament and the media, and that the ordinary
people have settled down to a dazed or irritated acceptance that they
are living in an age when the majority is always wrong (speech at
Rugby, reported in The Times, 29 November 1971).
Some of Powells most outspoken comments about conspiracy have

concerned economic issues, and in particular they are to be found in his
speeches against the economic policies of Harold Wilsons first Labour
government. Here thedefenceof/assez-/ai>eeconomicswasseen in terms
of a battle against an insidious conspiracy to impose an economicpolicy
which would destroy the nationsfreedom. The tone was as apocalyptic
as any conspiracy theorists (Lispet and Raab, 1970). Powell declared
The situation is perilous: the public had been so blinded that you
might almost suppose there was a gigantic conspiracy(14 October 1966,
quoted inPowell, 1969:191-2). Notlong later the almost conspiracy had
become a certain conspiracy: The whole prices and incomes policy was
a sustained and successful conspiracy against the common sense of
thepublic (speechin Gedling, 24 May 1968). He added that
Unfortunately it is not without precedent fora small groupof interested
people tosucceedinhoodwinkingthevast majority of their fellow-citizens
and thus bringing them under their control (Powell, 1969:106).
586 Michael Billig
Classic conspiracy theorists are not content to view one particular

governmental action as a conspiracy, but see all politics in terms of a
conspiracy. As has been shown, Powell sees hidden conspiratorial
designs lying behind various major issues, especially those on which
he is at variance with the Conservative leadership. Powell expressed
the general conspiratorial interpretation of politics when he declared
in 1968 that We live in an age of conspiracies and that The politics
of the last few years have been little more than a series of conspiracies
conducted by the politicians and the Press against the common sense
of the public (Powell, 1969: 109). The conspiracy is a mightier
instrument of mass repression than machine guns (p. 110); it aims so
to mesmerize people that they deliver themselves passively to the
guidance and domination of their betters (p. 109). Classic conspiracy
theorists claim that absence of direct proof of a conspiracy is a further
sign of the cunning of the conspirators (Billig, 1978). Similarly,
Powell alleged that The success [of the conspiracy] has been so
complete that we fail not only to be astonished at it, but even to
perceive it (p. 110).
In Powells demonology, each conspiratorial operation is a step in a

wider conspiratorial pattern. He claims that people were being
made vulnerable to the next stage of the operation, which is to subject

them to control, to a dictatorship (p. 112). Powell, like classic
conspiracy theorists, poses as the visionary who has seen through the
evil machinations and warns of the doom which will surely occur unless
he is heeded. In a speech delivered on 13 June 1970, he declared Britain
is at the moment under attack, just as it was in the two world wars. In
fact, continued Powell, The danger is greater today, because the enemy
is invisible or disguised so that his preparations go on hardly
observed. ... I assert then, that this country is today under attack by
forces which aim at the actual destruction of our nation (quoted in
Berkeley, 1977: 115-16). The conspiracy is not just directed against
Britain, but it is worldwide in its aims for dominance: All around the
588 Michael Billig
world in various forms the same formula for rending societies apart is
being prepared and applied by ignorance or design and there are those
who are determined to see to it that Britain shall no longer be able to
escape (quoted in Berkeley, 1977: 88).
These themes suggest the conspiratorial dimension in Powells

political mentality: ordinary explanations of politics are rejected in
favour of interpretations revealing hidden conspiracies. The nature of the
conspirators is not fully specified, but they seem to fulfil American,
rather than Soviet, aims (see, for instance, Powells criticism of
Thatchers belief that the Soviet Union seeks world domination in The
Times, 8 October 1983). Although it has been possible to construct the
outlines of a conspiracy mentality by piecing together themes from
Powells speeches and writings, there has been no direct link to the
conspiracy tradition itself. It is possible that this unusual politician had
constructed his own conspiratorial themes, in ignorance of the libraries

of conspiratorial publications. To locate Powell fully within this
ideological tradition, it would be necessary to show evidence of a direct
link.
Some sort of evidence is contained within an extraordinary speech

delivered in Bridgnorth on 28 August 1969 and reported in The Times. It
is a significant speech because it is his first upon Northern Irish matters.
Ostensibly, the speech comments upon the report that a French and a
German student had been gaoled at Londonderry for throwing petrol
bombs. Yet Powell reads wider designs into the report of the single
incident. The event was part of a pattern which is subverting nationality
and which has all but destroyed governments and states in Asia, in
Europe and in America. This wider pattern could explain all manner of
phenomena in contemporary life, including racial tensions in Britain and
590 Michael Billig
the payment of local property taxes or rates: The pattern is recognizable

enough and it belongs in an international, a worldwide context. The
moral of all this was that We must simply have more control over the
admission, the movement and the activities of aliens in this country.
These themes of national destruction and international plotting seem

familiar enough, but what marked this speech out was a remark at the
beginning. Having mentioned the newspaper report of the two gaoled
students, Powell declared that We shall do well to ponder the news item
deeply: for, as Douglas Reed used to say in the 1930s: This means
you. No individual is cited in the speech other than Douglas Reed. Nor
is any information given about Reed. To most of the audience and to
most of the readers of the written text the name of Douglas Reed would
be unfamiliar. But it would be recognizable to those with a knowledge of
the traditions of British anti-semitism and the conspiracy ideology. In
this way, knowledge of the subject area can lead to insights which fall
outside the range of methodological expertise.
Douglas Reed had been a prewar correspondent of The Times in

Germany. He left Germany in 1935, an opponent of Hitler but not of
National Socialism. He was a strong advocate of the Strasser brothers
form of National Socialism. When he died in 1976, The Times obituary
declared that Reed had his own hobby-horses, some of which were
indeed not so very unlike Nazi hobby horses. The writer specified the
allusion by mentioning Reeds virulent anti-semitism. It was not
merely that Reed was an anti-semite, but that he allied his anti-semitism
to conspiratorial fantasies. Reed, in books such as Far and Wide and The
Controversy of Zion, outlined his notions about Jewish plans to take
power of the world by subverting independent nations. As Thurlow
(1984) points out in his study of British fascism, Reed was one of the
592 Michael Billig
first anti-semitic writers to deny the reality of Hitlers persecution of the

Jews.
Today Reeds work is unread by mainstream political thinkers, but

continues to be much venerated by anti-semitic and fascist groups on the
extreme right. Reed is an especial favourite of the UK National Front,
which is currently following a Strasserite line. His books, alleging
Jewish conspiracies for world domination, are distributed in fascist
circles and frequently quoted in anti-semitic publications. For example,
Candour (July/August 1986) quoted conspiratorial notions from Reeds
Far and Wide in an effort to persuade readers that there is a very cogent
argument in support of the Conspiracy Theory (p. 62). The National
Fronts magazine, Nationalism Today, recently ran a series featuring
quotations from writers who laid the foundations of its ideology: issue
19 featured Reed and his ideas about Jews, conspiracies and Jewish
conspiracies. It is unsurprising that the sayings of Douglas Reed should

appear in National Front publications. It is more noteworthy, to say the
least, when they are quoted by Enoch Powell.
The procedures followed in this brief examination of Powell have

scarcely been methodological in the accepted sense. The starting point
was a single text, which needed to be interpreted in terms of other texts
before the ideological patterns could emerge. A methodologist might
prescribe a system of sampling other texts, but the scholar knows that as
much as possible should be read, lest something important be missed.
Nor can the results of reading be reduced to a quantitative matrix. A
single quotation might have more ideological significance than an oftrepeated one. George (1959), describing the analysis of Nazi propaganda
594 Michael Billig
during the war, stressed the importance of uniquely occurring messages.

Similarly, in the analysis of Powell, the ideological significance of the
quotation from Reed is not diminished by its unusualness; if anything, it
is increased by it.
Moreover, the scholar knows that the task of scholarship cannot be

reduced to getting through a list of set reading. It is not, for example,
merely a matter of ploughing through the collected works of Enoch
Powell, in the belief that all necessary reading then will have been
completed. Wider reading is also required. At all times in the foregoing
analysis there was an implicit comparison with conspiracy material and,
indeed, with the orthodox, non-conspiratorial versions of conservatism.
Thus, the analysis depended upon familiarity with wider material. Of
course it should not be presumed that, because the preceding analysis
sought to use the skills of traditional scholarship, it is satisfactorily
scholarly. In fact, it is a very limited sketch. A fuller analysis would have

required more details, and scholars must hope to deepen their expertise,
not merely as a result of having engaged in study, but as part of the
process of studying.
The main drawback of such traditional scholarship is that it places a

burden of responsibility upon the scholar. The procedures of
methodology make the individual expert anonymous, in the hope of
reducing the vagaries of individual bias. Yet this abolition of bias also
involves abolishing judgement. The traditional scholar cannot avoid the
task of judging whether a piece of evidence is important or not.
Moreover, the scholar cannot avoid responsibility for making
judgements which can be criticized by other scholars with different
views about the essential features of the issue in question. Thus,
596 Michael Billig
judgements about the patternings of ideology are potentially controversial. Scholars, with different complexions to their expertise, will
argue matters, putting different interpretations on each others evidence.
This is particularly true when one is asserting what are the essential
features of an ideological patterning. It cannot be expected that the
foregoing analysis is the last word on the subject of Enoch Powell and
the shape of his ideology. At best it raises further questions, even as it
suggests possibilities for answers.
Moreover, it must be recognized that the analyses of traditional scholarship, or the

judgements of traditional scholars, are potentially controversial. In this sense,
scholarship is located firmly within the domain of argumentative rhetoric (Billig, 1987;
Shotter, 1987). The traditional scholar, amongst many other responsibilities, also has
the responsibility not to shirk the possibility of receiving and administering criticism.
There are no neutral methodological procedures to hide behind. The analyses of others,
even of colleagues, must be criticized if the scholar finds their judgement to be
lacking. Moreover, each scholar exposes himself or herself to the danger of criticism,
especially to that most damaging accusation of being unscholarly. In consequence, the
so-called cosy world of scholarship is, or should be, a controversial place of criticism.
References
Abercrombie, N. (1980) Class, Structure, and Knowledge. Oxford: Basil Blackwell.
Abramson, L., M.E.P. Seligman and J. Teasdale (1978) Learned Helplessness in Humans: Critique and Reformulation, Journal
ofAbnormal Psychology, 87:49-74.
Adorno, T.W., E. Frenkel-Brunswik, D.J. Levinson and R.N. Sanford (1950) The Authoritarian Personality. New York: Harper
and Row.
Alvey Knowledge Based Systems Club, Explanation Special Interest Group (1987) Proceedings of the Second Workshop,
University of Surrey, England, January.
Anderson, N.H. (1968) Likeableness Ratings of555 Personality Trait Words, Journal of Personality and Social Psychology,
9: 272-9.
Antaki, C. (ed.) (1981) The Psychology of Ordinary Explanations of Social Behaviour. London: Academic Press.
Antaki, C. (1985) Ordinary Explanation in Conversation: Causal Structures and their Defence, European Journal of Social
Psychology, 15: 213-30.
Antaki, C. and C.R. Brewin (1982) Attributions and Psychological Change: Applications of Attributional Theories to Clinical and
Educational Practice. London: Academic Press.
Antaki, C. and A. Lewis (eds) (1986) Mental Mirrors: Metacognition in Social Knowledge and Communication, London: Sage.
Antaki, C. and S. Naji, Events Explained in Conversational Because Statements, British Journal of Social Psychology,
26:119-26.
Appelt, D.E. (1985) Planning English Sentences. Cambridge: Cambridge University Press.
Armistead, N. (ed.) (1974) Reconstructing Social Psychology. Harmondsworth: Penguin.
Atkinson, J.M. and J.C. Heritage (eds) (1984) Structures of Social Action: Studies in Conversation Analysis. Cambridge:
Cambridge University Press.
Austin, J.L. (1962) How to Do Things with Words. Oxford: Clarendon Press.
Bachnik, J. (1982) Deixis and Self/Other Discourse in Japanese, Sociolinguistic Working Papers, No. 99: 1-34.
Backman, C. W. (1985) Identity, Self Presentation, and the Resolution of Moral Dilemmas: Towards a Social and Psychological
Theory of Moral Behavior, pp. 251-89 in B.R. Schlenker (ed.), The Self and Social Life. New York: McGraw-Hill.
Barker, M. (1981) The New Racism. London: Junction Books.
Barthes, R. (1977) Image-Music-Text. London: Fontana/Collins.
Beardsworth, A. (1980) Analyzing Press Content: some Technical and Methodological Issues, Sociological Review
Monograph, 29: 371-95.
Beck, A.T. (1967) Depression. Philadelphia: University of Pennsylvania Press.
Becker, A.L. (1974) Person in Kawi, Oceanic Linguistics. 13: 229-55.
Bell, D. (1962) The Dispossessed, in D. Bell (ed.), The Radical Right. New York: Anchor Books.
Berkeley, H. (1977) The Odyssey of Enoch. London: Hamish Hamilton
598 References
.Bettelheim, B. (1976) The Uses of Enchantment. New York: Knopf.
Bettman, J.R. andB.A. Weitz(1983) Attributions in the Boardroom: Causal Reasoning in
Corporate Annual Reports, Administrative Science Quarterly, 28: 165-83.
Billig, M. (1978) Fascists: a Social Psychological View of the National Front. London:
Academic Press.
Billig, M. (1982) Ideology and Social Psychology. Oxford: Basil Blackwell.
Billig, M. (1986) Very Ordinary Life and the Young Conservatives, in H. Beloff (ed.),
Getting into Life. London: Methuen
Billig, M. (1987) Arguing and Thinking: a Rhetorical Approach to Social Psychology.
Cambridge: Cambridge University Press.
Billig, M. (1987a) Anti-semitic Themes and the British Far Left: some Social Psychological
Observations on Indirect Aspects of the Conspiracy Tradition, in C.F. Graumann and S.
Moscovici (eds), Changing Conceptions of Conspiracy. New York: Springer.
Billig, M. (1988) The Notion of Prejudice: some Rhetorical and Ideological Aspects,
Text, 8: 91-110.
Billig, M. (in press a) Historical and Rhetorical Aspects of Attitudes: the Case of the British
Monarchy, Philosophical Psychology.
Billig, M. (in press b) Common-places of the British Royal Family: a Rhetorical Analysis of
Plain and Argumentative Sense, Text.
Billig, M., S. Condor, D. Edwards, M. Gane, D. Middleton and A.R. Radley (forthcoming)
Ideological Dilemmas in Everday Thinking. London: Sage.
Bluck, S., E.J. Ballard, G. Baker-Brown, B. de Vries and P. Suedfeld (1985) The UBC
Scoring Manual for Integrative Complexity. Unpublished manuscript, Department
of Psychology, University of British Columbia, Vancouver.
Bowerman, W.R. (1981) Applications of a Social Psychological Theory of Motivation to the
Language of Defensiveness and Self-Justification, pp.00-00 in M.M.T. Henderson (ed.),
1980 Mid-America Linguistics Conference Papers, Lawrence, KA: University of
Kansas.
Brende, J.S. and E.R. Parson (1985) Vietnam Veterans: the Road to Recovery. New
York: Plenum.
Brewin, C.R. and C. Antaki (1987) An Analysis of Ordinary Explanations in Clinical
Attribution Research, Journal of Social and Clinical Psychology. 5: 79-98.
Brown, P. and S. Levinson (1978) Universal in Language Usage Politeness Phenomena, in
E. Goody (ed.), Questions and Politeness Strategies in Social Interaction.
Brown, R.W. and A. Gillman (1970) Pronouns of Power and Solidarity, chapter 1 in R.
Brown (ed.), Psycholinguistics. New York: Free Press.
Brunson, B.I. and K.A. Matthews (1981) Type A Coronary-prone Behaviour Pattern and
Reactions to Uncontrollable Stress: an Analysis of Performance Strategies, Affect, and
Attributions during Failure, Journal of Personality and Social Psychology, 40:
906-18.
Bryman, A. (1984) The Debate about Quantitative and Qualitative Research, British
Journal of Sociology, 35: 75-92.
Burleson, B., S. Wilson, M. Waltman, E. Goering, T. Ely and B. Whaley (November, 1986)
Item Desirability Effects in Compliance-gaining Research: Seven Empirical Studies
Showing why the Checklist Methodology Produces Garbage, paper presented to the
Speech Communication Association, Chicago, November.
Campbell, J. (1956) The Hero with a Thousand Faces (1949). New York: Meridian.
Carlson, L. (1983) Dialogue Games. Dordrecht: Reider.
Carroll, J.T. and J.W. Payne (1977) Judgments about Crime and the Criminals: a Model and
a Method for Investigating Parole Decisions, in B.D. Sales (ed.), Perspectives in Law
References 599
and Psychology (Vol. 1). New York: Plenum.
Cody, M.J. (1982) A Typology of Disengagement Strategies and an Examination of the Role
Intimacy, Reactions to Inequity and Relational Problems Play in Strategy Selection,
Communication Monographs , 49: 148-70.
Cody, M.J., D. Canary and S.W. Smith (in press) Compliance-gaining Goals, in J. Daly and
J. Wiemann (eds.), Communicating Strategically. Hillsdale, NJ: Erlbaum.
Cody, M.J. and M.L. McLaughlin (1985) Models for the Sequential Construction of
Accounting Episodes: Situational and Interactional Constraints on Message Selection
and Evaluation, pp. 50-69 in R.L. Street Jr and J.N. Cappella (eds.), Sequence and
Pattern in Communicative Behaviour. London: Edward Arnold.
Cohen, P.R. and C.R. Perrault (1979) Elements of a Plan-based Theory of Speech Acts,
Cognitive Science, 3: 177-212.
Cohler, B.J. (1979) Personal Narrative and Life-course, unpublished manuscript,
University of Chicago.
Cohn, N. (1967) Warrant for Genocide. London: Chatto/Heinemann.
Coleman, R.V. (1976) Court Control and Grievance Accounts: Dynamics of Traffic Court
Interaction, Urban Life, 5: 165-87.
Crano, W. and H.M. Schroder (1967) Complexity of Attitude Structure and Process of
Conflict Resolution, Journal of Personality and Social Psychology, 5: 110-4.
Cronbach, L.J. and P. Meehl (1955) 'Construct Validity in Psychological Tests,
Psychological Bulletin, 52: 281-302.
Davidson, J. (1984) Subsequent Versions of Invitations, Offers, Requests and Proposals
Dealing with Potential or Actual Rejection, in J.M. Atkinson and J.C. Heritage (eds.),
Structures of Social Action: Studies in Conversation Analysis. Cambridge:
de Waele, J.P. and R. Harrd (ed.) (1979), Personality. Oxford: Basil Blackwell.
Derrida, J. (1978) Writing and Difference. London: Routledge and Regan Paul.
Derrida, J. (1981) Positions. London: Athlone Press.
Dewey, J. (1975) Lectures on Psychological and Political Ethics (1898). New York:
Hafner Press.
Diener, C.T. and C.S. Dweck (1978) An Analysis of Learned Helplessness: Continuous
Changes in Performance, Strategy, and Achievement Cognitions Following Failure,
Journal of Personality and Social Psychology, 36: 451-62.
Doyle, J. (1978) Truth Maintenance Systems for Problem Solving, report AI-TR-419, AI
Laboratory, Massachusetts Institute of Technology.
Drew, P. (1984) Speakers Reportings in Invitation Sequences, in J.M. Atkinson and J.C.
Heritage (eds), Structures of Social Action: Studies in Conversation Analysis.
Driver, M.J. (1965) A Structural Analysis of Aggression, Stress and Personality in an Internation Simulation, paper 97, Institute for Research in the Behavioral, Economic, and
Management Sciences, Purdue University.
Duncan, S. and D.W. Fiske, (1977) Face to Face Interaction: Research, Methods and
Theory. Hillsdale, NJ: Erlbaum.
Eagleton, J. (1983) Literary Theory: an Introduction. Oxford: Basil Blackwell.
Edmonson, W. (1981) Spoken Discourse: a Model for Analysis. London: Longman.
Elig, T.W. and I.H. Frieze (1979) Measuring Causal Attributions for Success and Failure,
Journal of Personality and Social Psychology, 37: 621-34.
Ericcson, K. A. andH. A. Simon (1984) Protocol Analysis. Cambridge, MA: MITPress.
Feagin, J.R. (1972) Poverty: We Still Believe that God Helps Those that Help Themselves.
Psychology Today, 6: 101-29.
Feather, N.T. (1974) Explanations of Poverty in Australian and American Samples: the
600 References
Person, Society or Fate?, Australian Journal of Psychology, 26: 199-216.
Feuer, L.S. (1970) Ideology and the Ideologists. New York: Blackwell.
Figley, C.R. (1984) Catastrophes: An Overview of Family Reactions, pp. 222-48 in C.R.
Figley and H.I. McCubbin (eds), Stress and the Family: Coping with Catastrophe.
New York: Brunner-Mazel.
Fincham, F.D. (1985) Attributions in Close Relationships, pp. 203-34 in J.H. Harvey and
G. Weary (eds), Attributions: Basic Issues and Applications. New York: Academic
Press.
Fiske, S.T., D.A. Kenny and S.E. Taylor (1982) Structural Models for the Mediation of
Salience Effects on Attribution, Journal of Experimental Social Psychology, 18:
105-27
Fiske, S.T. and S.E. Taylor (1984) Social Cognition. New York: Random House.
Fodor, J.A. (1975) The Language of Thought. Harvard: Harvard University Press.
Foot, P. (1969) The Rise of Enoch Powell. Harmondsworth: Penguin.
Forsyth, D.R. (1980) The Functions of Attributions, Social Psychology Quarterly, 43:
184-9.
Frye, N. (1957) Anatomy of Criticism. Princeton: Princeton University Press.
Furnham, A. (1982) Why are the Poor Always With Us? Explanation for Poverty in Britain,
British Journal of Social Psychology, 21: 311-22.
Furnham, A. (1983) Attributions for affluence, Personality and Individual
Differences, 4: 31-40.
Furnham, A. and Lewis, A. (1986) The Economic Mind. Brighton: Wheatsheaf Books.
Gamble, A. (1986) The Political Economy of Freedom, in R. Levitas (ed.), The Ideology
of the New Right. Oxford: Polity Press.
Garfinkel, H. (1967) Studies in Ethnomethodology. Englewood Cliffs, NJ: Prentice-Hall.
George, A.L. (1959) Propaganda Analysis. Evanstone, IL: Row, Peterson.
Gergen, K. J. (1982) Toward Transformation in Social Knowledge. New York: Springer.
Gergen, K.J. and M.M. Gergen Narratives of the Self, in T.R. Sarbin and K.E. Schiebe (eds),
Studies in Social Identity. New York: Praeger.
Gergen, K.J. and M.M. Gergen (1984) Historical Social Psychology. Hillsdale, NJ:
Erlbaum.
Gergen, K.J. and M.M. Gergen (in press) Narratives and the Self as Relationship, in L.
Berkowitz (ed.), Advances in Experimental Social Psychology. London: Academic
Press.
Gergen, M. M. (1980) Antecedents and Consequences of Self-attributional Preferences in
Later Life, doctoral dissertation, Temple University.
Gergen, M.M. (in press) Towards a Feminist Metatheory and Methodology in the Social
Sciences, in M.M. Gergen (ed.), Feminist Thought and the Structure of
Knowledge. New York: New York University Press.
Giddens, A. (1979) Central Problems in Social Theory. London: Macmillan.
Gilbert, G.N. and P. Abell, (eds), (1983) Accounts and Action. Aldershot: Gower.
Gilbert, N. and M. Mulkay (1984) Opening Pandoras Box: a Sociological Analysis of
Scientists Discourse. Cambridge: Cambridge University Press.
Gilovich, T. (1983) Biased Evaluation and Persistence in Gambling, Journal of
Personality and Social Psychology, 44: 1110-26.
Goffman, E. (1955) On Face Work, Psychiatry, 18: 213-31.
Goffman, E. (1971) Relations in Public: Microstudies of the Public Order. New York: Basic
Books.
Goffman, E. (1981) Forms of Talk. Oxford: Oxford University Press.
Greenstein, F.I. (1975) Personality and Politics, in F.I. Greenstein and N. Polsby (eds),
Handbook of Political Science (vol. 2). Reading, MA: Addison-Wesley.
References 601
Grice, Fl.P. (1957) Meaning, Philosophical Review, 66: 377-88.
Grice, H.P. (1981) Presupposition and Conversational Implicature, in P. Cole (ed.),
Radical Pragmatics. New York: Academic Press.
Griffiths, R. (1983) Fellow Travellers of the Right. Oxford: Oxford University Press.
Groh, D. (1987). The Temptation to Conspiracy Theory, in C.F. Graumann and S.
Moscovici (eds), Changing Conceptions of Conspiracy. New York: Springer Verlag.
Habermas, J. (1975) The Theory of Communicative Action, Vol. 1: Reason and
Rationalisation in Society. Boston: Beacon Press.
Harre, R. (1979) Social Being: A Theory for Social Psychology, Oxford: Basil Blackwell.
Harre, R. (1981)Expressive Aspects of Descriptions of Others, inC.Antaki (ed.), The
Psychology of Ordinary Explanation of Social Behaviour. London: Academic
Press.
Harre, R. (1983a) Commentary from an Ethogenic Standpoint, Journal for the Theory
of Social Behaviour, 13: 69-73.
Harre, R. (1983b) Ethogenics, in R. Harre and R. Lamb (eds), The Encyclopaedic
Dictionary of Psychology. Oxford: Basil Blackwell.
Harre, R. and P.F. Secord (1972) The Explanation of Social Behaviour. Oxford: Basil
Blackwell.
Harvey, J.H., G. Agostinelli and C. Claiborn (1987) Attributional Mediators of Adaptation
in Vietnam Veterans, unpublished raw data. Iowa City: University of Iowa.
Harvey, J.H., R. Flanary and M. Morgan (1986) Vivid Memories of Vivid Loves Gone By,
Journal of Social and Personal Relationships, 3: 359-73.
Harvey, J.H. and K.S. Galvin (1984) Clinical Implications of Attribution Theory and
Research, Clinical Psychology Review, 4: 15-33.
Harvey, J.H. andG. Weary (1984) Current Issues in Attribution Theory and Research,
Annual Review of Psychology, 35: 427-59.
Harvey, J.H., A.L. Weber, K.S. Galvin, H. Hustzi and N. Garnick (1986) Attribution and the
Termination of Close Relationships: a Special Focus on the Account, pp. 189-201 in R.
Gilmour and S. Duck (eds), The Emerging Field of Personal Relationships.
Hillsdale, NJ: Erlbaum.
Harvey, J.H., A.L. Weber, K.L. Yarkin and B. Stewart (1982) An Attributional Approach to
Relationship Breakdown and Dissolution, pp. 107-26 in S. Duck (ed.), Dissolving
Personal Relationships London: Academic Press.
Harvey, J.H., G.L. Wells and M.D. Alvarez, (1978) Attribution in the Context of Conflict
and Separation in Close Relationships, in J.H. Harvey, W.J. Ickes and R.F. Kidd (eds),
New Directions in Attribution Research (vol. 2). Hillsdale, NJ: Erlbaum
Harvey, J.H., K.L. Yarkin, J.M. Lightner and J.P. Town (1980) Unsolicited Interpretation
and Recall of Interpersonal Events, Journal of Personality and Social Psychology,
38: 551-68.
Hastie, R. (1984) Causes and Effects of Causal Attribution, Journal of Personality and
Social Psychology, 46: 44-56.
Heider, F. (1958) The Psychology of Interpersonal Relations. New York: Wiley.
Heider, F. (1976) A Conversation with Fritz Heider, pp. 3-18 in J.H. Harvey, W.J. Ickes and
R.F. Kidd (eds). New Directions in Attribution Research (vol. 1). Hillsdale, NJ:
Erlbaum.
Henriques, J, W. Hollway, C. Urwin, C. Venn and V. Walkerdine (1986) Changing The
Subject. London: Methuen.
Heritage, J.C. (1984) Garfinkel and Ethnomethodology. Cambridge: Polity Press.
Heritage, J.C. (1985) Recent Developments in Conversation Analysis, Sociolinguistics,
15: 1-19.
Heritage, J.C. (forthcoming) Current Developments in Conversation Analysis, in P. Bull
602 References
and
D.
Roger
(eds), Interdisciplinary Approaches to Interpersonal

Bristol, Avon: Multilingual Matters.
Hewitt, J.P. and R. Stokes (1975) Disclaimers, American Sociological Review, 92: 11057.
Hill, C.T., Z. Rubin and L.A. Peplau (1976) Breakups Before Marriage: the End of 103
Affairs, Journal of Social Issues. 32: 147-68.
Horowitz, M.J. (1976) Stress Response Syndrome. New York: Aronson.
Janis, I.L. and L. Mann (1977) Decision Making. New York: Free Press.
Janis, I.L. and J. Rodin (1979) Attribution, Control and Decision Making: Social
Psychology and Health Care, pp. 487-521 in G. Stone, F. Cohen, N.E. Adler and
associates, Health Psychology: A Handbook. San Francisco: Jossey-Bass.
Jones, E.E. and K. Davis (1965) From Acts to Dispositions: The Attribution Process in
Social Perception, in L. Berkowitz (ed.), Advances in Experimental Social
Psychology (vol. 2). New York: Academic Press.
Jones, E.E. and D. McGillis (1976) Correspondent Inferences and the Attribution Cube:
aComparative Reappraisal, pp. 389-420 in J.H. Harvey, W.J. Ickes and R.F. Kidd (eds),
New Directions in Attribution Research (vol. 1). Hillsdale, NJ: Erlbaum.
Jones, E.E. and R. E. Nisbett, (1971) The Actor and the Observer: Divergent
Perceptions of the Cause of Behavior. Morristown, NJ: Silver Burdett/General
Learning Press.
Kelley, H.H. (1973) The Processes of Causal Attribution, American Psychologist, 28:
107-28.
Kelley, H.H. (1983) Perceived Causal Structures, in J. Jaspars, M. Hewstone and F.
Fincham (eds), Attribution Theory and Research: Conceptual, Developmental
and Social Distinctions. London: Academic Press.
Kohli, M. (1981) Biography: Account, Text, Method, in D. Bertaux (ed.), Biography and
Society, London: Sage.
Kruglanski, A.W. (1980) Lay Epistemologic Process and Contents: Another Look at
Attribution Theory, Psychological Review, 87: 70-87.
Kruglanski, A. (1987) Blame-placing Schemata and Attributional Research, in C.F.
Graumann and S. Moscovici (eds) Changing Conceptions of Conspiracy. New York:
Springer Verlag.
Labov, W. and D. Fanshel (1977) Therapeutic Discourse: Psychotherapy as
Conversation. New York: Academic Press.
Larkin, J., J. McDermott, D.P. Simon and H. Simon (1980) Expert and Novice Performance
in Solving Physics Problems, Science, 208: 1335-42.
Larrain, J. (1979) The Concept of Ideology London: Hutchinson.
Larrain, J. (1983) Marxism and Ideology. London: Macmillan.
Lau, R.R. (1984) Dynamics of the Attribution Process, Journal of Personality and
Lau, R.R. and D. Russell (1980) Attributions in the Sports Pages, Journal of Personality
and Social Psychology, 39: 29-38.
Layton-Henry, Z. (1984) The Politics of Race in Britain. London: George Allen and
Unwin.
Lehnert, W.G. (1978) The Process of Question Answering. Hillsdale, NJ: Erlbaum.
Levi, A. and P.E. Tetlock (1980) A Cognitive Analysis of the Japanese Decision to Go to
War, Journal of Conflict Resolution, 24: 195-212.
Levinson, S. (1979) The Essential Inadequacies of Speech Act Models, in H. Perret, M.
Sbissa and J. Verscheuren (eds), Possibilities and Limitations of Pragmatics.
Amsterdam: Benjamin.
Levinson, S. (1983) Pragmatics. Cambridge: Cambridge University Press.
Communication..
References 603
Lewis, A., Snell, M. and Furnham, A. (1987) Lay Explanations for the Causes of
Unemployment in Britain: Economic, Individualistic, Societal or Fatalistic?, Political
Lippman, S. (1986) Nothing but the Facts, Maam: The Impact of Testimony Construction
and Narrative Style on Jury Decisions, unpublished senior thesis, Swarthmore College.
Lipset, S.M. and E. Raab (1970) The Politics of Unreason. London: Heinemann.
Lord Raglan (1956) The Hero: a Study in Tradition, Myth, and Drama (1936). New York:
Vintage.
MacIntyre, A. (1981) After Virtue. South Bend, IN: University of Notre Dame Press.
Mannheim, K. (1960) Ideology and Utopia. London: Routledge and Kegan Paul.
Marsh, P., E. Rosser and R. Harre (1974) The Rules of Disorder. London: Routledge and
Kegan Paul.
McClosky, H. (1967) Personality and Attitude Correlates of Foreign Policy Orientation, in
J.M. Rosenau (ed.). Domestic Sources of Foreign Policy. New York: Free Press.
McDermott, D. and J. Doyle (1980) Non-monotonic Logic, Artificial Intelligence, 13:
41-72.
McGinn, C. (1983) The Subjective View. Oxford: Clarendon Press.
McLaughlin, M.L., M.J. Cody, and H.D. OHair (1983) The Management of Failure Events:
some Contextual Determinants of Accounting Behavior, Human Communication
Research, 9: 208-24.
McLaughlin, M.L., M.J. Cody and N.E. Rosenstein (1983) Account Sequences in
Conversations Between Strangers, Communication Monographs, 50: 102-25.
McLellan, D. (1986) Ideology. Milton Keynes: Open University Press.
Mead, G.H. (1973) Philosophy of the Act. Chicago: Chicago University Press.
Miller, F.D., E.R. Smith and J. Uleman (1981) Measurement and Interpretation of
Situational and Dispositional Attributions, Journal of Experimental Social
Miller, W.L. (1980) What was the Profit in Following the Crowd? The Effectiveness of
Party Strategies on Immigration and Devolution, British Journal of Political Science
15-38.
Mills, C.W. (1940) Situated Actions and Vocabularies of Motive, American Sociological
Review, 5: 904-13.
Moore, B. S., D. R. Sherrod .T.J.LivandB.U nderwood (1979) The Dispositional Shift in
Attribution over Time, Journal of Experimental Social Psychology, 15: 553-69.
Morris, G.H. (1985) The Remedial Episode as a Negotiation of Rules, pp.70-84 in R.L.
Street Jr and J.N. Cappella (eds), Sequence and Pattern in Communicative
Behaviour. London: Edward Arnold.
Mulkay, M. and N. Gilbert (1982) Accounting for Error: How Scientists Construct their
World when they Account for Correct and Incorrect Belief, Sociology, 16:165-83.
Moscovici, S. (1987) The Conspiracy Mentality, in C.F. Graumann and S. Moscovici (eds)
Changing Conceptions of Conspiracy. New York: Springer Verlag.
Nash, R. (1982) Measuring Up and Falling into Line: the Discourse of Maori Education',
paper presented to the NZARE Conference, December.
Neugarten, B.L. and G.O. Hagstad (1976) Age and the Life Course, in R.H. Binstock and
E. Shanas (eds), Handbook of Aging and the Social Sciences. New York: Van
Nostrand Reinhold.
Newell, A. (1980) One Final Word, pp. 175-89 in D.T. Tuma and F. Reif (eds), Problem
Solving and Education: Issues in Teaching and Research. Hillsdale, NJ: Erlbaum.
Newell, A., J.C. Shaw and H.A. Simon (1958) Elements of a Theory of Human Problem
Solving, Psychological Review, 65: 151-66.
Newell, A. and H.A. Simon (1972) Human Problem Solving. Englewood Cliffs, NJ:
604 References
Prentice-Hall.
Nisbett, R.E., D. Harvey and J. Wilson (1979) Epistemological Coding of the Content of
Everyday Social Conversations, unpublished manuscript, University of Michigan.
Nisbett, R.E. and L. Ross (1980) Human Inference. Englewood Cliffs, NJ: Prentice-Hall.
Norman, D. and S.W. Draper (1986) User Centred System Design. London: Erlbaum.
Novack, G. (1975) Pragmatism versus Marxism. New York: Pathfinder Press.
Oyamai, S. (1985) The Ontogeny of Information. New York: Cambridge University
Press.
Pennebaker, J.W. (1985) Traumatic Experience and Psychosomatic Disease: Exploring the
Roles of Behavioural Inhibition, Obsession, and Confiding, Canadian Psychology,
26: 82-95.
Peters, R.S. (1958) The Concept of Motivation. London: Routledge and Kegan Paul.
Peterson, C. and M.E.P. Seligman (1984) Causal Explanations as a Risk Factor in
Depression: Theory and Evidence, Psychological Review, 91: 347-74.
Poliakov, L. (1974) The Aryan Myth. London: Chatto/Heinemann.
Pollner, M. (1974) Mundane Reasoning, Philosophy of the Social Sciences, 4: 35-54.
Pollner, M. (1975) The very Coinage of your Brain: the Anatomy of Reality
Disjunctures, Philosophy of the Social Sciences, 5: 411-30.
Pollner, M. (1987) Mundane Reason: Reality in Everyday and Sociological Discourse.
Pomerantz, A.M. (1984) Agreeing and Disagreeing with Assessments, pp. 54-101 in J.M.
Atkinson and J.C. Heritage (eds), Structures of Social Action: Studies in
Conversation Analysis. Cambridge: Cambridge University Press.
Porter, C. A. and P. Suedfeld (1981) Integrative Complexity in the Correspondence of
Literary Figures: Effects of Personal and and Social Stress, Journal of Personality
and Social Psychology, 40: 321-30.
Potter, J. and F. Collie (1987) Community Care as Persuasive Rhetoric: a Study of
Discourse, unpublished manuscript, St Andrews University.
Potter, J. and M. Mulkay (1982) Making Theory Useful: Utility Accounting in Social
Psychologists Discourse, Fundamenta Scientiae, 34: 259-78.
Potter, J. and M. Mulkay (1985) Scientists Interview Talk: Interviews as a Technique for
Revealing Participants Interpretative Practices, in M. Brenner, J. Brown and D. Canter
(eds), The Research Interview: Uses and Approaches. New York1 AraHemir
Pr<=*Q
Potter, J. and S. Reicher (1987) Discourse of Community and Conflicts: the Organization of
Social Categories in Accounts of a Riot, British Journal of Social Psychology, 26:
25-40.
Potter, J., P. Stringer and M. Wetherell (1984), Social Texts and Context. London:
Routledge and Kegan Paul.
Potter, J. and M. Wetherell (1987) Discourse and Social Psychology: Beyond Attitudes
and Behaviour. London: Sage.
Powell, J.E. (1969) Freedom and Reality. London: Batsford.
Power, R. (1979) The Organisation of Purposeful Dialogues, Linguistics, 17:107-52.
Propp, V. (1968) Morphology of the Folktale (1928) (trans. and ed. Laurence Scott).
Austin: University of Texas Press.
Pruitt, D.G. and S.A. Lewis (1975) Development of Integrative Solutions in Bilateral
Negotiation, Journal of Personality and Social Psychology, 31: 621-33.
Putnam, H. (1975) The Meaning of Meaning, in Mind, Language and Reality.
Reason, J. (1987) The Chernobyl Errors, Bulletin of the British Psychological
Society, 40: 201-6.
References 605
Reeves, W. (1983) British Racial Discourse. Cambridge: Cambridge University Press.
Remler, J.E. (1978) Some Repairs on the Notion of Repairs, papers from the ninth regional
meeting of the Chicago Linguistic Society, pp. 391-402.
Rest, J.R. (1984) The Major Components of Morality, pp. 24-38 in W.M. Kurtines and J.L.
Gewirtz (eds), Morality, Moral Behaviour and Moral Development. New York:
Wiley.
Roberts, J.M. (1974) The Mythology of the Secret Societies. St. Albans: Paladin.
Rosaldo, R. (1986) Ilongot Hunting as Story and Experience, in V. Turner and E.
Bruner(eds), The Anthropology of Experience. Urbana: University of Illinois Press.
Ross, L. (1977) The Intuitive Psychologist and his Shortcomings: Distortions in the
Attribution Process, in L. Berkowitz (ed.), Advances in Experimental Social
Psychology (vol. 10). New York: Academic Press.
Rule, B.G. and G.L. Bisanz (1987) Goals and Strategies of Persuasion: a Cognitive
Schema, in M. Zanna, P. Herman and J. Olson (eds), Social Influence: the Fifth Ontario
Symposium in Personality and Social Psychology. Hillsdale, NJ: Erlbaum. Rule, B.G.,
G.L. Bisanz and M. Kohn (1985) Anatomy of a Persuasion Schema: Targets, Goals and
Strategies, Journal of Personality and Social Psychology, 48: 1127-40.
Sacks, H., E. Schegloff and G. Jefferson (1974) A Simple Systematics for the Organisation
of Turn-taking in Conversation', Language, 50: 696-735.
Sarbin, T.R. (ed.) (1986) Narrative Psychology: the Storied Nature of Human Conduct.
New York: Praeger.
Saussure, F. de (1974) Course in General Linguistics. London: Fontana.
Schegloff, E.A. (1968) Sequencing in Conversation Openings, American
Anthropologist, 70: 1075-95.
Schegloff, E.A. (1972) Notes on a Conversational Practice: Formulating Place, pp.
75-119 in D. Sudnow (ed.), Studies in Social Interaction. New York: Free Press.
Schegloff, E.A. (1988) Goffman and the Analysis of Conversation, in P. Drew and A.
Wootton (eds), Erving Goffman: an Interdisciplinary Appreciation. Oxford: Polity
Press.
Schegloff, E.A. and H. Sacks (1973) Opening Up Closings, Semiotica, 7: 289-327.
Schenkein, J. (ed.) (1985) Studies in the Organisation of Conversational Interaction.
New York: Academic Press.
Schoen, D. (1977) Enoch Powell and the Powellites. London: Macmillan.
Schonbach, P. (1980) A Category System for Account Phases', European Journal of Social
Psychology. 10: 195-200.
Schonbach, P. (1985) A Taxonomy for Account Phases: Revised, Explained, and Applied,
unpublished manuscript, Ruhr-Universitat Bochum.
Schonbach, P. (1986) A Theory of Conflict Escalation in Account Episodes, unpublished
manuscript, Ruhr-Universitat Bochum.
Schroder, H.M., M.J. Driver and S. Streufert (1967) Human Information Processing.
New York: Holt Rinehart and Winston.
Scott, M. and S. Lyman (1968) Accounts, American Sociological Review, 33: 46-62.
Searle, J. (1969) Speech Acts. Cambridge: Cambridge University Press.
Searle, J. (1975) Indirect speech acts, pp. 59-78 in P. Cole and J.L. Morgan (eds), Syntax
and Semantics: Speech Acts (vol. 3). New York: Academic Press.
Searle, J., F. Kiefer and M. Bierwish (eds) (1980) Speech Act Theory and Pragmatics.
Dordrecht: Reidel.
Seibold, D.R., N.K. Baym, C.R. Berteotti, S.C. Bergener, S.A. McCornack, L.P. McQuillan
and M. Rivers (1986) Garbage or Gar bag - Its All in Where you Place the Accent. A
Response to Burleson et al.s Item Desirability Effects in Compliance-gaining
Research: Seven Empirical Studies Showing why the Checklist Methodology Produces
606 References
Garbage, paper presented to the Speech Communication Association, Chicago,
November.
Seidel, G. (1986) The Holocaust Denial. Leeds: Beyond the Pale.
Shoeneman, T.J. and D.E. Rubanowitz (1983) Attributions in the Advice Columns: Actors
and Observers, Causes and Reasons, paper presented at the meeting of the American
Psychological Association, Anaheim, CA, August.
Shotter, J. (1975) Images of Man in Psychological Research. London: Methuen.
Shotter, J. (1980) Action, Joint Action and Intentionality, in M. Brenner (ed.), The
Structure of Action. Oxford: Basil Blackwell.
Shotter, J. (1983) Hermeneutic Interpretative Theory, in R. Harre and R. Lamb (eds), The
Encyclopaedic Dictionary of Psychology. Oxford: Basil Blackwell.
Shotter, J. (1984) Social Accountability and Selfhood. Oxford: Basil Blackwell.
Shotter, J. (1987) Rhetoric as a Model for Psychology, paper given to the British
Psychological Society conference on the future of the psychological sciences, Harrogate.
Silverstein, M. (1985) Who shall Regiment Language? unpublished manuscript.
Snyder, C.R. and R.L. Higgins (1986) Excuses: their Effects and their Role in the
Negotiation of Reality, paper presented to the second attribution-personality theory
conference, Los Angeles, September.
Snyder, M. (1979) Self-monitoring: The Self in Social Action, in L. Berkowitz (ed.),
Advances in Experimental Social Psychology. New York: Academic Press.
Spence, D.P. (1982) Narrative Truth and Historical Truth. New York: Norton.
Staw, B.M. (1980) Rationality and Justification in Organizational Life', in B.M. Staw and L.
Cummings (eds), Research in Organizational Behavior (vol. 2). Greenwich, CT: JAI
Press.
Steed, M. (1975) The Results Analysed, in D. Butler and D. Kavanagh, The British
General Election of October 1974. London: Macmillan.
Stone, W.F. (1980) The Myth of Leftwing Authoritarianism, Political Psychology, 2: 320.
Streufert, S. and W. Fromkin (1972) Complexity and Social Influence, in J.T. Tedeschi
(ed.), Social Influence Processes. Chicago: Aldine.
Streufert, S. and S. Streufert (1978) Behavior in the Complex Environment. Washington,
DC: Winston.
Streufert, S., and R.W. Swezey (1986) Complexity, Managers, and Organizations. New
York: Academic Press.
Studlar, D.T. (1974) British Public Opinion, Colour Issues and Enoch Powell: a
Longitudinal Analysis, British Journal of Political Science, 4: 371-81.
Studlar, D.T. (1978) Policy Voting in Britain: the Coloured Immigration Issue in 1964, 1966
and 1970 General Elections, American Political Science Review, 72: 46-72.
Suedfeld, P. (1978a) Characteristics of Decision Making as a Function of the Environment,
pp. 203-14 in B. King, S. Streufert and F. E. Fiedler (eds), Managerial Control and
Organizational Democracy. Washington, DC: Winston.
Suedfeld, P. (1978b) Die Messung integrativer Komplexitat in Archivmaterialen (The
Measurement of Integrative Complexity in Archival Materials), pp. 179-92 in H. Mandl
and G.L. Huber (eds), Kognitive Komplexitat. Gottingen: Hogrefe.
Suedfeld, P. (1981) Indices of World Tension in The Bulletin of Atomic Scientists,
Political Psychology, 2: 114-23.
Suedfeld, P. (1985) APA Presidential Addresses: The Relation of Integrative Complexity to
Historical, Professional, and Personal Factors, Journal of Personality and Social
Suedfeld, P. (1985b) Stressful Levels of Environmental Stimulation, pp. 83-104 in I.G.
SarasonandC.D. Spielberger (eds), Stress and Anxiety. Vol. 10: A Sourcebook of
References 607
Theory and Research. Washington, DC: Hemisphere.
Suedfeld, P. and L.E. Piedrahita (1984) Intimations of Mortality: Integrative Simplification
as a Precursor of Death, Journal of Personality and Social Psychology, 47: 848-52.
Suedfeld, P. and A.D. Rank (1976) Revolutionary Leaders: Long-term Success as a
Function of Changes in Conceptual Complexity, Journal of Personality and Social
Suedfeld, P. and P.E. Tetlock (1977) Integrative Complexity of Communications in
International Crises, Journal of Conflict Resolution, 21: 169-84.
Suedfeld, P., P.E. Tetlock and C. Ramirez (1977) War, Peace, and Integrative Complexity:
United Nations Speeches on the Middle East Problem, Journal of Conflict
Resolution, 21: 427-42.
Taylor, S. (1982) The National Front in English Politics. London: Macmillan.
Taylor, S.E., R.R. Lichtman and J.V. Wood (1984) Attributions, Beliefs about Control, and
Adjustment to Breast Cancer, Journal of Personality and Social Psychology, 46:
1192-211.
Tedeschi, J.T. and N. Norman (1985) Social Power, Self-presentation, and the Self, in B.R.
Schlenker (ed.), The Self and Social Life. New York: McGraw-Hill.
Tetlock. P.E. (1979) Identifying Victims of Groupthink from Public Statements of Decision
Makers, Journal of Personality and Social Psychology, 37: 1314-24.
Tetlock, P.E. (1981a) Personality and Isolationism: Content Analysis of Senatorial
Speeches, Journal of Personality and Social Psychology, 41: 737-43.
Tetlock, P.E. (1981b) Pre- to Post-election Shifts in Presidential Rhetoric: Impression
Management or Cognitive Adjustment? Journal of Personality and Social
Tetlock, P.E. (1983a) Policy Makers Images of International Conflict, Journal of Social
Issues, 39: 67-86.
Tetlock, P.E. (1983b) Psychological Research on Foreign Policy: a Methodological
Overview, pp. 45-79 in L. Wheeler (ed.), Review of Personality and Social
Psychology (vol. 4). Beverly Hills: Sage.
Tetlock, P.E. (1984) Cognitive Style and Political Belief System in the British House of
Commons, Journal of Personality and Social Psychology, 46: 365-75.
Tetlock, P.E. (1985) Integrative Complexity of American and Soviet Foreign Policy
Statements: a Time-series Analysis, Journal of Personality and Social Psychology,
49: 1565-85.
Tetlock, P.E. (1986) A Value Pluralism Model of Ideological Reasoning, Journal of
Tetlock, P.E., J. Bernzweig and J.L. Gallant (1985) Supreme Court Decision Making:
Cognitive Style as a Predictor of Ideological Consistency of Voting, Journal of
Tetlock, P.E., F. Crosby and T. Crosby (1981) Psychobiography, Micropolitics, 2: 58-81.
Tetlock, P.E. and K. Hannum (1984) The Berkeley Integrative Complexity Scoring
Manual. Unpublished manuscript, Department of Psychology, University of California,
Berkeley.
Tetlock, P.E., K. Hannum and P. Micheletti (1984) Stability and Change in Senatorial
Debate: Testing the Cognitive versus Rhetorical Style Hypotheses, Journal of
Tetlock, P.E. and A.S.R. Manstead (1985) Impression Management versus Intrapsychic
Explanations in Social Psychology: a Useful Dichotomy? Psychological Review, 92:
59-77.
Thompson, J. (1984) Studies in the Theory of Ideology. Cambridge: Polity Press.
Thurlow, R. (1984) Anti-Nazi, Anti-semite, Patterns of Prejudice, 18: 23-34.
608 References
Thurlow, R. (1987) Fascism in Britain. Oxford: Blackwell.
Tololyan, K. (in press) Cultural Narrative and the Motivation of the Terrorist, Journal of
Strategic Studies.
Totman, R. (1982) Philosophical Foundations of Attribution Therapies, in C. Antaki and C.

Brewin (eds), Attributions and Psychological Change. London: Academic Press.
Toulmin, S.E. (1958) The Uses of Argument. Cambridge: Cambridge University Press.
van Dijk, T.A. (1984) Prejudice in Discourse: an Analysis of Ethnic Prejudices in Cognition
and Conversation. Amsterdam: Benjamin.
van Fraassen, B.C. (1980) The Scientific Image. Oxford: Clarendon Press.
Voss, J.F., J. Blais, M.L. Means, T.R. Greene and E. Ahwesh (in press) Informal Reasoning
and Subject Matter Knowledge in the Solving of Economics Problems by Naive and
Novice Individuals, Cognition and Instruction.
Voss, J.F., T.R. Greene, T.A. Post andB.C.Penner(1983)Problem Solving Skill in the Social
Sciences, pp. 165-213 in G.H. Bower (ed.), The Psychology of Learning and
Motivation: Advances in Research Theory (vol. 17). New York: Academic Press.
Voss, J.F., S.W. Tyler and L.A. Yengo (1983) Individual Differences in the Solving of Social
Science Problems, pp. 205-32. in R.F. Dillon and R.R. Schmeck (eds), Individual
Differences in Cognition. New York: Academic Press.
Webber, G.C. (1986) The Ideology of the British Right, 1918-1939. London: Croom
Helm.
Weber, A.L., J.H. Harvey and M.A. Stanley (in press) The Nature and Motivations of
Accounts for Failed Relationships, in R. Burnett, P. McGhee and D. Clark (eds),
Accounting for Relationships. London: Methuen.
Weiner, B. (1985) Spontaneous Causal Thinking, Psychological Bulletin, 97:74-84.
Weiner, B. (1986) Attribution, Emotion and Action, in R.M. Sorrentino and E.T. Higgins
(eds), Handbook of Motivation and Emotion. New York: Guildford Press.
Weiss, R.S. (1975) Marital Separation. New York: Basic Books.
Wertsch, J.V. (ed.) (1985) Culture, Communication and Cognition. Cambridge:
Wetherell, M.S. (1986) Linguistic Repertoires and Literary Criticism: New Directions for a
Social Psychology of Gender, in S. Wilkinson (ed.), Feminist Social Psychology.
Milton Keynes: Open University Press.
Wetherell, M. and J. Potter (1986a), Discourse Analysis and the Social Psychology of
Racism, Social Psychology Section Newsletter, 15: 24-30.
Wetherell, M. and J. Potter (1986b) Majority Group Representations of Race and Race
Relations, paper presented at the BPS Social Psychology Section annual conference,
University of Sussex, September.
Wetherell, M. and J. Potter (forthcoming) Mapping the Language of Racism. London:
Sage.
Wetherell, M., H. Stiven and J. Potter (1987) Unequal Egalitarianism: a Preliminary Study
of Discourses Concerning Gender and Employment Opportunities, British Journal of
White, H. (1980) The Value of Narrative in the Representation of Reality, Critical
Inquiry, 7: 5-28.
Wittgenstein, L.W. (1953) Philosophical Investigations. Oxford: Basil Blackwell.
Wong, P.T.P. amd B. Weiner (1981) When People Ask Why Questions, and the Heuristics
of Attributional Search, Journal of Personality and Social Psychology, 40: 650-63.
Wood, D.C. (1979) An Introduction to Derrida, Radical Philosophy, 21: 18-28.
Wootton, A.J. (1981) The Management of Grantings and Rejections by Parents in Request
Sequences, Semiotica, 37: 59-89.
Yarkin, K.L. and J.H. Harvey (1982) The Structure of Attribution, Journal of Social
Yarkin, K.L., J.H. Harvey and B.M. Bloxom (1981) Cognitive Sets, Attribution, and Social
References 609
Interaction, Journal of Personality and Social Psychology, 41: 243-52.
Yearley, S. (1985) Vocabularies of Freedom and Resentment: a Strawsonian Perspective on
the Nature of Argument in Science and Law, Social Studies of Science, 15: 99-126.
Young, K.G. (1987) Taleworlds and Storyrealms. Boston: Martinus Nijhoff.
Zukier, H. (1987) Conspiracy as a Crime in

Solidarity, in C.F. Graumann and S. Moscovici
(eds), Changing Conceptions of Conspiracy. New York:
Springer.Index
accountability, 5, 128, 156-67; interactional,

138, 144; and social relations in
Japanese, 160-6
accounts,30-1,34,94-112,113-26,127-44,
185; analysis, 121-3; coding of defendants,
118-21; for combat stress, as unstructured
attributes example, 37-41; role of, 140-1;
and self-other relations, 136-8; and the
structure of social action, 139-41; on trial,
113-26; types, with respect to imposition of
penalty, 123 adjacency pairs, 129,147-8
Adorno, T.W., 201 advice systems, 145
aggravation-mitigation continuum, 116-17
Agostinelli, Gina, 32-42 Alvarez, M.D., 35
ambiguity, 22-4, 25
Antaki, Charles, 1-14, 22, 34, 35, 60-73,
145-55
Appelt, D.E., 17 Archers,
The, 191-6 argument,
structure of, 21-2 Aristotle,
76
artificial intelligence, 15, 20, 26, 145
attribution theory, 5, 7-8, 9, 32-42, 60-1, 94;
data, 11-12; methods used, 10-11
attributions: classification of, 34; coding,
40; free response, see attributions,
unstructured; identifying, 3242;
unstructured, 32,
34-7; analysis
of, 39-41
Bachnik, J., 161, 163
backing, 21,76 Backman,
C.W., 114 Beardsworth, A.,
206 because, 19-20 Becker,
A.L., 166
behaviour, power of explanations to change,
1,4
belief, structures of, 60-73; defined, 62-3
beliefs: frozen, 33; information and, 20-1;
studies in causal, 30; see also attribution

theory Bettman, J.R., 35 biased reporting, 3,
123-5 Billig, Michael, 199-215
biographies, see self-narratives
Bisanz, G.L., 123 Bowerman,
W.R., 72 Brewin, C.R., 34
Brown, R.W., 157-9
Campbell, Joseph, 95 Canary, D., 123
Carlson, L., 152 Carroll, J.T., 35 categories,
explanatory, 8 causality 28; beliefs about,
30; questions about, 18-20 causes, 24, 31
challenges, 119 claims, 21, 76 clines, 161,
162 coding skills, 14 Cody, Michael, J.,
113-26 cognitive adaptation, versus
rhetorical style, 50-4
cognitivist tradition, see also attribution
theory; social cognition theories comedy,
98, 100 communication, 1-14, 146-7
community talk, 170, 172-3 completion,
145-55; cognitive, 149; internal, 150-3;
structural, 148 complexity, conceptual, 44-5
complexity coding, integrative, of verbal
behaviour, 43-59; system, 44-50 complexity
theory, integrative, the future for, 58-9
concessions, 114, 120, 122
confiding, 33, 34
conservatism, signs of conspiracy mentality
in, 203-14
conspiracy theory of politics, 201-14
construction, of discourse, 171-2 content
analysis, 206-7 context, and meaning, 17
conversation, completion and, 146-8,
dynamics in, 153-4
conversation analysis, 12, 15-31, 127-44,
146-8; data collection and analysis, 1301, 141-4
conversational interaction,
analysis of sequencing, 128-3
Index 611
0conversational structures, 64-8; causal, 657; justifications, 67-8 counter-arguments, 77
courtroom observation, 4, 118-23
covariance analysis (Kelley), 7, 14 culture
fostering, 178-80
Danish language, 158 data (Toulmin), 21,
76 Davis, K., 61 de Waele, J.P., 95 decoding
skills, 14 deconstructing accounts, 184-98
deconstruction, definition, 186-9 deixis,
160-1, 163-4 descriptions, explanatory, 17
deviant case analysis, 131, 133-5 Dewey, J.,
150 differentiation, 43 disclaimer, 176-7
discourse analysis, 9, 168-83; definition,
169-73
dispositional shift, 36 disruptive stress
hypothesis, 52 Draper, Stephen W., 15-31,
145 Drew, P., 170 Driver, M.J., 45
dynamics, in explanation seeking, 145-55
effects, and intentions, 27-8 English
language, 156,157, 159-60 enthymeme, 76
Ericsson, K.A., 74 ethnography of speaking,
160 ethnomethodology, 9, 10, 127-44
events, naturally occurring, unstructured
attributions in, 35-6
excuses, 4,30-1,113-15,120-1; definition,
114
explanation: OED definition, 17; privately
held, 4-5; in public domain, 1-4; social
function of 28: types of, 6-8 explanations,
everyday, analysis of, 15-31, 196; as
wholes, 60
failure events, accounting and, 113-26
false starts, 151
Fanshel, D., 116-17
fascism, 203, 213-14
Fox, George, 159
French, Norman, 159
Frye, Northrop, 98
Garfinkel, H., 127-8
George, A.L. 214 Gergen,
Mary M., 94-112 Gilbert,
N 172 Gillman, A., 157-9
grammar, determination of thought thesis,
167; see also pronouns, role of Grice, HP.,
17
Harre, Rom, 9, 73, 95, 156-67 Harvey, John
H., 32-42 Heider, Fritz, 61, 94 Heritage,
John, 127-44 hermeneutic circle of
meaning, 185 Higgins, R.L., 115
idea units, in protocol analysis, 83-90

ideology: definitions, 200; explanations as,
190,199-215
illocutionary acts, 26, 116-17, 147
impression management strategy, and
manipulation of integration complexity
level, 57
indexicality, 156,161
information: and belief, 20-1; content of
explanations, 16-17
integration, 43-4, see also complexity
theory, integrative
intelligibility, social of action, 139-40
intentions, effects and, 24, 27-8, 187
interpretative repertoires, 172-3; examples,
177-82; identification of, 168-83
interpretative tradition, 11
Japanese language, 157,160-5 Jones, E.E.,
61 journalism, 3,12-13
justification, 4, 24, 30-1, 63, 113, 121;
definition, 114; structures of belief and,
60-73; see also accountability;
accounts
Kami (old Javanese), 166 Kelley,
H.H.,7,61,72 Kimura, Shinya, 160, 164-5
knowledge, socially distributed nature of,
20-1
Kruglanski, A.W., 33 Labov, W., 116-17
language: encodes social orders, 156-67;
function, 168-71; naturally occurring,
11-12, 28-9; structuralist conception of,
185, 186-7
language of thought, 152 Latin, 158
leaders, revolutionary, studies of, 45, 50-4
Lehnert, W.G., 18
Leudar, Ivan, 145-55
Levinson, S., 147
linguistics, 9-10; see also language
Lipset, S.M., 201
literary theory, 184, 186
logical proofs, 120, 121-3, 125-6
Lyman, S., 113
McLaughlin, Margaret L., 113-26
McLellan, D 200 Mani, Vimula, 165
Mannheim, K., 200 markers, see surface
cues Mead, G.H., 145,146,150-3 meaning,
shared, 29,150 memory, 34, 36
metacognitive background of explanations,
12-14
methodology, in understanding ideological
explanation, 199-215; definition, 199-
612 Index
200
Miller, W.L., 205 mitigation, 144,116-17
Mulkay, M 172 myths, 95-6
Naji, S. 22, 35
narrative forms, 97-102; progressive, 99100; regressive, 99-100; stability, 99
narrative structures, in social explanation,
94-112
Newell, A., 74, 77 Norman, D 115,145
normative aspects of conversation analysis,
128, 129
Norwegian language, 163
observation, 123-4 open souls doctrine, 73
openness, 147-8
operators, of problem-solving and reasoning
structures, 83
oral explanations: identifying attributions in
32-42; in traffic court, 113-26; see also
verbal behaviour Oxford English
Dictionary, 17
Paragraph Completion Test, 43-5
paraphrases, 17-18
Parker, Ian, 184-98

Payne, J.W., 35
perlocutionary effect, 26
personality: psychometric approach, 110;
social constructionist approach, 102-12
polarities in deconstruction, 188-9, 195-6
politics: conspiracy theory of, 201-14;
discourse of, 196-8 post-structuralism,
187 Potter, Jonathan, 168-83 Powell, J.
Enoch, 204-14 power, meaning in terms of,
184-5, 186, 189, 196-8 Power, R., 26
pragmatic realism, 178,180-1 pragmatics,
145,146
problem-solving, and reasoning in illstructured domains, 74-93 pronouns, role
of, 156-67 propaganda, 2-3 Propp,
Vladimir, 94 protocol analysis, 74-93
psychohistorians, 54-5 psycholinguistics,
157 Putnam, H., 20
Quakers, 160 qualifiers, 77 question threat,
124
questions: about causation, analysis of, 1819; cognitivist, 8-9; pedagogic and
sincere, 27; relevance answers,
18,140
Raab, E.,201
race relations; in Britain, 204-5; construction of, 173-82 Raglan, Lord, 95 Ramirez,
C., 55-6 Rank, A.D., 45, 50-2, 56 reason

maintenance, 20, 22 reasoning: in illstructured domains, 74-93;
and social action, 127 reasons, offered
in explanations, 20-6 reduction of
explanations into simple causal categories,
61-2 Reed, Douglas, 213 refusals, 114, 119
register, 161
reinterpretation, 189, 196 repair, 151
research, explanations, 5-12 response bias,
self-presentational, 41 response sheets,
fixed-format, 10-11, 12
Rest, J.R., 114
rhetoric, argumentative, 9, 215 rhetorical
style, versus cognitive adaptation, 50-4
romance, 98, 100 Rosaldo, R., 95 Rule,
B.G., 123 satire, 98, 100 Schegloff, E.A.,
136
scholarship, in understanding ideological
explanation, 199-215; definition, 200
Schonbach, P., 113-14 Schroder, H.M., 45
Scott, M., 113 Secord, P.F., 73 selfcompletion, 146, 150-3 self-narratives, 10012 self-presentation, 114, 118 Silverstein,
Michael, 159 Simon, H.A., 74 Smith, S.W.,
123 Snyder, C.R., 115 social action,
reasoning and, 127-44 social cognition,
defined, 5 social cognition theories, 1-14;
assessment of, 13-14; types, 9-10 social
constructionism, 157, 166; see also social
explanation study social explanation study,
94-112 social psychology,7,61,94,184;new,
184-6 social relations, and accountability in
Japanese, 160-5
solidarity, in grammar of pronouns, 157-8
speech, distinction from writing, 187-8
speech act theory, 147,167
Spence, Donald, 108
status, in grammar of pronouns, 157-8
story line method, 99,102-10
Streufert, S., 45
structuralism, 185-7
Studlar, D.T., 205
Suedfeld, Peter, 43-59
surface cues, 17-20
Tamil, 165 Tedeschi, J.T., 115 Tetlock,
Philip E., 43-59 text: boundaries of the,
197-8; nature of, 189-90; selection and
presentation of, 206-8; use of term, 188
Thurlow, R., 213 togetherness, 178, 181-2
Tololyan, K., 96 Totman, R., 61 Toulmin,
S.E., 21, 76-7 traffic court, oral arguments
in, 113-26 tragedy, 98, 100
transcription, problem of, 29-30, 41, 182
truth maintenance, 20, 22 Turnquist, Dawn
Index 613
C., 32-42 turns, extended, 151 TV soaps,
190-6
US presidential candidates, changes in
rhetorical style, 52-4 utterances, 28-9,146-7
variation, 171,174-7
verbal behaviour, integrative complexity
coding, 43-59; see also oral explanations
Vollmer, Fred, 163 Voss, James F., 74-93
Vygotsky, L.S., 152
warrant, (Antaki), 62-3, 67-8, (Draper), 22,

24, 31, (Toulmin), 21, 76 Weiner, B., 37
Weiss, R.S., 35 Weitz, B.A., 35 Wells,
G.L.,35 Wetherell, Margaret, 168-83
Wittgenstein, Ludwig, 15 Wong, P.T.P., 37
writing, distinction from speech, 187-8
written explanations, identifying attributions in, 32-42
written structures, 68-72;

causal, 69-70; justifications,
70-
2
SAGE Publications
London Newbury Park Beverly Hills New Delhi

Antaki

Uploaded by

Copyright:

Available Formats

Antaki

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Antaki

Uploaded by

Copyright:

Available Formats

ANALYSING

SAGEExplanations are given and received in all areas of

which will prove invaluable to all social scientists.

Jacket design by The Pinpoint Design Company

SAGE Publications 28 Banner Street London

Edited by Charles Antaki

London Newbury Park Beverly Hills New Delhi

SAGE Publications Inc 2111 West

SAGE Publications Inc 275 South

Analysing everyday explanation: a casebook of methods.

Printed in Great Britain by Billing and Sons Ltd, Worceste

Explanations, communication and social cognition 7

1 Explanations, communication and social cognition

STRUCTURE AND CONTENT

3 Identifying attributions in oral and written explanations John H.

5 Structures of belief and justification

6 Problem solving and reasoning in ill-structured domains

7 Narrative structures in social explanation

III ACCOUNTS IN CONTEXT

8 Accounts on trial: oral arguments in traffic court

9 Explanations as accounts: a conversation analytic

10 Completion and dynamics in explanation seeking

11 Accountability within a social order: the role of pronouns

IV RHETORIC AND IDEOLOGY

12 Discourse analysis and the identification of

14 Methodology and scholarship in understanding

Explanations, communication and social cognition 9

Gina Agostinelli, Department of Psychology, Spence Laboratories of

Michael Billig, Department of Social Sciences, University of

Stephen W. Draper, Department of Psychology, University of Glasgow,

Explanations, communication and social cognition 11

Mary M. Gergen, Department of Psychology, Pennsylvania State

John H. Harvey, Department of Psychology, Spence Laboratories of

Ivan Leudar, Department of Psychology, University of Manchester,

Margaret L. McLaughlin, Communication Arts and Sciences,

Explanations, communication and social cognition 13

Ian Parker, Department of Psychology and Speech Pathology,

describe ways of dealing with the explanations that we give and

Some contributors are concerned with explanation as an index of a

Explanations, communication and social cognition 15

These three families of interests are represented in separate parts of

Explanations, communication and social cognition 17

Explanation can mean many things, and ordinary explanations can be

In this chapter I want to make some remarks about how ordinary

Explanations, communication and social cognition 19

We can choose to look into the explanations in these examples in many

Explanations, communication and social cognition 21

What is immediately noticeable about both explanations is the

legitimating. In both cases the explanation has the power to redefine

We can pursue this feature into explicitly propagandistic literature to

Explanations, communication and social cognition 23

The explanation is often a reprise of the general message of the

Seeing the partisan flavour of explanations in literature which is

are quick to exploit it in headlines and leading paragraphs. In the paper

In certain extreme cases, there is no need for subtle analysis, as this

Explanations, communication and social cognition 25

example about the Libyan embassy shootings of April 1984 shows:

The revelatory power of explanations is something that has made

Explanations, communication and social cognition 27

So far we have treated explanations as being offered by someone to