Sunteți pe pagina 1din 112

DISCOURSE AND

SOCIAL PSYCHOLOGY
Beyond Attitudes and Behaviour

Jonathan Potter and Margaret Wetherell


·<

-- -~ --- '' --. ··.

®
Sage Publications
London. Newbury Park. Beverly Hills. New Delhi
© !987 Jonuthan Pottcr and Margare! Wcthcrdl Contents
First Publishcd 1987

All right.~ rcscrved. No pan of Lhi~ bunk muy be rcproduccd or utilizeU in any fonn nr by Introductian
any mcans, electronic nr mechanical, including phntocopying, rccording, or by uny infonmninn Titree iilu.strarive examples 2
:aoragc and rctricvul systcm. withnut pcrmission in writing from thc Publishcrs. Definitions of discourse analysis 6

1. Foundations of discaursc analysis 9


Choms/..1' and psychology 9
SAGE Publicutions Ltd Word.s as deeds: speech act them}' 14
28 Banncr Strcct
'Doing' talking: etlmomethodnlogy 18
London EC 1Y 8QE
Signs of structure: semiology 24
SAGE Publicmions Inc Social srudy ~f language 28
275 South Bcvcrly Orive Problems ami limitations 29
Bcvcrly Hills. California 902!2
und 2.'',- Unfolding discourse analysis 32
SAGE Publications lnc Some majar components of discourse analysis 32
2111 Wcst Hillcrcst Orive Variable discourse ami traditional socinl psychology 36
Ncwbury Purk, California 91320
Tiw suppression of accowlf variabiliry 39
SAGE Publicutions .India Pvt Ltd Altitudes in discour.se 43
C-236 Dcfcncc Cnlnny Beyond altimdes 53
New Dclhi 110 024
3 _ Making rules work 56
77w etlwgenic perspective 56
Problems with ethogenics 61
British Llbrury Cululoguing in Publicution Duta Discourse analysis and the srudy of sciellfisrs' mies 64
Testabiliry, .wccer violence all(/ the wwlysis of rules 71
Pottcr, Jonathan
Discoursc and social psychology : bcyond
altitudes and bchaviour. 4. Accounts in scquence 74
l. Oral communication - Social aspccts Accounts and social psychology 74
l. Titlc JI. Wetherell. Margare! Com'ersation analy.sis and accounts 80
302.2'24 P95 Accounrs in cqurt 87
Discourse and accowus 93
Library of Congrcss Catalogue Curd Numbcr 87-060198
-5. .Speaking subjccls 95
\.:ISBN 0-8039-8055-8
ISBN 0-8039-8056-6 Pbk
· . !_-..· Tradirional images of the se/f 95
1 Towards a ne\\' conception of the subject 101
The ideological self 108
Phowtypesct by Sys\cm 4 Associatcs, Gcrmrds Cross, Buckinghamshirc Di.scourse analysis CJl work on rhe .relf 110
Printcd in Grcut Brita!n by J. \V. Arrowsmith Ltd., Bristol 1
-~r·-\
116
~~)
Catcgorics in discourse
Social psychology ami .wcial catc'gories 117
J[ ~~ Prublems wirh rrcu/itimwl categoriz.nrioll researcJ¡ 120
~¡ "\
Prolotype.r cmd variability 122
Morives and membership categories 126
Categories, comem and community 132 About the authors
Categories and the constmction of discourse 136

7. From representations to repertoircs 138 Jonathan Pottc.r and Margaret Wetherell are both Lecturers in Psychology
71w theory of social representations 139 at the Umvers¡ty of St Andrews, Fife, Scotland KYI6 9JU. They are the
Problems with social representations 142 authors (wJth Peter Stnnger) of Social Texts and Contexr: Literarure and
Imerpretative repertoires 146 Social Psycho/ogy ( 1984) and many scholarly articles.
Language, representation and rept!l1oire 155

8. How to analyse discourse 158


Phi/osophy, sociology ond mer/wdo/ogy 158
Ten stages in the wwlysis of discourse 160
Con clusi rms 175

9. Controversia\ tapies and future dircctions 177


Discrmr.H' ami !he world 'wuler the skul/' 177
Discourse and rhe \\lorld 'out there' 180
Di!i'C:Ollrse and reflexivity 182
77Je de1•elnpment of discourse ana/ysis 184

Appcndix: transcription notation 188

Rcfercnccs 190

Jndcx 205
Acknowledgements Introduction

As with any work of this kind many peop!e have conlribuled direclly This book is ubout languuge and its importance for social psycho!ogy.
or indirectly to the final product. We would particularly like to thank It looks at the sublle ways in which language orders our perceptions and
those people who have read and commented on versions of. chapters: makes things happen and thus shows how !anguage can be used to construcl
Dominic Abrams, Charles Abraham, Mick Billig, Paul Drew, N1ck Emler, and create social interaction and diverse social worlds.
Nigel Gilbert, Quentin Halliday, Dennis Hilton, Ruth MeFadyen: Andy We wil! be concerned with social texts of al! kinds; that is, lhe conversa-
McKinlay, Mike Mulkay, David Myers, Gün Semm, Heather Sm1th and tions, ncwspaper stories, novels and soap operas which are a central and
inescapable parl of everyday life. Indeed, the term 'texl' will be used in
Rod Watson.
\Ve have also benefited from an Economic and Social Science Research a broad sensc, to include not only writing prima facie but also the written
Council Grant 1 which enabled us to conduct sorne of the interviews w~ich record of the spoken; the transcript of an interview will be as much our
are discussed in the book. Bridget Rothwell retyped chapters and Dmne concern as the prepared narrative. .
Millar lhe references. Farrell Burnett at Sage has made many helpful We had lhree main goals in wriling lhis book. Firsl, to illuslrale the
suggestions about the content and organization of the work. breadth of understanding which has emerged from research on sociallexls
We are gratefu\ to Academic Press for permission to reproduce a tab\e in the last few years. We deliberalely chose lo discuss examples of work
from G. Scmin andA. Manstead (1982) Tiw Accowuability oJCmuluct, which asks diffcrent sorts of qucstions of highly varied types of discourse.
London: Acadcmic Press; to Pau\ Drcw for pcrmb;sion tn rcprmllH:c Sl!cond, wc hopcd thc bonk could sl!rvc as a rl!sourcc for thosc intcrcstcd
data extracts from J. M. Atkinson and P. Drcw (1979) Order in Court: in doing thcir own rescarch on social tcxts. To this cnd, wc havc tricd
Tite Orgaflization of Verbal Jweraction in Judicial Sertings, Lo~d~m: to providc a full description of certain studies rather thnn a complete! y
Macmillan/SSRC; and to Cambridge Univcrsity Prcss for pcnmss10n comprchcnsivc, but more supcrllciwl, covcragc of thc cntirc wrca. In
lo reproduce data extracts from G. N. Gilberl and M. Mulkay ( 1984) addition, Chapter Eight is dcvotcd cntirely to thc analytic issucs which
Opening Pandora 's Box: A Socio/ogical Analysis of Scientists 'Discourse, arise in this kind of work. Third, we hoped to demonstrate how .liome
Cambridge: Cambridge University Preos. ofthc most fundamental theorctical notions in traditional and more radical
social psychology can be illuminated by an analysis of discoursc. In fact,
we will suggcst that thefáilure to accommodate to discourse damages their
theoretical and empirical udequacy. Overull, we hope thal this work might
cncouragc othcrs to recognizc thc importancc and interest or analysing
social tcxts.
Thc book is organized as fo!lows. In the firsl chapter, thc lheoretical
roots or discoursc analysis in linguistic philosophy' ethnomethodology and
semiotics· are described. In Chaptcr Two we overvicw the pcrspcctive of
discoursc analysis, and illustrate its utility with respect to thc social
psychological notion of attitudcs. Then fo\Jow !lvc substantive chaptcrs
which deal in turn with the theorctical concepts of rules, accounts, thc
se! f. calegories and social representations. Chapler Eight then describes
the practica! stages through which rescarch on social texts progresses, and
discusscs the issue of validity. Final! y, Chapter Nine tackles so me broader
issue.s and identities futurc rcsearch directions.
Befare we begin, however, thcre are two things we want to do. We
would likc to discuss thrcc simple examplcs ofdiscoursc to illustratc some
of thc phcnomena with which wc will be conccrncJ throughout thc bao k.
2 Discourse CIIJ(l social psychology !ntroducrion 3

Thifi should givc a preliminary sketch ofwhat we mean when we say social thatthcre wns general and deep dissatisfnction in social psychology in the
texts play a constructive role in our sociullives. Then we will give a br.ief early 1970s to be checked? It is most unclear what a 'deep' dissatisfaction
account of our rationale for the research included in the book, and descnbe is. Perhaps the dissatisfaction musr ha ve been deep beca use it led to revolu-
sorne of the potentially confusing terminological issues which bedevil this tion- but this is just circular. The idea of revolution itself fits in well of
course, because it follows on from the notion of deep dissatisfaction.'
arca.
The point is, then, that even in an apparently simple passage the idea
of plain reference to pre-existing entities does not do justice to the.
complexity of the material. From the many possible Ways of character- . .,)".
Three illustrative examples
izin~ the relevant group ofpeople, sets of befiefs, añdPfOCeSses this ¡;
parttcular·-one··was chosen:-During the course of díTS-bOOK·--\ve·-wnl 1'.·
repeatedly ¡JOiril'óurthe-ever-present possibility of alternative descriptions-1."',
Erample one and categorizations. -- ·---
The first example is a reflexivc one in the sense that it is an instance of 1b A description of this kind in vol ves a whole set of choices, and these
thc kind of text social scientists themselves might write. This passage of ~-.bhmces may ha ve many consequences. Take the use of the term 'positiv-
discÜursc could, in facl, have stood pcrfct:tly sensibly as thc first paragraph istic'. There is a lot of debate about what precisely should be counted as
positivist research - is it confined to the idea that science progresses
of this book.
through the accumulation of facts, or does it include all research which
In thc lust !iftccn ycurs a rcvolution has takcn plucc in social psychology. This ~nvolvcs thc conversion of data into numbers? At thc present moment there
revolution, somctimes known as thc 'Crisis in Social Psychology' (Eims, 1975), IS much more agrecment that being 'positivistic' is a bad thing. It is a
was thc consequcnce of decp dissutisfaction wilh the stnte of rescarch and lheory. tenil which comes ready packaged with its evaluation. Another theme
Works likc Hnrré und Secord (1972), Israel und Tujfel (!972), Gcrgen, (1973) which will regularly recur in this book is the clase interdependence of
ami McGuirc (1973) uskcli fundamental qucstiom; about thc nnturc of thc descriptive and evaluative language.
discipline and, in particular, about its strongly positivis~ic re~iancc on cxpcri- \V_e can also look at how the passage is organized as an effective
ments as thc main research method. In thc aftcnnath ofthts radtcal reassessment
introduction. It offers a narrative of the development of the discipline so
there is now a necd to develop systematic methods of analysis which ... (from
that our present project (writing a book on the analysis of discourse)
an carlier draft of this book)
becomes both a proper and important thing to do. In this sense it serves
When reading research papers, reviews and books we are constantly faced as an initial justification of our activity. Again, we will see that justification
with writing ar this kind which we recognize as providing a hel~ful of various kinds is a very general feature of accounts given in a wide range
introductory orientation or context for the reader. But we could al so V!CW of situations, and one which, on the whole, psychologists ha ve been rather
this piece from a very differcnt perspective, focusi~g on it~ nature as a slow to pick up on.
social texr, and analysing how_ it constructs a ·specific_ reali_~~---- These phenomena are not, of course, unique to scientific texts. Far from
Thc lirst thing to note is that this passagc is itsclf an cnormously it. Texts are not part of sorne natural process like a chemical reaction or
ambitious piece of social psychology, identifying social categories (social electrons moving around a circuit. They are complex cultural and psycho-
psychologist), statcs of belief shared by the membcrs of the category logical products, constructed in particular \vays to make things happen.
(dccp c.Jissatisfaction), causal proccsscs (dissatisfaction leadmg to change) In thc course of this book wc will arguc that any social text can become~­
and cvents (rcvolution). This ambitiousncss is hiddcn, partly by the the object of rcscarch. E ven our own writing should not be immune from
apparent simplicity of thc description anci partl.y by its co~ventional this kind of examination.
ami familiar nnturc. But therc is n great deal gomg on evcn m a short
tcxt such as thi.s.
Thc cutcgory '.sm:ial psychology' is far from straightforward.:-It could Erample rwo
rcfcr w a sct of ideas, u body of writings in journul.s, textbooks and so
on. orto a group of pcoplc. Rcfcrring to social psychology is not qu!tc The following passage comes from a conversation where Ruth is discussing
Jikc referring to doughnuts in thc bun shop, as in '1'11 ha ve t~at one w~th an acquaintance who is supposedly mentally ill (transcription symbols are
thc extra jam'. Thc statc of bclicf is cqually difficult. Hmv ts thc clmm cxplaincd in the Appendix).
4 Discourse cmd social psycho/ogy /ntrocluclion 5

Rttth: He was a rncing driver and [)he had one panicular spccial cur which It is quite possiblc to crect a version which docs not dcpend on the
he'd ncvcr Jet anybody elsc go in. But he used to takc my mum out for drives idea that the man has departcd from thc world of the sane. In doing

--
in il, and, you know, cvcrybody fch that that was alright and cvcrything. this wc are not, of course, claiming we can tcll what 'really happcncLl',
Looking back on it they dccided, my grandmOlhcr decided, that shc should but by displaying the possibility of alternatives wc can start to look al how
ha ve realized that there was something odd about it. Becausc []he wouldn't ·-····-··
let anybody clsc ncar thut car. And my mothcr wus very vcry naivc about different readingsare produced. Throughout the book wc will encounter
evcrything m that stagc. [] Anú he suddcnly said ta her in thc kitchcn une day: vá"riabilit}~·orLfiTSllrl~l b~i~~;ecn vcrRions, and in Chupter Five wc willlook
'l think we'l\ havc to stop meeting like this'. And she hnd no idea what he specifical\y at how selvcs or typcs of people come to be constructed in
was talking aboUI. lt just complctcly bafflcd her. And thcn he slnrtcd writing discourse.
really strangc, long, long. rambling leltcrs - very abusivc lctters - to my
grandfamily, wha realizcd [J that something must be up. (Mi trnns 4:1)
Etample three
Justas with Example One, there ís nothing peculiar or nonsensical about
this section of discourse. As readers we do not have any difficulty in We will take one more bricf example, from another arca altogether. The
reading it as a literal description of someone who is starting lo 'go mad'. following piece is from the bcginning of a newspaper articlc.
Yet there is a lot more to notice about it. One of the interesting things
about this account is that it is an account of an account. Accounts of lslamic Tcrrorists Blow Up Planc
accounts are a chronic feature of ordinary life: we read a story in a
In Beirut last night thc hijackcrs of thc British Airways 727 fina\ly rclcascd
newspaper and then describe it to friends, or we gossip lo Kevin about
the passengcrs and crew. As the gunmcn lcft they detonuted a !urge quantity
what Caro! told us about Jane and Steve. Accounts are often versions of of explosive and the planc was quickly gutted by firc. This followcd a pcriod
versions. The present one is Ruth 's version of events; but it is nota first- of in tense negotiution in which the authoritics madc it clcar that tbcy werc not
hand version, it is based on her grandmother's version of what went on. going to meet the hijackcrs' main demands.
And it is clear that evcn the grandmother's version is not first hand; her
account of what occurred in the kitchen is prcsumably bascd on the Perhaps one of thc most striking things about an uccount ni' this kind is
mother's statement. its familiarity. It conforms to a standard pattern and we could probably
Another interesting feature ofthis piece afta! k is its status as a descrip- give a reasonably detailed speculation about both the cvents leading up
tion. In fact it is part of a much longer response to thc question 'have to those described and how this story will continuc. Therc is a standard,
you ever known anyone who you thought might be mentally ill?' Three virtually stereotypical, narrative which accompanies the cvcnts wc categor-
features of the man 's behaviour are singled out for comment: taking the ize as 'hijackings '. As competent newspaper readers we can generate this
mother for drives, thc episode in the kitchen and the letters. Part of what scenariojust from the tcrm 'hijack'; the specitic story that follows bccomcs
makes thcse features cvidcnce of mental illncss is their inappropriatc or a variant on a general thcme rather than a uniqucly ncw cvcnt.
unusual nature. The listener is not provided with a rational story to integrate As in the first examplc, some of the terms in this story come ready
them together and so they function effectiVely in an anecdote about mental cvaluated. This is true of'thc gunmen', 'hijackers' and, most espccia\ly,
illness. However, thcre are many ways toread texts and it is possiblc to 'tcrrorists'. A\1 are constuntly used with ncgativc associations in tcx.ts such
look for other plots or patterns which might fit. as the above. A cursory look at the use of these terms might lead us to
This account could be seen as a classic story of a lave affair where the expect that they ha ve a precise reference - small groups of politically
parents disapprove. There is nothing odd, Jet alone mad, about taking your motivatcd peoplc who blow things up and create terror- and that thc
\oved one out for a lot of drives. Neither is the statement in the kitchen negative connotations are generated by rcpcated association of the term
- '1 think we'll have to stop meeting like this' - unexpected in this kind with horrific events. However, it is important to be aware that the evalua-
of narrative. Using a clichéd jokey formulation is one way of broaching tion comes into the system at a much earlier point.
delicate questions about a relationship. Pcople are often in situations where lt is well kna\v·n that whcn such people are seen as acting in a good
they raise tapies obliquely to test the water befare giving their whole- cause they are described not as 'terrorists' but as 'frcedom fighters'. Often
hearted linguistic commitment to something. And in this version the abusive this goes· along·with evaluating their acts as 'necessary for encouraging
letters make sense, because later in the account we find the letters only social change' rather than 'causing pointlcss carnage' and evaluating their
started after thc grandmother had forbidden her daughter to see the man. motivations as arising from 'moral and politicul ideals' rather than 'warped -e: ~\t- ¡Of!rl)f
. 1
1
:¡V!'-7'0\1 i¡oJ~IJ-t::
S DiscoHrse and social psychology

sociology and literary theory. We will start. the boa k with a review of
cenain developments in thcse disciplines which ha ve la1d the loundatlons
for our current understanding of the operation of social texts. l
Foundations of discourse analysis
·( ¡ C;. ~.(~.-

There are a number of very good rea.sons why psychologists should be


intcrestcd J.n language. Languagc is so central to all social activities it is
casy to tukc for granted. Its very familiarity somctimes makes it transparent
to us. Yet imagine conveying a complex idea such as 'meet me Thursday
in my room for a discussion of semiology' without language. It is not easy
to see hmv it could be done.
Communication of this kind which involves abstract notions, actions
and cvents removed in time and space, dclicate shades of meaning, and
logical distinctions dcpcnds on peoplc sharing a complcx symbolic
represcntational system. Moreover, language is not justa cacle for com-
munication. It is inseparably involved with processes of thinking and
rcasoning. Justas it is difficult to imagine sophisticated cornmunication
without languagc, it is hun.l to scc how complcx ubstrac.:t rcasoning could
be performed by peoplc without a language.
The study of language is particularly vital to social psychology because
it simply is the most basic and pcrvasivc form of interaction bctwccn
peoplc. Wc spend a lot of our social Ji ves tnlking to cach other, reading
the papers, watching rnovies and writing shopping lists. Most forrns of
soci~l interaction involve pcoplc talking together or reading each other's
scribblings. Furthcrmorc, whcn raising issucs of thc naturc of culture,
on the one hand, and thc nature al' lhe self, on the other, it is virtually
impossible to clearly disentangle thcm from questions about language and
its role in human affairs. As we will se e shortly, it is clcar that a large
part of our activities are perforrncd through language; O)Jr tal k and writing
do not live in some purely conceptw.ll real m, but ar~ mediums for action.
Des pite the imponance of language, it has not always been a tapie which
has intcrested psychologists. lndeed, it only really too k off as a majar field
of study in the late 1950s and !960s under thc influence of the linguist
Noam Chomsky. The best way to understand thc developments discussed in
this book is to contrast them initially with the very diJTcrcnt Chomskinn
psycholinguistic tradition which predominated in psychology during the
1960s and 1970s.

)
Chomsky and psychology

Nnam Chomsky (imrn 192H) dcvclorcd his linguistic ideas at thc Univcrsity
of Pennsylvania and latlcrly at thc Massachus~tts lnstitutc of Tcchnology.
6 Discourse and social psychology lntrodHction 7

pcrsonalitic~·. Thc contrasting accounts ofthc African National Congress developing, Iinguistic practices (Foucault, 1972). Sorne workers make a
givcn by Wcstcrn liberal~ and thc South African govcrnment are prime contrast bctwccn discoursc analysis and convcrsation analysis on the basis
examples of such a split. Here the political conflict is exactly paralleled of different theoretical and methodological strategies (Levinson, 1983),
by linguistic contlict. It is important to immediately dispcl any illusions while others want to make a very different contrast between discourse
that thc study of languagc is lhc study of" a cal m, idealizcd real m bcyond analysis and text analysis with thc aim of scparating the study of underlying
politics, argument and conOict. theoretical structures from actual Iinguistic performance (Halliday, 1978).
In discussing thcse three examples wc have tricd to provide some All in all, there is a great deal of pütential fur confusion.
initia\ simple illustratinns of thc sort of phcnomena which are thc subject It is importan!, then, to clarify what we mean when we use these terms.
tal' this book. Wc havetrie_d to_ show hQw .. so~ial texts d_o not mcrcly \V_r;_ -~UL. _us~. .'-~i-~co~~se_'_. .in.. its mo~t pp¡::n _sense, following q_i_lbert and J
rejlecr or ,¡¡,.,.m~bje~-t~.- ~~~ñtS. a~ct-~lltegories pr~-e-xistirÍg. in thc social Mulkay (1984) to cover all forms of spoken interaction, foriririlañd
'nnd-rúitüfal world. Rather, thcy activcly construcr a version Ofthose things. informal, and written texts of aJJ kinds. :So when we talk of 'discourse j
Thcydonotjusldescribe things; thcy do things. And bcin_!L!l<:tivc_,t~ey analysis' wc mean analysis of any ofthe$e_ forms of discourse. We will
bllve social.and_ppJitÍ_f~_UrnP.l.ications. We ha ve secn how description is giVC a· fulle'i-accourir bfthc·-nature·at discour~~--~-~a-lys-Í~~s we understand
iiect to cvaluation and how different versions of events can be constructed it, in Chapter Two. For the moment, Jt is important to emphasize that
to justify or blame thesc events. As wc will scc, this approach to lunguagc 9.~I_C:QD~_Gtn.is. not purely _with_discourse.persc; that is, wc are not linguists
has profound implications for lhe practice of social psychology which, attempting to add social awareness to linguistics through the addition of
in many cases, presupposes that Ianguage is unproblematically and simply the study of pragmatics. We are social psy~hol 0 gists expecting to gain
dcscriptive. a be~tcrun~-~-~Handing.of.sociall!f~ _ ~nd social interaction from our study
of sodaf"icxts. · ~ · · ... · -· ·
By focusing the book in this way we have excluded three main areas
Definitions of discoursc analysis of work. First, we ha ve avoided research concerned with the interface
bctwccn discourse ancl cognition. This includes work on the comprehcnsion
Perhaps the on\y thing all commcntiltors are agreed on in this arca is that of texts, their mental organization when stored in memory, and the role
tcrminological confusions abound. The problem arises becausc of schemata and scripts for discourse processing (e.g. Bower and Cirilo,
dcvclopments ha ve bccn happcning concurrcnlly in a numbcr of" diffcrcnt 1985; Frederickson, 1986; van Dijk and Kintch, 1983). Second, we havc
disciplines (psychology, sociology, Iinguistics, anthropology, Iiterary avoidcd research which is mainly concerned with what we see as primarily
studies, philosophy, media and communication studies), using a panoply linguistic qucstions; for. example concerning the way reference is made
of theoretical perspcctives. For example, the Jabel 'discourse analysis' from onc sentencc to another (e.g. many of the vol u mes in the series
has becn used as a generic tcrm for virtually al! rescarch concerncd with Advances i11 Discourse Processes). Finally, we ha ve not cave red socio-
Ianguagc in its social and cognitivc contcxt (Brown and Yule, 1983; linguistic work concerned with the study of variations in the language use
Coullhard, 1977; van Dijk, 1985), as a description for studies focusing of different social groups (e.g. Labov, 1972a; Milroy, 1980) and the
only on Iinguistic units abovc thc Icvcl of thc sentcncc (Stubbs, 1983), psychological intergroup dynamics associatcd with thesc (Giles, 1977;
as the corrcct tcrm for research concemecl with cohesion ancl connectedness Giles and S t. Clair, 1979).
across sentences or turns of ta\k (Tannen, 1984; van Dijk and Kintch, Even with these exclusions, there is still a huge amount of relevant
\983), and to cover developments stemming from structuralism and rescarch left (for thc most comprehensivc collection, see van Dijk, 1985).
semiolics (Foucault, 1971; Pcchcux, 1982). Jt is a field in which it is From this work, we ha ve selected our tapies according to what fitted best
perfectly possible to have two books on discourse analysis with no overlap into our scheme of prcsentation in terms of the social psychological
in contenl at all (el'. MacDonell, 1986; Stubbs, 1983). concepts we wish to covcr. Naturally we have also been guided by our
In line with these diffcrences, thc term 'discourse' itself has becn uscd value judgments asto what is the most productive and interesting. Othcrs
1 in many varying ways. Sorne researcher~e~l.~~~e~~?__ ~7~~-all form~ will loo k at the field differenlly, but we hope that no onc will be offended
of talk and writing (Gilbert and Mulkay, 1984), others take the term to by either exclusions or selections.
:·app!y-órily ta··¡¡;e \Vay talk is meshed together (Sinclair and Coulthard, Discourse analysis is a relatively ncw approach in social psychology.
\ 1975). While ut the othcr extreme, sorne continental discoursc analysts It should not, however, be thought that the approach fcll out of the air.
such as Foucault takc 'discoursc' to refer to much broacler, histórically It has its roots in a variety of more establishcd perspectives in philosophy,
JO Discourse and social psychology Fowulations of discourse anaiysis 11

His basic conccrn was to produce a gcncrative grammar, that is, a limited . A t~eory ~hose ?ata ~re tran?formed in this way is in danger of slipping
set of rules which generales grammatical sentcnccs and only grammatical . mto.cJrc~lanty. I~ Js a~ 1f one d1d research on the visual system using only
senlences (cf.Chomsky, 1965). vertJcallmes as sttmuh, and concluded that the visual system was espccially
Chomsky's work was crucially important for psychology becausc he adaptcd for the perception of vertical lines. Only certain data are admit~
vicwed thcse rules not mere! y as a more cconomic form of dcscription, ted as 'propcr' performance data, and from this base an ordered set of
like thc Library of Congress system of cata1oguing books say, but as generative rules is produced.lt is not at aJI clear, however, ifthese rules
psychologically real cognitive structures. He took the rules as representa- c.ould. cope with natural performance data, with ordinary talk in ordinary
tions af what speakers must actually know, that is, as representations of SituatlOns. The problem is not idealization per se; after all, Newton's Jaws
thcir cognitive systcrns. Indeed, Chomsky has suggested that these rules ?f mechani~s ar~ ex.act only for ideal frictionless environments. The worry
may actually be a part of every pcrson's genetic make-up, inherited at IS that thc Ideahzations Chomsk-y used miss essential features of natural
birlh (ChomRk-y, 1966). speech and informal cvcryday conversations (for an excellent critique of
The details of this nt times highly technical thcory are not of interest thesc kinds of idealizatons in linguistics generally see Harris, 1980).
he re. Our cancern is more with the general nature of Chomsky's linguistic There is.' however, an interesting response to this point built into the
programmc. \Ve can best show the contrast between his work and discourse Chomskmn approach, which is lo stress the importance of speakcrs'
analysis by concentr<.~ting on three tapies: thc compctence/pcrforrnance intuitions.
distinction, the role of speakcrs' intuitions, and the importance of speakers'
crcativity.
Speakers' imuitions

The comperencelpelformance distinction Chomsky indentifies the linguistic perforrnances which are suitable for
analysis on the basis of speakers' imuitions about which sentences are well-
Chomsky argueú that it i.s pos.siblc to di.stingui¡;h betwcen, llrst, thc formed and which are not. What this means is that he does not try to find
underlying ability to produce grammatical sentences and, second, the actual out what utterances conversationalists themselves treat as well-formed or
productian o·r particular sentenccs in particular situations. More formally, badly constructed, rather he uses his own intuitions about what is and what
he úistinguishcd bctwcen thc sct of rules which allows thc Rpeaker to is nota 'prnpcr' scntcncc. If his general perspectivc is correct, this is a
genera te grammatical sentences and the performance itself, which is subject s~nsible and appropriatc procedure because Chomsky holds that al/ of us,
ta a panoply of speech errors, lapses of memory, distractions and effects hJmself mcluded, know what is grammatical and what is not. If we all
duc to context. Quite apart from the question of whether this theory is hav~ a Iinguistic compctence, a set of rules for the production of gram- ,
empirically carrect (and it has not fared well on thc whole- for a rcvicw mat1c~l scntcnccs ~hich also enables us to tell when other people are
of experimental work on the theory sec Johnson-Laird, 1974) there are spea~mg grarnmatJcally or nan-grammatically, then it seems perfectly
conceptual difflculties with this kind of approach. sens1ble to use these intuitions to decide on appropriate speech data.
The most' fundamental of these centres on thc problem of thinking of Thc problem with this assumptian is that it risks compaunding the
competencc indcpendcntly of performance. How does the researcher get existing circularity with another. This process of relying an intuitions is
dircctly at the underlying rules if nDl by way of performance data; via acceptab1e only if Chomsk-y's lheory is correct. And when idealization
actual language of various sorts? In fact there is no indcpendent access. is combined with reliance an intuitions it becomes hard ta see how the
But Chomsky skirts round this problem by working from idealized theary could ever be falsined. Data which might patentially ovenhrow
sentences and using these to make infercnces about the nature of com- the thcory are likely to be excluded tout court.
petcnce. Thc linguist Jolm Lyons ( 1967) notes that there are threc senses Agai~st this background, the idea of looking at naturalistic speecll data
in which Chamsky's specch data are idealizcd cornpared with ordinary (t.ranscnpts of cveryday conversations or documents of various kinds) as
:-;peech. First, the d<.~ta is regularized so that the pcrvasive erran;, hesita~ d1scoursc analysts suggest, is a radical one. With its ernphasis on com-
tions, sclf~corrections and so on are removed. Second, it is stalldardiz.ed. petcncc and spcakcrs' intuitions, Chomskian psycholinguistics does not
That is. it takcs no account of variations in pronunciation, say. Third, acccpt this as a necessary stcp. But there is more to it than that; for the
it is decol1le,ttlfa!ized, separating thc semences out from any specilic context ~homskian perspective maintains ordinary langu<.~ge performance data will
in which they might be used. m any case be unmanageablc becnuse of the creatil'iry of speakers.
Foumlwionr r~l tli,\·course analysis 13
12 Discourse and social psyclwiogy
N: ~nywa::y,=
Speake;s' creativity
H: =pk! A:nywu [:y •...
Qne of the central tenets of Chomskian psychol inguistics is that speakcrs N: So.:.:,
are capable of producing an infinite number of sentences simpl~ through (.)
the use of the various generativc and transformationa\ rules wh1ch make H: .p=
up a grammar. Chomsky idcntifles crcativity as onc of the most csscntial N: =Ynu'll come <Jhou:l (.) eight. Right?=
features of languagc use. The importance of this point should no.t ?e H: =Ycu::h,=
N: =Okuy
underestimated. At least in part, for Chomsky. it is the 'fact' ofcreat¡vlty
(.2)
which makes children's learning of a languagc so amazing, and which
N: Anylhing ebc to rcE_ort,
makes the behaviourist explanation of this \earning, in terms of rewards,
shaping and reinforcement schedules so unconvincing. If there is such H: (.3)
a huge diversity of sentences in a child's environment, .many of th~m N: Uh:::::: m:::,
novel, thcn how can thc child possibly start to dca\ wlth thcm usmg (.")
the normal principies of associative \carning? As is wcll known, Chom~k?'
H: Gcuing my ~uir cut tihmorrow. =
be\icves that only the prcsence of a gcnetically wired in '\anguage acqu~sl­
N· =Oh ~il!¿'!
tion device' can make the surprising phenomcnon of language \earnmg
(Buuon and Cuscy. 1984: 168)
possible.
The postulation of this radical creativity also has import~nt methodo-
logical consequences. If we accept that speakers are so cr~auve that many Gcnuinc performance data of this kind seem very far removed from the
or perhaps the majority of their utteranccs wi\1 be umquc then therc idealizcd scntcnccs used by gcncrativc grammariuns. Evcn a succcssful
would indeed be little point in looking at performance data, at open-ended gencrative grammar has virtual! y nothing to say about many of the fea tu res
ordinary talk. The task of systematizing such an infinite body would that are immediately relevant to thosc who anal y se this kind of transcribed
be daunting to the point al' impossibility. The researcher could o~ly tal k. For examplc, ultcranccs are. regular! y ungrnmmatica.l_ in cveryday
achieve arder by focusing on the speakers' competence or set of gencrat1ve tal k withOu~eliCitiilg-C5niñli!nt~ they ·cohcre intó seqúeñtial d.iscO·urse, and
rules. Unlike performance data, these are manageable in complexity they are coinriióruy· the joint achievement of two or more people. All these
and number, and anyway are taken as providing an explanation of al\ features are apparent in thc cxtract abovc.
Furthermore, thc creativity evident in the naturallanguage data examined
possible performance data.
~p to now fallsshort ofthe unmanagcable and unrestricted plenitude implied
m the Chomskian tradltlon. Indeed researchers ha ve found that much natural
Namral language la_n~~-~~e._u.s.eJ~ h_iglllY stereotyped and.. quitc predictabíe.-Far Koñí-be.ing
ImpOSSibly UlllqUe, pe~forl11".._11_CC dat~ is often boringly-repetitive. 11.ttV"Ic,:.,. ¿ b,.,,.c
The Chomskian approach is undoubtedly a very elegant one and its different The fact that performance data can be systematiciilly handled pulls out :¡
facets fit together with a coherence unusual in a lot of psychology. lt one prop from under the abstract psycholinguistic approach. A study of
focuses specifical\y on the gram_~atical features of discourse and given transcript also questions thc central role playcd by speakers' intuitions.
this ·goal we wmild not wish tó í:ritiCize it for fai1ing tO provide a complete For verbatim transcripts are very different both from our intuitions about
explanation of language. Nevertheless, advances in language .s~udies ha~e speech and from our la ter rccollcctiom; of cnnversations. Part ofthe rcason
depended on researchers questioning sorne of the presupposltlons of th1s for this is probably thc difficulty of cscaping from our literary conventions
perspective. One of the most important stimuli for developing an altemative ~or representin~ speech in written form. Wc tend to think we speak rather
has come from natural speech data of various kinds. h~e characters m a play, but actual vcrbatim spccch is not at all likc play
If a verbatiiñ-tiññScrfpt -arrear
peoplc having an ordinary conversation dmloguc. And, as Sacks (1984) anO other convcrsation analysts havc
is examined the Chomskian approach to the three issues above looks shown, the detail that is edited out and lost in reconstruction and intuition
decidedly fr~gile. Take the following piece of transcript (slightly simpli- is extreme! y rich and often vital for a full understanding of what is going
fied from the original - the transcription symbols are cxplamcd m the on. This is not surprising, nf coursc. bccausc as !ay talkcrs wc almost
ncver ha ve to formulate how or what we do when we talk: we just tal k.
Appendix).
14 Discourse ami social ps_vchology Folll!dations of discourse analysis 15

'¡ In contrast to thc Chomskian lradilion, thcn, discoun;c analysis has whose central f-unction is the description of states of affairs. Whm Austin
cmbraccd performance data in all its mcssy ami ungrammatical complcxity. se.t.o~~. l() do :vas to undcrmine thc notion that an understanding of 'truth
1It has not been cowed by a fear of impossible crcativity, and it has let condltlons', states of truth and falsity, is central to an understanding of
spcakcrs' intuitions takc a back seat. As a rcsult, it has had to face up language, and in doing this he produced a madel with far-reaching
to the ract thnt Janguagc is notan ubstracl rcalm. Out thcrc in thc real consequences.
world it is ma:d~p-from par:ticular utterances performc~ in part~cul.ar
contexts. MorcovCr, discourse becomes placed ftrmly in the arena of socml
bchaviour. Wc bccomc lcss intcrcstcd in what is going on undcr pcoplc's Staring \'er.ms doing
'' skulls anda lot more intcrcsted in how peoplc are actual! y using languagc
with each other in the course of diffcren~ kinds of interactions. Our research Austin began with the observation that therc is a class of sentences which
(
questions thus· open out into a full-blown social psychÜIÓgy of discourse. are principally important for what thcy do, not because thcy describe
\?olvt tv~·t T
1
(
For a better· understanding of the development of ideas about \anguage things. For instance, tl~.~.-~.e;nte.nce:
function wc now nccd to turn to events taking place in another discipline
entirely, namely the philosophy of \anguage. We need to look at John I declare war on thc Philippines
Austin's (1962) notion of speech acts.
is not a description of the world which can be seen as true or false but
?-~-a~t wi~ practi.cal.consequences; when uttered in the right circumstances
Words as dccds: spccch act thcory 1t bnngs mto bemg a statc of war. Austin cal!ed scntences of this kind
performarives. Other examples are:
In 1955 at Harvard the psychologists wcre buzzing with excitcment about
thc lcctures bcing givcn by Noam Chomsky, although perhaps this had I namc this ship tbc Lady Penelopc
more todo with his outspoken criticism of B. F. Skinner's bchaviourism Bcwarc of thc bull
than specifically his theory of Tro.nsformational Generative Gra~mar. 1 sentcncc you to six months bnrd lnbour
Languagc in \955 wus intcrcsting becausc, us wc have noted, _when cluldren
lcarn languagc thcy sccm tu del)' thc bch<Iviourist, simple ."itimulus-rc."ipnn~c In cach or thesc cases, thc primary role of thc scntcncc is not dcscription a."i
view of the world. In that sume year the British philosophcr John Ausun such b~t to make ccrtain thin~s happen; thcy are sentenccs pcrfonning acts.
( 1911-60) was al so at Harvard, prcsenting thc prestigious Willium James Au~lln contrasted thcsc kinds of sentenccs with others whose primary
!ccturcs. A fcw p.sychologi..;;L"i drifted into thcsc seminars. On thc whole role .d1d appcar to be the description of states of affairs. That is, scntences
thcy were nonplusscd by thc urbanc, failllly ironic manncr of Austin's makmg slatcments and asscrtions which can be chcckcd as cithcr true or
prescntation, as well as being mystificd by his apparently self-destructive false. ~ustin.' who had a p.enchant for coining ncologisms, named them
procedure of first arguing Jor a distinction between two basic forms of consrattves (m the hope al' avoiding the ambiguity of 'statement' which
utterancc, arid then, later, arguing against it. Al\ the same, the general could mean cither a description or any spokcn utterance).
perspective on laY'tguage Austin proposed ultin:atcly had an enorm~us One of the intercsting features of performative utterances as we ha ve
impact on a wide rm,ge of disciplines and has rad1cally rcshaped our VICW noted, is that unlike constatives, they cannot be straightforw~rdly true ar
of language and its operation. fa! se. But they can be defective in other ways. Think of the utterance used
It is important-ror a clear understanding of Austin's ideas to ~1ave so.me by a monarch when knighting.
indication of thc situation in philosophy which he was both reactmg agmnst I dub thee Sir Lance\ot
and commenting upon. His main target was the \ogical positivist vicw that
sentenccs which cannot be verified, that is sentences for which t~(:!~e is There are various ways in which this utterance will fail to institute thc
no way of checking whether they are true or fa\se, should simp\y be treated act of knighting, or w.ill make the act problematic in sorne way. For
as mcaninglcss. From this viewpoint, for instance, the statement 'God example, the person m1ght ha ve been misidentified: perhaps Mer\in has
does not exist' should be trcated as nonsensical sincc the truth ofthe state- m~stnkenly been eaught tieing his shoe laces. Or perhaps the appropriate
ment can never be validatcd. In addition, Austin's argument was directed W1tnesses, the sword and so on are not present. It could even be that the
at a wide swathe or views of \anguage which takc it to be an abstract system whole thing is a joke set up to amuse the.other Knights of the Round Table.
16 Discourse and social psychology Foundations of discourse wwlysis 17

In each case there would be a problem with the utterance but the problem distinctly problematic, oras Austin put it, unhappy about this utterance:
is not with its truth or falsity in any simple sense. The crucial point is That dog is called Lassie but 1 don't believe it
that certain conditions have to prevail for perfonnatives to be accomplished
successfully. These conditions are called feliciry conditions, and they are The problem here is not truth or falsity- the dog mayor may not be called
listed schematically in Figure 1.1 along with an example vio1ation of the Lassie and the speaker may or may not accept this fact; the problem is
conditions. the lack of the appropriate beliefs which go along with claims of this kind.
In terms of Figure 1.1 it vio1ates condition C(i) which states that people
must have the appropriate thoughts, intentions etc. to properly carry out
FIGURE !.1 speech acts.
Felicity conditions In the light of a large number of similar considerations Austin abandoned
(A.i) There must exist un accepted conventionnl procedure huving u certain conventionul the simple distinction between two classes of utterance. However, the
effect, and furthcr, exploration of this distincti:--n has raised sorne ve!'Y important issues
(A.ii) the particular pcrsons nnd circumstunccs in n givcn case must be uppropriate for thc demonstrating,, _firr-st, t~.~-t-~~!l.!.~!lE~i!~f.l.§_\Y~~~~~-ri~.~-thi~gS-~lnd,
invocntion of the particular proccdurc.. second, that there areiP.J!.vemLQltS_ocfelir;jty_con~iti9n~ \V.hfCh: Iiñk utter-
The procedure must be executcd by ull purticipants both (i) cnrrectly und (il)
(B) ill]!=.~"s.and_,s_o_t;:j~l_at;t_i_Vities. For example, in a ship-naming cere.rñOily the
complete! y.
Qften (i) the pcrsons must huve ccrtain thoughts, intenlions etc. which are spccilicd force of '1 name this ship ... 'dcpends on a set of conventions concerning
(C)
in the procedurc, (ii) thc procedure spccifies ccnnin conduct which must be ¡¡dhercd to. the correct sequence of events, the proper people who must be involved,
the presence of the ship and so on. Austin replaced the performativel
Violutions constative distinction with what he called tl,le general t_Jle~ry of spec;_ch acts.
A(i) Suying '1 Jind you guilty' to sornconc whcn s~cing thcm shuplift in n supcrmurkct. ... ~- . - ',

A(ii) Awurding un honornry degrcc to thc wrong pcrson.


B(i) Snying 'OK' instend of '1 do' ut u wedding.
B(ii) Not suying 'you'rc on' whcn uccepting ¡¡ wager.
77w general rheory of speech acts
C(i) A jurar voting 'guilty' when she thinks thc uccu~cd i~ innoccnt.
C(ii) Making u promisc und thcn not following it. The general theory does not_distinguish between sentences which do things
and sentences which say things, between performatives and constatives,
but casts this distinction in a different way. The fundameiltal tenet of the
Austin's work is unusual in that he deliberately set out to test exhaustively ;¡theory is that al/ utterances statethings and dolhings. That is, all utterances
his own notion that these two sorts of utterances can be distinguished. ¡:·~:t~a..m'eañiiig.--átid"ilotce.. In fact, Austin suggested that with any
In the course of his William James lectures, subsequently pub1ished as · lutterance a speaker is simultaneously doing three sorts of things.
How To Do Things With Words, he started with this notion of the two First, the speaker is uttering a sentence with a specific meaning - it
classes of utterance but went on to show that this is too unsophisticated has a certain sense and may refer to specific eveñts,--pérSot\Sor objects.
a way oflooking at things, that in fact al/ utterances have both performative Second, the sentence is uttered with a particular force. We know what
and constative features. Take, for example, the suggestion that constatives the words 'shut the door' literally mean~"!J~t- they·-~an be used with the
can be true or false and that this distinction is not relevant to performative force of an arder, a request or even a question. Force is thus an clement
utterances. If we examine utterances like the following we can see sorne of utterances which is dissociated from their meaning, although it is often
difticulties with this idea. indicated by the use of a certain verb: promise, arder, state and so on.
The third feature rcfers to the effects or COI]_S{!;!=Jl.Um.ccs or the first two.
1 bet you a pound that Ludwig wins the 3.15 at Kemptown The sentence 'shut the door' may be uttered with the force of an arder
This is an utterance performing the act of bctting; it is not describing the but it may ha ve the effect of making the hearer shut the .door or it may
bet, but doing the bct. Howevcr, it bccomcs cxtrcmcly problematic ifth~rc simply makc thc hcarer annoyed.
is no 3.15 race at Kemptown or if Ludwig is nota horse. Performat1ve This third fcature is much less determínate than thc·lirst two. For
utterances are thus not entirely independent of matters of truth and falsity. instance, if the appropriate penmn in the right circumstances uttcrs thc
On the other hand, dcscríptívc statcments- constutivcs - cannot only be sentcnce 'I declare war on thc Phillipines' it is not rcnlly clcur if thc
true or false but can, like performativcs, be infclic;_Ul9_!::i.s. There is somcthing
18 Discourse ond social psychology Foundations of discourse analysis 19

consequcnces are the initiation of the war, the en tire set of military events as a result of being a member of the jury but brought them into the court
which make up the war, or even those events and everything subsequently as part of an organized body of knowledge and skills used all the time
influenced by them. Austin has not provided a systematic way of deciding (Garfinkel, 1974).
what precisely counts as the consequences of the utterance. Ethnomethodology as a discipline is concerned with a wide range of
Austin's :theory represented a radical departure from much of the features of the way sociallife is put together. However, it is of particul8.r
previous philosophical work on language because instead of viewing interest here because of the lessons it pro vides about how language is used
it as an essentially abstract corpus, which can be dealt with in the same in everyday situations. The best way to get to grips with these is through
way as logic and mathematics, he recognized th~l~~.~':!ag~~ a human looking in detail at a classic ethnomethodological study by Lawrence
JUllCLice.. People use language, like a tool, to get things done, whether Wieder (1974a and, most accessibly, in 1974b) which looked at Jife in
these things are bets, relief from the draught through an open door, a half-way hostel for narcotics felons.
or thc start bf a war. This feature of the theory makes it very attractive to
social psychologists~ as there is little in the psycholinguistic tradition
which even begins to show how a researchcr might deal with language Telling the code
function.
Austin offers a highly social vicw of language. He draws our attention Wieder's study was based on an intensive period of participant observa-
to the role played by the web of social conventions in the achievement tion in the half-way hostel. The goal ofthe hostel programme was to help
of actions through talk and thus sensitizes the researcher to features of ex-convicts readjust gently to life in the outside world and lead them away
the social cOnrext surrounding \anguage use. Despite these positive aspects, from the world of drugs. In these terms it was a dismal failure. In fact
howcvcr, his thcory was primarily dcvclopcd to combat alterna.ti.vc only a third of the inmates left in the approvcd manner, the rest jumped
philosophical vicws and little consideration was paid to the practical!ttes parole or offended again and wcrc rearrested. The research involved
of actual\ y :applying it to everyday talk in natural situations. In fact, the Wieder spending time around the hostel, watching what went on, and most
rcscarch tradition which Austin spawncd has in many ways staycd as importantly building up a rapport with the inmutes and staff so he could
abstract and decontextualized as its Chomskian predecessor (cf. Searle spend long periods talking with them about their lives.
et al., 1979). In contrast, the ethnomethodological research we ~ill One of Wieder's objectives was to illustrate the difference between an
examine in.thc ncxt section has bccn guidcd much more closely by detmled ethnomethodological anda traditional social scientific approach to language
use by displaying the shoncomings ofthe latter. To this end he first adopted
empírica! studies.
\- i ! ,;--._ ( . ·~·'-'""- - .i_ (1 \.· \ ·, '·'..:
1 the traditional social scientific play of identifying a set of informal rules
r,.•) : .( ',-:,<:_\ ' ·-~-; operating in the institution. It is a commonplace finding in traditional
4 Doing' trilking: ethnomelhodology research that prisons, hospitals and similar total institutions ha ve a set of
informal rules which are different from, and often oppose, the official
As its namc suggests, the subdiscipline of ethnomethodology is concemcd ones (Becker, 1963; Goffman, 1961). In this kind of work the rules are
with the study of (ology) ordinary people's (ethno) methods; the methods seen as a template guiding behaviour, so if you could describe the rules
in questíon bcing those used for producing and maklng sense of cveryday you would have an explanation for the inmates' actions. Through his
sociallife .. Ethnomethodology treats the goals and aims of ordinary people participant observation in the hostel Wieder was able to identify just
as similar to the goals and aims of the social rescarcher. That is,. peo~lc, such a sct of rules- which he called 'The Code' (see Figure 1.2)- and
Iike the scientist, are constantly attempting to understand what ts gomg he goes on to show how these rules could be considered as having
oh in any situation and using these understandings to produce appr~priate motivational force.
behaviour.of their own. A central figure in the development of th1s per- In the first place, the narcotics felons had their own sanctions for those
spcctive was the sociologist Harold.Garfinkel (born.l_917- se.e Gar~nk~l, who departed from the Code. Inmates who deviated were categorized by
1967; Heritage, !984). Whilc working on the dec1~10~ making ofJunes their fellow inmates into a set of obnoxious social types; 'kiss ass', 'snitch'
at the University of Chicago in the 1950s he was mtng~ed wnote that and 'sniveller'. 'Kiss asses' were too friendly with the staff; 'snitches' were
jurors - just likc social scicntist.s - were intercsted m dtstmct!OnS such informers and 'snivellers' constantly complained and pleaded with the staff
as that betwccn ·fact' and 'opinion' and in what one can and cannot do These categories were backed up by verbal and sometimes physical abuse.
with evidencc. Furthermorc, they did not seem to learn these distinctions In the same way, the rules could be used to explain certain styles of
Foundations of discourse .analysis 21
20 Discourse and social psycltology
abstracting this comment as an indicator of an explanatory rule and part of r~·r
FIGURE 1.2
The convict codc the Code, Wieder asked what the utterance was doing and what it achieved. C
It occurred to him there were severa! different things going on at once.
8 Basic Ma:dnH' In part, this piece of tal k was fonnulating the immediate social environment
1. Above n\1, do not snitch. in constrained ways. First, it constructed what had just happened in the
2. Do no\ cap out. . convcrsation as Wieder asking the inmate to snitch. It then allowed the
3. Do not take ndvantngc of othcr ressdcnts.

4.--Yoo-
5. Help othcr rcsidents. . , . .
6 Do not mess with other restdents mterests.
i'
\.
resident to not answer Wieder's question because his demand was now
characterized as illegitimate, and to respond would be morally inappro-
priate. Thus the resident had provided himself with a motive and reasonable
7 : Do not trust staff - staff is he~l. ~.' explanation for not replying. Finally, this response rerninded all concerned
S. Show your loyalty to othcr ressdents. ; that this conversation was going on between an inrnate and someone defined
asan 'outsider' rather than a friend, for snitching only operates betwecn
. 1 Rule Eight (show your loyalty to the residents) ¡ inmates and outsiders. In this way the inmate's comment provided a vcrsion
behavtour: For exa.mp e, . 'ded lively conversation with staff and ~ or definition of what was going on in the immediate circumstances; it
can explam why mmates avm. . . . · ¡¡ the Code can sustain Í established thc intcraction in one way rathcr than anothcr.
the lack of enthusiasm fo.r staff lmU~!V~'j,~l;~h~s'e who tlout the official 1
deviant behavio~r, reqluel~l~;a~~~r=~~ :~:id
rules, cover up 10r ru
informing staff about deviant i In general, ethnomethodologists suggest that talk a!ways has this quality.
The naturc of interaction does not arrivc pre-packaged and; prc-ordaincd
but is reproduced on each occasion. To put it another way, the participants
goings on. do not passively respond to what is going on but active! y produce it. In
this case, for example, Wieder and the resident are notjust passively acting
out the Codeas a pre-written script for how inmates and outsiders should
From resource to tapie behave- rather that this role relation is rcproduced for the occasion on
hand by thc inmatc saying 'you know 1 won't snitch'; othcr pieccs afta! k
Through first id~ntifying this Code, ~ieder wa.s ~rying t~ ~~~~~~~rra~~ might have reproduced other role relations.
ho w the traditional social researcher
h
mtght explam mmlatesk 1 sely
hauld now oo · more e o As well as formulating what is going on, 'you know I won't snitch'
~~eh~::~~ ~:d~~~sd~~~~ds~;;~:~ ~:~i:~l:r, at ~~~:~~~~:~sa~~ ~~~:~~~~
icked out in the course of conversatlDOS wt
also leads to certain practica! consequences. First, it acts as a rebuff to
Wieder by negatively characterizing his line of questioning as a requcst
:~: ihen abstracted from those conversations to serve as a.n e~p~anat~~
to snitch. Second, it calls for a stop to that line of questioning, and gets
it. Third, because the question remains unanswered Wieder remains
f r thc researcher. In other words, how certam e atms
resourc~ ~ loyed when the inmates talk in the institution c~me to ~e ignorant about the answcr. Fourth, this utterance signals that the conver-
categones ~~e researcher as the basis for a causal explanatlOn of tts sation might turn nasty if the questioning continued. If Wieder refuses
~~;~end sbyWieder proposed, asan alternative to the tradi~ional procedur~, to end this Jine of questioning he risks being seen as an incompetent. as
an anal;ti~ approach which takes this resource, the Codettse~f, ~s ~el rof;;~
someone who does not undcrstand the Code.
d asks how the Code is used, m practtce, m t e. os_e.
~i~~~~;c~~e a similar approach to rules in Chapter Three: Thls shtft f:~~ Rejlexiviry
using features of tal k as an exp.lana~ory resour~e to ~oo~~~;~~~~:Clogy
ic for research in their own nght ts a standar one m e .
top d p 11 r 1971) In general ethnomethodolog!Cal resear- With these examples from his interaction with the inmutes, Wieder
(Zimmennan an 'do s'¡mplyne • · '
adopting Iay explanations and accounts as
· draws attention to what he describes as the multiformulativc and
chers try to avot k h th object mulliconsequentiaJ nature of tal k. That is, the examples show tal k is not
bulwarks for their own explanations. Rather they ta e t em as e
mere! y about actions, events and situations, it is also a potent and Constitu-
of their study · . 'th ample Wieder tive pari;jthose actions, events and ·situations. The utterance 'you know
We can best see the distinctiveness ofthts move Wl an ~x b . bru tly
I wolf'tsnitch' is not just a description of a rule, it also formulares the
found his friendly co~ve:satiokn~s wi~ in~~:e:n:~:.ld ~~~~:~~~~m~~diaiely n;1ture of the action and the situation and has a number: of practica!
stopped by them saymg you ow wo
Foundations of discourse analysis 23
22 Discours.e aod social psyclwlogy
· . . · Fol\owing Garfinkcl (1967), lndexicality
conscqucnccs. w!thin that sh¡tuatJOn .. of this kind as reflexive fea tu res
ethnomethodo)og¡sts refer to p enomena . A final fea tu re of thc tal k in the half-way hostel is its indexicality. The
basic idea, derived from philosophy (Bar-Hillel, 1954), is a simple one.
of :SI\Vieder looked more closely at the use inmates and staff made If one person says 'my stomach hurts' and then someone else utters the
of thc Codc in their tal k: its rcflcx;vc cha~~~~r :~~=:~t~~;:f;~~ :~~~ same sentence, although the sentence is the sarne, the reference is different.
apparent. For instance, mmates o ten u~e . For example Different stornachs are indexed by the same sentence. Or take 'it's a nice
to encourage: or discourage cert~in thmbgs ohfa~iaef~~~;·organizc a pooi day'- this sentencc could be used as a surprised description of sunshine
'd t w·¡s asked by a mcm er . or perhaps an ironic comment on further rain. In general, indexical
whcn onc rcst en • . . • k 1 n 't organizc a pool
tournament , e
th inmute rcphcd you now
't
ca
Id Jook like kissing ass'. By characterizing expressions are expressions whose meaning alters with their comext of use ~-i;(:
tournam~nt, becausfe t~e w;~de it is displayed as something prohibit~d, (Barnes and Law, 1976). One of the central claims of ethnomethodology
the act m terms 0 f · 11 · period ofobservauon is that the vast majority of expressions used are indexical. That is, their
and this justifies not performing lhe act. In a~t, m ¡s a re ly of this sort. sense and reference are settlcd by looking at features of their context or
Wieder never sawha ~t~fif h:~:~~rg~~~:~:~~!e~pmight well ha ve been occasions of use. Thus for virtually any utterance the listener will 'fill
Yet an excuse suc as in' its meaning with the aid of information about, for instance, who the
ineffective. speaker is, what their status is, what they ha ve said previously, what is
likely to happen next and so on.
In the half-way hostel, then, to makc sensc of utterances, Wieder and
T!ze etcetera ;.clouse the othcr residents had to sort out such issues as the identity of the speakers,
' . d the 'open flexible texture' whether they were staff or inmates, who they were ta!king to, and what
5
Another featurc emphasized m t~: s~uoJe ::s only refe;rcd to in the most the tapie of the conversation had been. For example Wieder would ha ve
ofthc Code On most occasmns e · · r d t novel interprcted the statcment 'you know I can't organize a pool tournament'
al f t~rms and at the same time was constantly bemg app te o l quite differently ifhe had heari:l it said not by an inmate toa staff membcr
~=~,"~f ~i;e~rnstances. ;thnornet~eo1~1~~isrtsa~~~::t~~~t ~;:t~~~~ 1~7~~
1
but from one staff mcmber to another, where it might ha ve been taken
or prescnptt?ns are m et~rm.m~ "necessa to elaborate on a rule to imply the speaker does not know enough about pool or something
Zimmcrman, 1971). That JS, Jt ¡s always !y the:le in a definitive and similar. Ethnornethodologists suggest that people in conversations are
d th ature of the ctrcumstances to app al constantly engaged in interpretative ·work to accomplish the meaning of
an e o ·. (M lk y !979a) In Garfinkel's (1967) terms, rules ways
clear-cut way u a¡' h: h llows novel or unforseen instances to uuerances using their knowledgc of context to help them. Producing smooth
embody an ~rcetera e ause w te a conversation is a highly skilled art which only appears to be natural and

be :::~~~~e~~~e~/~~e ue7c~~~: e~~~~: ;~~\~~a~i:r~z~~~~:tsoint~:r;~~~ unproblematic because we are so practised at it.


Overall, Wieder's study illustrates the difference between the ethno-
genera te the possibility tha~.mm~~esc~:; t~~y desire. It might ha ve been, methodological and traditional approaches to language and interaction.
the Codc specJI'ically to ac ¡c~e :t out of organizing the pool tournament In particular, he shows the limitations of traditional work which has tended
for instance, that the mmate-:' ~ g eally just lazy rather than to treat references to rules basically as descriptions of externa!, objective
by claiming it would be kJssmgl as~ w;~v:r the fact that 'the Code has constraints on behaviour and ignored the flexible application of these rules.
conccrned about followmg the ru e. o . t~ decide what would count In contrast to t.his Wieder looked in detail at what the inmates were doing

~~i~~:;;r ~;~~~~n::~~~tt~::h:~ t~:~ns~~:t:~l~:~~~:~:cs~s;et;;:si~~~c:f:~~~


with their talk of rules; he looked at the way language was operating
practically, within the scene of the action. In Garfinkel 's tenn, he looked
etcetera claúse bUJ!t mto rule systems mean . f nds for a variety of at the reflexive character of tal k.
. . ty of ways to lead to a vanety o e . Although speech act theory and ethnomethodology ha ve been enonnously
d
be use m a vane . . . ses is that it LindernlirieS atterúpts
motives. The problem thJs ambJgUlJlY ;o .sense for this draws u pon the important in the~ development of discourse analysis, and have turned
to use the code in a c\ear-cut exp ana ory ' ) attention to the functional, constructive nature of language, discourse
idea that rules work as a causal template (see Chapter Three . analysis also has its roots in a third theoretical tradition - semiology -
and it is to thís we now turn. _---------.
Fmmdations of di.H:our.l'e mwlysi.r 25
24 Discourse and social psyclwlogy
phenomenon, independenl ~~ ~~~ per aJ~ even an entirely physialogical
what tasles nice is a purely t 1 h
Signs of structure: semiology a few othcr cultures to be disabu ~ref an... anguage, we only ha ve to visit
The Swiss linguist Ferdinand de Saussure (1857-1913) autlined the idea lying behind these ideas the as~ 't o lhJs Jdeifa. Saussure calls the principie
· r I ranness o the sign.
of a science of signs, or semiology, in three courses of lectures given at
the University af Geneva shartly befare he died. These ideas wcre nat
published during his lifetime, but after his death two linguisls pul togelher The arbitrariness of rhe sign
lecture notes taken by his students into an integrated book called Course
in General Linguistics (1974). ll is probably easiesl to get hold ofthe basic Basic to Saussure 's thinking is d' . .
associated speech sound For e a ltstwcthJOn between a concept and its
idea of the notion af semiology lhrough a non-linguistic example.
~~urry ammals WJth four legs which bar k) d lh . h. . .... p
'dou' tfl . : xamp e, on t e one hand, there is the canee t
Imagine going into a restaurant and giving your arder to the waiter.
He brings the first course to you and - horror! - it is the Black Forest sound 'dog'. Saussure calls the h ' an e:e JS t e speech
r~--- speec sound the · ijj h
gateaux and not the vichyssaise. Clearly something has gane lerribly tne signified. and the comb¡'n•a1.IOn o f the two th SlgW. r fr, . . t e. concept
wrong. But what would a semiologist say about lhis? The semiologist would argument for the arbitrariness of th . e mgutsttc stgn. The
suggesl thal the problem wilh the mea\ is exactly analogous ta an ungram-
neither the naturc of the si n· ~ st~n rests on the demonstration that
ordetermined Th' d g ¡fier, sJgnil'ied or their rclau'.onship is fixcd
matical sentence: a specific rule of combination has been brokcn, namely · Js emonstratwn depends d .
that in our culture, in this particular historical period, Black Forcst gateaux. that things could easily have b h on pro ucmg examples to show
een ot erwtsc
fits into a slat towards the end of a meal and vichyssoise flts into a sial h .
betwecn the signifier and th . ·~ d b at t ere JS na natural relation
In the first place, Saussure sug ests th
at the beginning. In between the two we might have Caeser salad or coq
au vin and afterwards cheese and then coffee. There is a rule of arder animals with four legs whic~ s~~~~ ~~h ~tween the word 'dog, and furry
here concerning the way dishes can be combined iota sequences or in speech sounds confirms this· in E. r eh ac~ that other cultures use ather
Saussure's terms there is a syntagmatic rule. It is important to realize that use 'chien'. Any sound could be u~:t w e~e ~e use 'dog' the French
it is not a prescriptive rule that is being identified here, of the kind criticized is essenlially arbitrary Humans abslthe s¡gmfier for d. og, the choice
· are capa e ofpr d · ··
by ethnomethodologists. Synlagmatic rules set conditions for meaning; of sounds and d¡'f'
'
t
.eren 1anguages use cert .
° ucmg an, 1mmensc rangc
d !
meaning and ignore others Th . . am soun s : for conveying
a ce I~ sounds should
they do nat regulate behaviour in the same way. · ere ts no necess 1ty th t na·
Back at the restaurant, when we are ordering our mea! we know there carry same mcanings and oth T
however is that . 'f' d. ers not. he most controversia! claim
are various options for each slot, perhaps prawn cocktail instcad of thc ' sigm te s, cancepts are th ; . ·
examples used to support this claim reÍ m . emse ~es arbJtrary. The
1
vichyssoise or paella instead of the coq au vin. There is another principie
Far instance, in English a distinct' . y k ami y on lmg~Jstlc evJdence.
at work here, one of substitution rather than arder. Semiologists callthesc
rules af substitution paradigmatic rules. The power of this idea lies in and sheep meat, by the use of th:~~~ 7:r ed. bctwe~n she,ep, the animal,
the fact that with twa simple rules it is possible ta endlessly generate more in French this distinction stays unmarked: rt~: s .sh~ep an~ ' mutto~ •. while
or less acceptable meals. If we wished to research meals it would be far both. In English the distinction between ' . m~ e te';" mo~t?n Js. used
importan! to identify these rules; they are a very economical way of making related ta size: rivers have mo l fl . nver and stream JS bastcally
, . . re wa er owmg down th · hil .
sense of the structure af any particular meal. The claim that an underlying a. dtstmctlon is made between •r·IVIere .. , an d 'fleuv ' hem, w1 1e m French
system involving rules af acceptable sequences and cambinations can mto the sea and the former does not The . ~ w ere t le atter flows
genera te and make sense of cultural phenomena is basic to semiology. conceplually partitioned in endle d.'f' Jdea JS that the warld can be
TI · . ss 1 ,crent ways
It is important to note Saussure's suggestion goes further than merely . us •. then, ~~ what Saussure means by the arbitrari~ess of'the .
providing a convcnient and less redundant way of organizing a mass of ts nothmg whtch determines the nature of . . : stgn. There
material. He argues that the underlying system is essential far full lhe signif1ed; therc is no na . . the sJgn¡fier ar the nature of
knowledge af the nature af given items. Or, la pul it another way, we The radical consequence ofthtu,.:~l osr mtnnst~ relationship between them.
., 1Df aussure ts that · . 1
could not fully understand the nature of Black Forest gateaux, say, without be seen as a naming process u . . usmg a unguage cannat
1
an understanding of its place in the system of sequence and substitution lo lhe thing that it names Inste,adSitng al 1St of words each corresponding
. · 1 IS a ways depende t . · · h
which underlies meals. This may sound like a rather odd claim. After all, ot Black Forcst gateaux 0 . . . n • as Wtt the nature
Think of . . ' n a syslcm of rclatJOnships.
it might seem that the essenríal nature of Black Forest gateaux is the sweet, a railway tJmelable. What makes the 8 ·-?5 Geneva-to-Pans . . tram
.
stick)', chocolate flavour on the tengue. Yet, while we tend la lhink that
26 Discoursq and social psychology
Foundations of discourse analysis 27
what it is? It is not the physical nature of the train - it might have
completely di[ferent coaches on it every day. Nor is it the time that it FIGURE 1.3
leaves - for it•is still the 8.25 even if one day it lea ves at 8.35 or even Second arder signification - myth
9.30. What defines it is its place in relation to other trains, as indicated
by the railway' timctable. Thc timetable is the underlying system used to
make sense of particular instances, particular trains.
lt is this ba~ic insight concerning the relationship between particular
meanings and :·an undcrlying system of differences which led Saussure to Leve] II (myth)
suggest that a Separate scíence ofsigns within socicty is canceivable. This
sciencc of semiology (named after the Greek work for sign) would not Signified
be concerned :only with language but would study any human realm to
which meaning is systematically applied (Culler, 1976). And indeed many
studies have :·naw becn carried out an subjecL.;; as diverse as fashion
(Barthes, 1985), road signs (Krampen, 1983), literature (Barthes, 1974), Leve] I
telcvision programmes (Fiske and Hartley, 1978), rack music (Hebdidge,
Signifier
1979), buildings (Broadbent et al., 1980) and many other cultural forms
which can be considered sign systcms.
Probably more than any othcr, the person responsible for developing
this klnd of ~ork was thc multitalentcU Frcnch scmiologist Ro\and Barthcs E.g. Leve! I- car~ Jaguar XJ6; Leve! n- 'wealth' 'luxu ' , '
1
ry 1 speed . 'glamorous living', etc.
(1915-80). He was a central figure in Parisian intellectual life for three
decadc.s befoi"e Oc died. One aspect that Barthes stressed in particular was
the possibilily of additiona\ lcvcls of mcaning in cultural scmiological complex meaning systems associated with h
artifacts andl moreovcr show· th h suc apparently simple cultural
are not natural, inherent ~rope~i~~ of%~s=~·taken-for-gra~ted me?nings
systems (1964, 1972).
eulturally eonstructed eonve t' Th' mgs but essent¡alJy arbJtrary
· n Jons. mgs eould ·¡ h '
wtse. Take the essay on wrestl" B rth eas~ y ave been other-
Second leve! significarion: myth to think of a culture where wr~n~.. a es notes that Jt would be possible
be a sport where the goal is to ds mg !S very much like boxing; it would
Saussure suggested that signifiers (speech sounds) and signifieds (coneepts) . emonstrate physieal .
eu l ture wresthng is not like this at all. Wh prowess. Yet m our
are connect~d together in signs by the process of signification. Barthes not on wrestling? Why do w tl y do people bet on boxmg and
makes the pbint that this process need not stop here; for a sign can take d o they eontinually break theres ers moan wah ag
rules? Barth '
d .
onyan pam? And why
part in another leve! of signification where it beeomes the signifier to a
contrasting sets of meanings at w~rk her:s answer IS that there are two
new signified. Thus the word 'Jaguar XJ6' signifies the eoncept of a
specific car:with a certain body shape, performance and so on. But this . Cl~arly there is pain in both wrestlin a~d b . . . .
JS quite different. In boxing the e g o.x¡~g but Jts SJgmficance
sign can itsOlf beeome the signifier of a new signified at another leve! (see defeat whieh undermines the im xpr~ssJOnfofpam Js a sign ofimpending
Figure 1.3), What is signified here is much more diffuse but is likely to portray. Boxers get up quickly :~ess;:n o exeellenee the boxer tries to
include such notions as 'wealth 'luxury'~ 'speed', 'glamorous living'
1
1
the appearanee ofhurt. Wrestlin ~~w oeked down and try to minimize
the good guy fights it out with t;~ bast:;de~~~s¡¡" moral dJmension where
and so on. 1f we see the line in a novel 'Mark got into his Jaguar XJ6
and drave ,off we are pravided with a set of expectations about the
JS part of the spectaele· it hel l . . outs the rules. Pam here
.status, wealth and pcrsonality of the owncr without any direct information and justice. To hide pain a b ps e adnfy the display of suffering, defeat
concerning· these things. s oxers o would und . h
momcnt intclligibility of the moral d ~rmme t e moment by
Throughout the mid !950s Barthes wrote brief articles for a French
are part of this drama for without th rama .. Suffenng and rule-breaking
ncwspapcr;on second-level signification, or myth as he ealled it (1972). justice would be last. ese thmgs the essential element of
He wrote essays on the semantics of trivia: wine, steak and chips, writers
abroad, Einstein's brain, striptease. Yet their interest lie in displaying the Following on from Saussure's general a
us that thc meaning of wrestlin b .rgument, Barthes' essay reminds
g or oxmg does not arise naturally out
28 Discourse and social psychology Foundations of discour~·e analysis 29

of movements and behaviours. At the first leve! of signification, a particular a: a form of action performed in discourse between jndividuals with-' ~··~·~~-:
fist movement may correspond to a punch, but at the seco~d leve! of ~tffer~nt g~als we ar.e forced to take the social context into account,'l~i-:-1l¡1-.'
signification the meaning of that punch will depend whether Jt Jsthr.own h.ke.wts~, w~th the .not10n that a web of felicity conditions ora systcm 0 ( ·• ·
in a boxing match or a wrestling match. Thus ulum.ately tts st~n~fic~nce dJstmctJOns Js reqUJred for Janguage to be used meaningfully. These things
will be a product of the underlying set ?f conven.tlons and dtstmct.ton~ are not the property or creation of individual persons but;are of necessity
which people use for decoding thc meamng of act.I0~5 and evc~t~ (!or a sharcd across collectivities (Harré, 1979). Taken together, thcsc featurcs
more detailed exploration of Barthes and the semtotics of wrcsthng sec suggcst lunguage should be of enormous interest to soCial psychology,
Webley, 1986). although as y~t there are only isolated and preliminary ,moves to put it
at the hcart ot a progrumme of rescarch. If we wish lo know more about
the way language is actually operating we are going to need to think about
Social study of language social psychological issues.
Despite their importa~! insights, spcech act thcory, ethnomethodology
.In the last three sections of this chapter we ha ve outlined sorne of the basic and sem10logy _are not ':"Jthout their flaws. They are nota fully salid basis
· contributions of speech act theory, ethnomethodology and semiology to on wh1ch to bUJid a socml psychological approach to language. But befare
the study of ]anguage (for reviews and introductions to these. fields we can sta~ t~ d~velop a more viable approach we need 'an appreciation
see Levinson, !983; Heritage, 1984; and Sturrock, 1986 ~espect¡vely). of both the1r ms1ghts and their problems.
This work tums away from the Chomskian/psycholingUJstJC tradJtJOnm
which language is viewed as a formal system principally concerned w1th
describing ór representing the world. This work also retre.ats ~rom ~he Problems and limitations
idea that language is best understood outside the specific occastons m whtch
Speech acts
·.it.is used. lki
Speech act theory and ethnomethodology both emphasize that ta ~ ng
¡5 a species of action. In speech act theory this is enshrined in t~e p:mctple One ofthe main difficulties with specch act theory is that it' was devcloped
that the same sentence, the same string ofwords, can be used m dtfferent as a philosophical thesis criticizing alternative philosophical perspectivos
ways: with the force of a request, an arder, a quest~on a?d .so on. Eth~o­ rathcr. tha~ as a theory able to cope with the vicissitudes of real tal k. As
methodologists ha ve taken this idea cven further by JdentJfy¡~g a reflex1vc such, 1~ m1rrors much ofthe idcalization ofpsycholinguistiC work: it dcnls
dimension to talk. While Austin concenlratcd on categones o( actlon b~sl w1t~ 'm~dc up' sentcnces takcn outside gcnuinc con'texts of use or
derived essentially from an analysis of English verbs, Wieder stressed that h1ghly ntuahzed speech forms in ceremonies like weddings and ship
an utterance can formulate both the nature of the action it is performmg ~~mmgs. I? ordmai_Y.talk participants often do not makc the Uct their spccch
and the relationship between the parties involved in the talking. In additi.on IS performmg cxpllclt. When we want thc salt wc oflen say 'can you pass
he noted that utteranccs typically do not ha ve just one but a whole senes me the salt?' not 'picase pass me the salt'. Yet in this c3se we are not
of consequences for the talkers and subsequent in.terac.tion. after information - although our request is couched in :the form of a
Rather than emphasize function in this way, s~~-~-~-~o~tsts h.ave wor~ed question ~ whar wc want is action: the salt lo be passed. Áttcmpts to try
on undermining simple models of language which conceptua!Jze meanmg to dea! w1th these problems by introducing a notion of indirect speech
as a consequence of the relation between isolated words and obJeCts. acts ha ve appcared rather post hoc and ha ve certainly reduced the heuristic
Instead semiology lays great stress on the existence ~f. a syste~. of power of thc theory (Lcvinson, 1983).
oppositions and differences - 'la langue' -as a prerequJSJte for g¡vmg Whe~ we try to apply speech act theory, to categorize a piece of
sense to individual words. This forces us to thmk about descnptton 1~ a transcnpt mto speech acts say, these sorts of problems beco me acute. In
rathermore sophisticated way taking into account both thewords wh1ch particular, the idea thHt conversation is buill out of discrele spcech acts

l
cif~.u~~c;l.for _des~ription an.d tho~e which a.re not. .F~r semtol.ogtsts have meshed together through sequencing rules (e f. Jacobs and Jackson, 1983;
shown that what is absent ts as tmportant m provtdmg meanmg as what Labov and Fanshell, 1977; Sinclair and Coulthard,l975) is beset by a host
i is present. . of problems. For example, single utterances can perform a liumber of acts
The treatment of language as action moves toward a mo.re socml pers~ec­ at once or acts may be spread o ver a numbcr of Ulterances. :Furthermorc
tive than traditional psycholinguistics. When language JS conceptuahzed in practice the decision about what acr an ulterance is perfofming is afie~
30 Discourse i and social psyclwlogy
Foundations of discourse analysis 31
made by refer~ing to the response rather than to any features of the look not only at those situations wher ' . ' .
utterance itsel(Thus if 'is it nine o'clock yet?' elicits the reply 'yes, just situations where sorne other t . e terronst Is used but also at those
gane' it can beiinterpreted as a question; but if the response is 'alright, erm Is used The ab · ·
as the presence. This is afien a difficult th: sence. !S as Important
1'm just going ;to get thc drink' it scems plausible to interpret it notas researcher to shirk this task and d mg to do and It Is easy for the
a qucstion but t~:'s a rcque.'it to carry out a prc-arrangcd duty (for a dctailcd studies which attempt to gct t en udp condu~ting traditional interpretative
discussion ofthese issues see Levinson, 1983: 286-94 and the important . .. a secan arder Signifi t' h .
J~tuittve means. Indeed, Barthes' (197? e Jea Ion t rough basical!y
work of O rice; 1975). highly suggestive, can be criti . d -) ssays on mythology' although
Semiology tends to produce ~~z:e an. exactl~ ~hese. grounds.
problem has been es . II ' . agam, statJc Jdcahzed analyses. This
Ethnomethodology pecm y rccogmzed by tho h h d .
proponents of semiology The e h . 1 se w o a prev!Ously been
rather than specific use~ ('par~~')asis on 1o~k.i~g at structure ('la langue')
One of the diff¡cultics with ethnomethodological rcsearch is that the kind meaning at a single time rather tha~ coup e wtth a f~cus on examining
of ethnographk study methods (Hammersley and Atkinson, 1983) adopted ruled out a numbe f . processes occurrmg over time has
r o tmportant and int · · '
are prcy to many of the same problems which ethnomethodologists identify analysts such as Barthes, Foucau]t and D er~stmg questiOns. Continental
in other work.,! Por instance, Wieder's description of inrnates' tal k about of the crucial insights from semiolo e eurnda hav~ suggested that sorne
the convict code is based on his interpretation of how they formula te what
is going on and the resulting consequences. In this sense Wieder's work
the stress on underlying structure ;.¡ o Id be mamtamed, particular! y
use and processes of change Th.' w Id':l,"Ying ~ore attention to language
is nota great advance on the more traditional work. Although he is taking as 'post-structuralism') has had Is ~o 1 ~ed sem!Ology (sometimes known
talk as the res,earch tapie, we find this talk pre-packaged when we read we will see, is now starting to :n utge/mpa7t on hterary theory and, as
his work. Put!another way, the reader of an ethnographic report of this in research on the self (see Ch~pt:/;¡~:)ocml psychology' particular! y
kind is depcndent on the researcher's description both for what they know Overall, then, these perspective f . .
about the datri. and for their evaluation of the researcher's analytic con- important insights about languages ~o~. outside psychology embody
clusions (Atkihson and Drew, 1979). Pretty much the same point has been meaning. Although none have articul nc Ion .and the . ~rganization of
mude by disCourse analysts in their criticisrns of a whole range of they provide the basis for this artic~It=~:nsu~~:n~d emp¡~llcal progr=e
sociological !csearch (Mulkay ct al., 1983; Yearley, I984a). next chapter. • e we Wt tum to m the
In effect, it is often difficult to assess the findings of ethnomethodological
work becaust! it uses unexplicated comrnonsense assumptions (although
it is unclear ~hether such assumptions could ever be entirely eliminated
- Ashmore, 1985; Atkinson, 1985). Thus, evcn though cthnomethodologists
ha ve offercd ~ much more subtle approach to Ianguage use, the empirical
bash; is unclc6r. As a response to this problem many cthnomethodologists
ha ve adopted!a ncw analytic strategy known as convcrsation analysis bascd
primarily on !verbatim transcripts of interaction rathcr than an field notes
or the reseafcher's remembered experiences. This will be described in
Chaptcr Four.

Semiology

Thc empirisal basis of semiologica\ rcscurch has also been C¡llestioned


(Culler, 1975). The problem whcn carrying out semiological rescarch,
of course, iS adcquatcly dealing with a system of distinctions where the
vast majority might be nbsent in any one case examined. 1f we are can-
cerned with· the use of lhe category 'terrorist', for example, we need to
Unfolding discourse analysis 33

'would yo u mind ifl borrowed your calculator this evenirig?' ore ven more
obliqucly: 'it is going to drive me mad daing all those statistics by hand
2 tonight' (Brown and Levinsan, 1978). It may be to the speaker's advantagc
to. make a request indirectly bccause it allows the reclpient to reject it
Unfolding discourse analysis w1thaut making the rejection abvious (Drcw, 1984). On thc whale, people
prefer ta head off undesirable acts like rejections befare they happen
In the first part of this chapter we would like to begin the process of (Drew, 1986; and see Chapter Faur).
describing what we mean by discourse analysis although, of course, our . .The ~nalysis offuncti~n thus cannot be seen as a simple matter of categor-
description will only be complete when the book is finished. Wc hope tzmg ~1eces ~f s~ee~h, _n ~.~m.c.rsi.$ . 4P.9.t:l P!~ analx~t ·r~-~-fH!!K~.-~h-~--~C?ntext.
to indicate how a new style of so~io"-psy~Eg!~ica!_!esearch can be erected There IS nathmg mtnns¡c toa camplaint about doing stati~tics by hand 'that
on the foundations of speech aéúheory, ethnomethOCfology and semiotics. makes ita request. It can only be recognized as this fro~ the context.
Throughout this chapter, and the boa!: as a wholc, we will be assuming In th~ calculator examplc onc spccific function is b:eing pcrformcd:
that discourse analysis will become more than a new field of study in social requestmg. However, functions can ulso be more glob~l, and this is the
psychology- that is, discourse added to the list of attributions, altruism, second point to make, A person may wish to present himself or herself
bystander apathy and all the other tapies which have interested social in a favourable light, for examplc, or to prcscnt someo~c thcy dislikc in
psychologists over the last thirty years. Discourse analysis is a radical a paor light. Global seltCprcsentations can be achieved with particular kinds
new perspective with implications for all socio-psychological tapies. The of formulations which emphasize either gaad or bad features. As with
second part ofthis chapter will try to make this claim come alive through the ~r~vious .examp.Ie, it can be sensible to be inexpliciL Por one thing,
demonstrating how other socio-psychological methodologies ha ve ignored exphcitness nsks bemg less persuasive. To prescnt yourscil fas a wonderful
or covered over the constructive, active use of language in everyday life human being to someone, you perhaps should not say ·t am a wonderful
and then the implications of this repressiori for the central socio- human being', but yau might madestly slip inta the conversation at sorne
psychological concept of attitude. 'natural' point that you work for charities, have won an: academic prize,
It is important to emphasize at this stage, that we are not naivcly read Goethc and so on.
expecting all social psychologists to down tools and start working on .' In general, we find that iftalk is orientated to many different functions
discourse. The point is that discourse analysis asks important questions 'global and specifjc, any examination of language over time reveals can:
of conventional research, it provides a workable methodology and we will 'siderab)e variation. A person's account wi11 vary accordirig to its function.
pe satisfied if we can demonstrate that the study of social texts can, in That is;-it will vary a:cording ta the purpose of the talk. For example,
\tself, be a fascinating and potentially significant occupation. 1f we take two descnpt10ns of a particular individual, we Will expect them .
to vary ~n accordance with the :eelings of the person do in~ the describing.
1

If y~u hke a person yo u may, m the course of everyday gossip, describ i, (,)_
Sorne majar components of discourse analysis partJcularly likeable characteristics out of the many available. Someone \) 'i '
wha dislikes that person may emphasize very different characteristics, ar
Fwzction, construcrion and variarion the same Iikeable characteristics may now bccome disag~eeable. Alterna-
tively, imagine yo u ha ve to describe a pcrson to a claSe friend on orie
One of the themes strongly stressed by both speech act theory and ethno-
occus.io~ and ~o a parent on another. Again, what will b'e picked out for
methodology was that people use their language to do things: to arder
descnptmn will vary. For instancc, you are prob<;~bly nclt going to focus
and request, persuade and accuse. This focos on \anguage function is also
th~ parental account on Lhis person 's daring acts of dclin:quency, but this
one of the majar components of discourse analysis. Function, however,
m1ght be a more appropriate facus for a clase friend.
cannot be understood in a mechanical way. Unfortunately, as we all know,
~~~t ___ i~ ~~pperiing in these cases (s that peoplc are using thcir languagc 1\\ ' ·
whcn people are pcrsuading, accusing, rcquesting etc. thcy do nol ulwuys
. tqG·on~·rr~Lct..pcrsions. of ~he ~ocia! \\fOrld. Tj1e principal~f discoursc
1

do so explicitly. When someone makes a requcst - perhaps they want to


analysts 1s that functwn mvol ves construction of version~. and is demon-
barrow your calculator- they do not always politely but explicitly ask:
strated by language variation. The term 'construction' is a)Jposite for three
'could I borrow your calculator this evening, picase?'. Often they are lcss
re~sons. ~ir~t, it_rem~n~s us that accounts ofevents are built out ofu variety
direct than this, perhaps couching the requcst as ·an abstract qucstion:
al prc-ex1stmg lmgutstlc rcsources, almost as a house is c'onstructed from
34 Discoursi! ami social psychology Unfolding discourse analysis 35

brick~. bcam~ and sn on. Sccoml, con~truction implic~ active ~clcction: focusing on discourse as u tapie in its own right. That is, wc are not trying
sorne resourceS are included, some omilted. Finally, thc notion of con- 1 lo rccover events, beliefs and cognitive processes from participants'
struction emph~sizes the potent, consequential naturc of accounts. M~ch i discourse, or treat language asan indicator or signpost to sorne other state V.
·~ of social interaáion is based around dealings with events and people whtch of affairs but looking althe analytically prior queslion of hgw discourse~ · · .
G•0:5
l_

".- . ·\ ¡are experie. nce~ only in tcnns of speci11c linguistic versions: In a prof?_~~d or accounls of these things are manufactured (Gilbert anct-M'i'i!kay, 1984;
·...... t sense account:S '.construct' reality.
Poller, Stringer ana"WahNelt;-l9SZI). ~·- ~ ~ · '
l -wc\ío. not :,wish, however, -to make the process seem necessarily Take the idea of attitudes. If someone espouses attitude X on one occasiou
deliberale or itltcntional. It may be that thc person províding the account and th_e contradi.ct~ry attitude y on another, the ~naly~t clearly cannot treat ;_ __ _
the extstence of att1tude x or y as an unproblemattc gUide to what the person -.~ ·. ~
1
is not consciou~ly constructing, but a construction emerges as they merely
try to make scnse of a phenomenon or engage in unselfconscious social actually. believes. B~t it is possible to. treat the _account containing the _ x:J
actívities like blaming or justifying. It is important to note that in these ~xpr~sston of the attttuct: as the focus ttself, asking: on what occasions ·~)
cases, too, thCre is variability in accounts, because different forms of ts att1tude x rather than att1tude y espoused? How are these attitude accounts .
descriplion may be righl for differenl occasio~s, but the persa~ ~ay be constructed? And what functions or purposes do lhey achieve? It is .
jusl '<Juing whal comes nal).ltally' rather lhan mlenllonally decJdmg lhis questions of lhis kind which are.at the .heart of discourse analysis .._....··
ralher thañiliiit~gu~ge wi!l be appropriate. Indeed, we expecl To summarize, then, this first stage, discoursc analysts are suggesting-. 1
this to be the d!uch more common silualion in lhe hurly-burly of ordinary that: ::
Ianguage use, 'and we are nol beguiled inlo thinking lhal sorne classes of l. language is used for a variety of functions and its use has a variety/ -~.
tal k are mere! Ydcscriptivc whilc othcrs are deliberately constructive. All of consequenccs; / .
Janguage, evcn language which passes as simple description, is constructive 2. language is both constructed and constructive; \
and consequential for the discourse analyst. 3. the same phenomenon can be described in a number of different ways; \
Overall, thcn, discourse analysts' propase that people's language use 4. thcre will, therefore, be considerable variation in accounts; '¡

is -much- mord_yariabl_e than. is .indicalí_!d by _the widely hcld 'realistic' 5. thcre is, as yet, no foolproof way to deal with this variation and to 1_
descriptive mbd~Í- ~-f language - which treats discOurse as a relatively sift accounts which are 'literal' or 'accurate' from those which ar~.. -
unambiguous :pathway to actions, bcliefs or actual cvents. 13--~~~ªf_C_~.ers rhetorical or merely misguided thereby escaping the problems variation
who presuppose the \ealislic niode as sume lhal when people descdbe the raises for researchers with a 'realistic' model of language;
same evenl rtcliañ·ar-oelieftheir accounls will, broadly speaking, be 6. lhe construclive and flexible ways in which language is used should
consistenl. And, for melhodological purposes, lhey will take consistent themselves become a central tapie of study.
accounts to m:eanJ.b.aLC..vents did hapQ.e.ll..as._d_e_~f!!l'~t;d. Brenner provides So far we ha ve illustrated the need for an analysis of participants'
a good examrear this realistic position when he claims that 'descrip_tions discourse from theoretical principies, although we have been guided by
and explanations of evenls and experiences may be regarded as h1ghly the research associated with the positions discussed in the Jast chapter.
valid, when they are, by and l.i[g~,Jh.,..~_ame _acrossaccl.lunts' (1985: !56). Nevertheless, it is clear lhat our perspeclive depends on an empirical claim
There are tWo basic problems with thlS: First, consistency in accounts as its mainstay, namely that there is considerable variation in participants'
is often overstated by the various aggregating techniques commonly used accounts; or, more specifically, that there is sufficient variation in accounts
by psychologísls, as we will indicate later in the chapter. Seco~d, there to cause problems for the realistic approach.
¡5 no reason to supposc that consistency in accounts IS a surc Indicator We intend to provide documentation for this claim later in this chapter
of ctescriplive validily. This consistency may be a product of ~ccounts and in subsequent chaplers. However, readers may be wondering why,
sharing the s~me function; that is, two people may put the1~ dtsc~ur~e if this variability is as common as we ha ve suggested, it does not feature
together in tl1c same way because they are doing the same thmg w1th il. widely in the current empirical literature of social psychology. How can
,c!.r· (;,;j( ;:~.. ~!'~·( ._,_., __ , .. ----~~ . ,:,_,· it ha ve been overlooked? There are two answers to this question. The first
¡ '· (. ( .,, .~ . ' .... is that it does feature in social psychological research - but it is not

l) \~
'·. ! •
(
.., understood within the bread discourse analytic framework. The second
thnt_y_n.~J~~i!i~Y.!t:JJICCQt},ni_~.l~ ..,'~,anaged
'
Discourse as: tapie
answer is . through analytic strategies
Discoursc an'aiysts havc responded to the all-encompassing functional/ of restriction, categorization and selective reading. We will take these facets /
constructivc hature of accounts by suspcnding the rcalistic approach and in turn. .~'-1. • - 1 ·. - ' ·
36 Discourse and social p.!.t)'chology Unfolding discourse ana/ysis 37

Variable discourse and traditional social psychology Se(f-presentation

There are a number of strands of experimental social psychology which Therc is a large literature in social psychology conccrnCd with pcoplc's
illustrate account variability, The areas of social perception, impression self-presentation or impression management (Baumeister, 1982; Tctlock
rnanagement, cognitive dissonance and speech accommodation have in and Manstead, 1985). One ofthe most reliable findings in this area is that
particular brought inconsistencies in language use to the fore. people modiry their behaviour, including their talk, in accordance with
different social contexts. To ingratiate themselves, people afien resort to
gross flattery and insincere compliments directed towardS those they wish
Social perception to influence. When they have succeeded in their aim, lhc compliments
may be replaced by hostility. Variation in self-presentatiqn is particular! y
In the field of social perception, studies such as Duncan's (1976) have relcvant, of coursc, in experiments where subjects can change their actions
demonstrated that people provided with the same kind of scenario will to picase experimenters or in linc with the researcher's subtle cues, meeting
nonetheless describe that scenario in quite different ways. Duncan showed the demand set up in the experimental situation (Orne, 1969; Rosenthal
his experimental subjects a ftlm in which either a black or a white man and Rosnow, 1969).
shoved another person. In each case it was exactly the same action, al1 Once again, account variability is illustrated in social psychology, but
that changed was the race of the protagonists. The viewers, who were the theoretical perspective is rather different. Discou~se analysts see
white students, then had to categorize this event, for example as 'playing vadation in accounts as a consequence of people performirig a whole range
around' oras 'aggressive behaviour'. Duncan found that his subjects were of different acts in their tal k. Sorne variations may be dueto considerations
rnuch more likely to describe the event as aggressive behaviour when it of face saving and creating a good imprcssion, but, as !we will shortly
was the black who was doing the shoving, thus apparently indicating thc try to demonstratc, othcrs will rcsult from thc necd to caristruct discoursc
effect of their st<;_L~P.types o~. their perception of events. to achicve an effect - a blaming, say - or arise from a broader concern
TmsS!mlfc.értainly shows variability in accounts- even where people with making accounts locally coherent, where there is little sense in talking
ha ve seen the same kinds of actions with their own eyes. This finding is of active sel f-prcsenlation. Variubility is conlincd to a narrl~W arca of social
generally interpreted, however, according to theories ofsocial perception as/ life, if we see it as resulting mcrely from impression r-Danagcmcnt.
an example ofhow racial prejudice and stereotypes can distorl perception.'¡
Account variability of this kind is no_naken.as a more ge_n_e_ral and normal'
pl;¡_~~9_f!.l~On, as is the case with disco~xs~~ª!l,alysis. Moreover: no attempt Cognitive dissonance
is made to look at the construction Of the accounts or their possible function.
Indeed, this kind of anaJYsiSTs' precluded by the experimental design which Research in social perception and self-presentation active! Ystresses varia-
allows participants only an extremely limited range of descriptions, and tion. Cognitive dissonance or cognitive consistcncy theories, on the other
provides these descriptions for the participants in advance. hand, focus on people's desire to appear consistent ahd invariant to
, ,( j ~~ The point we ~1sh to _!!!~~ _is that people are always constructing versions themselves and others. Festinger ( 1957) argued that if a pf'rson perceives
., .. 'i anO re-describing events, not merely when prejudiced or stereotyped. The dissonance or inconsistency between two or more of lheir:cognitions they
\ 'i's'tiidy of social perception largely concerns how people tal k about other will experiencc an unpleasant state of psychological tenS.ion and will be
~ople; it is a linguistic studyas muchas an investigation ofvisual processes. motivated to reduce the tension by changing their attitudes and thoughts

Y( Uescripti~ns of others are i0evitably 'distorted'. not simply occasionally,


but peremal1Y, m the sense that they are always constructiOns for sorne
p~se. Different pieces oftalk and prescription have, of course, differcnt
functionsagd different consequenccs. A racially prejudiced description is
so they are once more consistent. For instance, if somcmle trying w lose
weight also fancies having a cream cake for afternoon! tea, they may
rationalize the contradiction by telling themselves that they, will go without
dinner as compensation. The dissonance pcrceived is red4ced by the new
highly p·er.niC\d~s in its effects but other descriptions are no less functional self-stricture. In this perspective variability is not secn as a normal or
and constructive. This point is overlooked by social psychologists who natural feature bul as a psychologically unpleasant statc'.
investigare 'bias' in information processing, assuming a rcalm where sorne It is often the case that íf people become aware of cdntradictions or
descriptions will simply and ncutmlly rcncct rcality. Thc mcaning of thc inconsistcncics immcdiutcly, as thcy are talking, thcy will httcmpt tn gloss
variability found in these studics is thus frcqucntly misundcrstood. over or makc sen se of thc inconsistency. In our socicty i,riconsistency is
\. 1 • ·• - •• C.~ '.
_1'.-~t::. ..._¡ \ (_ Í\ ~'-. .1· ,-. ..: ..1 \ • 1

--/c}•.r .\._.··.~ 1. ·_.r:: ~-- : .•• _·: '+ .~:·:..\''":"'. :1


38 Discourse rmd social psyclwlogy Unfolding discourse analysis 39

rarely a desirabl~ way ofprescnting oneself. However, dis~~urse analys~s say and write rather than how they say it in terms of phonology, intonation
al so demonstratcis that these cases of self-conscious recognttmn and repmr and so on.
of variability are relatively small in number. As we will try to show in
the Iast section bf this chapter, variability is still pervasive in people's
accounts. Givcri just a small separation of accounts in time, across one The suppression of' account variability
or two turns of t<ilk, people cease to attend to the variation in their languagc
use. It will becdme evident, in other words, as we progress through the Social psychology not only tends to play down the issue of variation in
boa k that variab\lity is an expected usual featurc of conversation and social arcas crucially conccrned with it, but it is al so the case that the procedures
texts despite th~ fact that people often try to reduce it when it is pointed psychologists regular! y use for dealing with diseourse have, often inadv'éi'tl' '""'
out to them or ~vhen it becomes salient for sorne other reason. ently, acted as management strategies for Sb)Jpre~s·fng variability. For the
Another poini should be made he re. N ame! y, the type of in~onsistency sake of simplicity, we can distinguish three broad strategies which obscure
described by cognitive dissonance theory is not a formal or stnctly log1cal variability in accounts. These involve restriction, gross coding and selective
inconsistency. Jt is a psychological inconsistency. Somethmg may be d'
rea mg. \_(·~
<

• '
_··"''-- \) .::··'-' .
-')-~ dJ :_,..{~ v:t.~'c._tu.~'"' "
perceived as indonsistent by a logician but not by the pe.rson conr,e~ned, ._\ ·;-1-· '7~J'.., '_):;l.de.t.L•-' .1 ,;-._.:c.~:r_.·..l.-J ....
and vice versaL Consistency hnd inconsistency are h1ghly negotJable
occasioned phe~omena, in fact. On sorne occasions sorne variati?n~ in Restriction
accounts will be!seen as inconsistent, on other occasions the same vanatmns
will be seco as sbnsiblc and rational after all. To put it crudely, consistency The restriction strategy involves techniques which prevent the variability
and inconsisten~y are variable stntes themselves and one of the things which in participants' discourse becoming in any way apparent. Experiments use
interests the disdaurse analyst is how they are uscd, variably, as argumenta- restriction very effectively, beca use the control exerted over the subjects
tivc or rhctoric:al strategies. and the situation to measure the effects of the independent manipulated
variables also works to exclude information which might potentially throw
up variability in discourse. Of course sorne experimental work can highlight
Speech acconii~JOdation variations in language use, as our discussion above indicated, but typically,
very rcstricted parts of participants' talk are sampled in very constrained
Work on speech aecommodation in social psychology has looked at the circumstances.
factors that surtound speech and the variability in this domain. Ha~, for Experiments normally take place in social psychological laboratories;
instance, speakcrs modify their accenl, dialcct or intonation pattern etc. the cxperimenter sets up a scenario within the laboratory for the subjects
in diffcrcnl gr6up contexls (Giles, 1977; Giles ancl St. Cl~ir, 1979). For who are asked to respond to it in a way which can be numerically
example, Wclsh speakers ha ve becn shown to broaden thCJr accents ~~~en quantified. For example, thcy may have to play a game and choose between
talking toan upper-class English speaker, with the goal of emphas1zmg spccified options where the number of choices for one option across the
thcir group disiinctiveness (Bourhis and Giles, 1977). Vanabd1ty m these sample becomes the dependent measure. The experimenter's goal is to
kinds of speesh dimensions can readily be related to the context and constrain the situation in such a way that it can only be perceived by
function of sp~ech. What distinguishes discourse .analysJ~, however: IS the subjects as the researcher intended. Other interpretations must be
that we also stlldy variability in linguistic content In relation to funcuon ruled out, although it is clear that this ruling out is frequently ineffective
and give thi5 <l priority. . : and controversia! (cf. Mixon's, 1972, discussion of Milgram's, 1963,
Qverall, the~efore, there are rcsearch tapies withín expenmcntal soci~l obedience research).
psychology w~ich document variabili.ty in acc~u~ts. But non~ of t~Js The subjects' reaction to the situation must also be as constrained as
research addrcsses the broader implicauons of vanatmn. Companson Wlth far as possible if the experiment is to be successful. There must be no
these research !arcas clarifies our perspective. We are cóncerned with the ambiguity about this reaction, responses must fit into pre-categorized
functional aspects of all kinds of tal k, notjust that seen as 'biased' in sorne entities. Subjects cannot behave in just any way, they must select their
way, or invol"cd in impression management. All types o.f ~anatJOn are response according to the options offered by the experimenter. Similar! y
interesting, nOt simply logical inconsistencies or vanatrons people they cannot usually give more than one response, they cannot choose two
recognize imrriediately, and we examine linguistic content or what people or three options or mark severa! points on a scale, and their response is
40 Discourse and social psyc!wlogy Unfolding discolirse analysis 41

generally measured only once. That is, it is not usual for the experimenter Gross categorization
to return ten minutes later and say to the subject 'what do you think now?'
There is usual! y no scope for recording uncertainty of response, varied The techniquc social psychologists havc traditionally ~sed to dcal with
reactions or changes in opinions. In other words, experiments are situations participants' open-ended discourse is content ana1ys¡s (Holsti, 1968;
where the value placed on consistency ofbehaviour in our culture is made Mostyn, 1985). This in vol ves the generation of categorics which can be
particular! y salient to participants. In general, experiments are designed reliably coded and imposed over thc data for thc purpbses of hypothesis
to wipe out variability of interpretation and response, indeed, that is testing. For instancc, in McLelland's (1961) wcll-kn'own rcsearch on
supposedly their strength and rationa1e, a1though they may be obscuring 'achievement motivation ', content analysis of achicvemcrlt themes in books
one of the most interesting and important features of social life in the was used as an index of the stress on achievcment in a Purticular society.
process. Coders were trained to rcliubly catcgorizc the.mes slress;ing high uchieve-
In surveys or opinion poli research, restriction arises similarly from mcnt so thc prevalcncc of achicvcment themes could b!e correlatcd with
collec.:ting a very constrained selection of participants' discourse on broad indicators of national development such as gross 'national product.
ane discrete occasion. lt is often only possiblc to rcspond 'yes', 'no', In principie, therc is nothing wrong with this proccdurc!. Yet, in practice
'don't know' or 'agree', 'strongly agree' etc. toa survey question. The it is more suited to research where the discourse is und~rstood primarily
poss_i!üLi_ty_ o[ a .respondent giving contra~[lng views on a_t~pic is again as an indicator of something lying beyond. lf wc ad'opt a theoretica1
prééluded; ambivalence, the expression of flexible opinions tailorcd perspcctive in which language is seen asan essentiully FuOctionalmedium,
to the context and inconsistent responses are ruled out by the response where the meaning of terms is closely related to their Particular context
forma t. of use, content analysis is Iess helpful.
As with experiments, these restrictive practices do not prevem informa- Function is a problem for content analysis beca use of the difficulty of
tion pertinent to people 's variable language use appearing altogether. For dealing with the sheer subtlety of a situation whcre pa~ticipants muy be
instance, there is a considerable literature showing that subtle variations constructively using their language to produce different sorts of effects.
in question wording can lead to large differences in responding (see Marsh, As ethnomethodologists have emphasizcd, when a person is trying to
1982) and that peop1e contradict themselves and make 'incohcrent' claims persuade, for examplc, part of thcir cffcctivcncss may: wcll depcnd on
when responding to opinion questionnaires (Kinder and Sears, 1985; producing their talk in a way which conceals that it :is an attempt to
Schuman et al., 1983; Turner and Krauss, 1978). Unfortunate1y, these persuade; for what could be more persuasive than the 'jnere description
phenomena are rarely treated as theoretically interesting indications ofthe of facts'? Context is a problem because, if the meaning of a section of
way peop1e dep1oy their 1anguage, but are generally treated as obstacles discourse depends on the context in which it appears.! the criteria for
to the production of reliab1e research findings. · what counts as an instance in a particular coding categqry may become
Both experimentaÓon and survey research carne under sustained attack impossibly complicated. '
during the 1970s; experiments because of supposed1y unwarranted assump- More importantly for thc present argument, the broadicategories often
tions about the causal constraints on behaviour (Harré, 1979; Harré ,' used in content analysis (e.g. achievement themes) can easily obscure
and Secord, 1972) and doubts about the generalizabi1ity of findings : theoretically interesting differences in discourse. Broa~ categories are
(Argyris, 1975; Gergen, 1978); survey research because of the severe ,particularly likely to be used when researchers expect co~sistency. At the
Iimitations on the responses it is possib1e to make (Rosier, 1974; Harré, same time, the fact that different coders can re1iably operate the same
1979) and because of the difficu1ty of deciding on the proper interpreta- category - the acid test of a well-conducted content analysis - does not
tion ofthose responses which are produced (Cicourel, 1974). In the Iig_12t\. resolve this problem. For two coders, trained in the !same way, can
of these k.in. ds of criticisms, considerable interest was expressed in the \.l(· pcrsistently repeat the same confusions and categorize tqgether different
use of_Qrulli:-~nd..k4lt:~l!:!:'iews and procedures based on the interpretation ·~ \ sorts of utterances. This 1eads to a big prob1em in eva1uating content
of dÜcuments, which ha(fbeen more common in arcas of micro-sociology : analytic research, because even in the most detailed rese~rch report very
(Piummer, 1983) and ethnography (Hammers1ey and Atkinson, 1983). few examples of the application of the category to the discourse or the
With the advent of this work, the variability in participants' discourse training of coders are reproduced (Abraham, 1984).
could no longer be dealt with merely by restriction;_ new strategies -~
which _\V·~-~jlJ_ Gall.gross.categorizaü9.Q an~ selective ré<i"(ifñi"'-= ..~a-d to be
adoj:lled.
42 Discourse ¡and social psyclwlogy Unfolding discourse analysis 43

mi~ht throw it into doubt. What is required is an analysis of discourse


wl11ch focuscs on variability and thc constructíon ofaccounts. Howevcr .
Problerns of thi~ kind have !cd sorne social psychologists, notably Rorn befare making any more moves we must illustrate in more detail how ~ ¡
Harré (sec Cha¡)ter Three), to advocatc qualitative research which works discoursc analyst approaches accounts. This goal can be attained by
with the accourhs themseivcs as opposed-tcr-::lñU"'íñeriCally transformed demonstratrng how we would deal with one of the most fundamental social
version of themJ This approach has been commonplace in micro-sociolagy psychological notions: attitudes.
(Bogdan and l]ay!or, !975; Plurnrncr, !983) and when adopted it is
ccrtainly more !difficult to ignore variability in participants' discourse.
Ncvertheless, ¡fthe analyst is operating with a commonsense 'rea!istic' Attitudes in discourse
rnodel of Iangu'age, the rnodel can be sustained by selective readmg. ,
One forrn ofj.se!cctive reading takes pla~e when the analyst ha~dhng\ Our approach to attitudes should reveal that distinctiveness of the discourse
interviews or texts selects out those wluch appear to be stgmficant p~sition ?"d put some flesh on the notion of variability in accounts alang
when !istcning f to the tape or reading the docurnent. The great danger : wtth the Jdea that.~counts ~§lli!.g_~dJqhave specific c;onsequences.
here is that th:e res~archer n;aking selections will sirnply rnirror his,'.,('> Thcse are sorne ofthe thernes which will recur frequent!y in !ater ch~apters.
ar her prior expectatmn. In thts sttuattan the data c.an be used to ~r.mpl~J' J Fmt, let us clarify what social psycho!ogists mean by the term 'attitude'.
buttress the fa¡oured analytic story rather than bemg used to cnticall1, The concept of attitude is one of the oldest theoretica! ideas in social
evaluate 1t. ¡ psycho!ogy and, in manifold ways, has supported a huge body of sorne-
If the resear;cher does notice variability he or she can use another times disparate research. Despite this history, the exact rneaning of the
form of reading which selectiv.ely reijies and/or ironizes the discours.e. notmn has remamed somewhat obscure. Indeed, in a classic overview
Reification ref~rs to the process where abstractions are treated as matenal Gordon Allport (1935) suggested that the very irnprecision of the rneaning
things or, in t~is case, where words referring to objects ~r processes of atl!tude IS one of the features which has rnade the notion so popular,
are treatcd as guarantees of the actual existence of thcse objects or pro- becausc Jt glossed over difticult theoretical conflicts. A recent review
cesses. Far example, a spoken account af an event which describes it as (McGuire, 1985) also describes a nurnber of radical! y different perspectivos
a quarrel migh\ be taken to indicate that thc ~vcnt was in fact a quar~cl. on the nature af attitudes, and we will return to the most important of
Ironization is !thc rcvcrsc. Irony occurs, of coursc, whcn a meanmg thesc later. Howevcr, McGuirc gocs on lo suggcst that it is possible to
different ar oPposite to the literal one is in tended. lroniza~ion refers. to formulatc a working dcfinition which would command a wide degree of
the process wh~re descriptive language is treated not ~s genumely ~escnp­ canscnsus. .. .. ·
livc but as ha~ing another purpose or as a dcceptton. The notrons af McGuire claims that empírica! studics of attitudes work with at least
rcification anct' ironization can be combincd to e reate a selective reading im.plicitly, thc t'oll~w.ing basic definition, When peoplc are ex~ressing
of accounts arld thus eliminate variability once again. att!ludes they are g1vmg responses which '!acate "objects of thought"
With these t\vo approaches to reading at their disposal, analysts working on "dirnensions ofjudgment"' (McGuire, !985: 239). That is, when they
with open-end~d discourse find it. relatively easy to co~struct o~e c~herent are speaking or actmg, people are taking sorne idea ar object of interest
story of events, processes or behefs out of the rnatenal at the1r d1sposal. and giving it a position in an evaluative hierarchy. We will describe a
Versions whibh are in line with their preferred story can be re1fied and concrete exarnple to make this concept sharper.
others which ~onflict with it, can be ironized. Reification and ironization
can be' very suptle, but nevertheless can be revealed by ~ ~areful, and critical
examination or the way analysts draw upon parttctpants dtscourse Traditiona/ approaches to altitudes and racism
(Mulkay, !98,1; Mulkay et al., !982; Potter and Litton, !985; Potter ~t
al., !984). In the next chapters we will demonstrate how a c!ass1c ethogeruc In 1976 a British researcher, AJan Marsh, asked a randorn sarnple of !,785
study based ~n Harré's perspective adopts these devices. . . people to express their attitude to 'coloured imrnigrants' by placing a mar k
Qverall, it ¡is possible to see how the use of these three strateg¡~s.- .¡ ~\!1 on ~ scale which ran frorn 'cornpletely sympathetic', through to, 'no
restriction, gnpss coding a~d .selective.readi~g- can suppress the va~abthtyfj¡ !J(J ~ feelmgs about them either way', to 'complete!y unsympathetic'. In
discoursc anal)1Sts claim ts mherent m saeta! texts. These strategtes pre-:! McGuire's terrns the object ofthought would be the 'coloured immigrants',
'
suppose the ·~ealistic model' of language use by obscunng . any data w h're h.J:1' i
while the d1rnens10n ofjudgernent would consist ofthe 'syrnpathy' which
44 Discourse and social psychology Unfolding discot~rse analysis 45

the respondent can offer or refuse. Marsh's survey resembte: my:iads this is reflecied in the very title of his article, which is called 'Who ha tes
of other surveys, the techniques he used are extremely common .In ~ttltude the Blacks' not 'Who hates those people who are bothi recent settlcrs in
research. Having collected his responses, Marsh went on to spht hts scale Britain and black defmed'. His terminology is not neutral. Ir you ha ve
up 'logically' into categories. These are labelled 'very hostile', 'hostile', lived in a country for the whole of your life you might be concerned if
'neutral' and so on (see Figure 2.1). people start calling you an immigrant- a tenn often used to connote aliens
or outsiders (Reeves, 1983; Sykes, 1985). ·

FIGURE 2.1
ID A second problem arises when we examine the transformations which
Marsh makes to his subjects' responses. Ifwe look at Figure 2.1 we can
Distribution of sympathetic and unsympathetic feelings towards see that Marsh has transformed one dimension, running lfrom 'complete! y
coloured immigrants
unsympathetic' to 'complete! y sympathetic', into a mote complex set of
No feelings about Complete\y labels: 'very hostile', 'hostile', 'unsympathetic' etc. There is no coherent
Complete\ y
unsympnthetic them either wny symputhetic justification for making transformations of this kind. For example, it is
probably wrong to suggest respondents mean the same tJ]ing by the words
46-55 56-79 80-99 IDO 'very hostile' and 'complete! y unsympathetic'. For onio thing, the term
o 1-20 21-45
hostility is often used to imply an active disposition, Y,hiie if someone
20% 10% 3%
12% 13% 17% 25%
lacks sympathy, they are without a certain kind of activ~ dispositian. By
Symputhctic Positive Vcry making this transformation thc analyst is riding roughshod over subtle
Very Hostíle Unsympathetic Neutral
positive distinctions that may play a crucial role in the participanis' discourse, and
hostile
certainly in thcir mcthods of making scnse of the !wrJey qucstions.
No. in samplc: unwcightcd = 1.7H5. wcighted = 1.41C!; 'dnn't km,w' {cxcludcd =4% ). (j;.¡ A thinJ prnblcm al!lo conccrn.s tran.slaliun: in thi.s ca.sd thc rc.scarchcrs'
_,., transHnion of purticipants' responses into the underjying thearetical
Source: Marsh, 1976.
category of attitudc. The aim of attitudc sea les is nol me~ely to show how
pcople lill in thcsc scalcs, but lo identify attitudcs. Thal iis, in McGuire'.s
From the point of view of a discourse analyst,. th~re ~r~ a numb~r. of terms, to identify where on a specífic dimensiona personjlocates an object
interesting points to be made, both about McGuJre s mmtn.wl defimtJOn of thought; in the current example, where the respondent~ !acate 'coloured
of attitudes and the kind of practica! research procedures Jllustrated by immigrants' on the dimension of 'sympathy'. Thc cruci'UJ assumption of
Marsh 's se ale; we will concentrate on three issues. attitude researchers is that Lhere is sometbjng enriJJdng within eeople which(~~
First, there are obvious problems with the status of 'coloured immi- the scale is measuring- the attitude. '
grants' asan object ofthought. Onc way oflooking at the term 'coloured 1 Discourse analysis points to many difficulties with this.! Wc need to ask,
immigrants' would be as a simple category Iabcl for a group of people, !\for instnnce, whether peoplc filling in an attitude scal~¡ are performing
in fact those people who fit the descriptions 'coloured' and 'immigrant'. a neutral act of describing or expressing an interna! m~ntal state, their
However, things are a lot more complex than this. For example, there ,attitudc or whether they are engaged in producing a specific linguistic
is no clear-cut neutral way of deciding how to apply the category 'coloured )formulatian tuncd lo the context at hand. From thc di~course analytic
-.L~immigrant'. That is, there are no objective criteria for category me:rnbership )perspective, given different purposes ora differcnt contex~ a very difTerent
· ' (this is a theme which will be explored in detail in Chapter Sjx). 'attitude' may be espoused. Put another way, if a certain attitude is
"!' The proper application of 'coloured' is dependent on unstated. theo;ies of expressed on one occasion it shoüld not necessarily lcad Us to expect that
\:) race and biology. But modero theories of genetlcs and populatJOn g1ve no ..the same attitude will be expressed on another. lnstea~ there may be
support to the idea that •races' of people can be distinguished in terms of systematic variations in what is said, which cast doubt bn the enduring
unambiguous, underlying physical, and ultimately genetic, differences homogeneous nature of the supposed interna! mental a~titude.
(Husband, !982). In addition, 'immigrant' means (in the dictionary sense) How, then, should wc deal with lhese ihrcc problcms Which are by no
a person who comes into a foreign country as a settler. Yet Marsh (1976) means uniquc to Marsh: first, thc meaning of interprcta~ion given to the
does not address the problem of splitting 'coloured immigrants' from terms in the attitude scale~ sccond. the translation betw~cn participants'
'coloured residents', and it is clear that he takes the term 'colaurcd immi- ctiscoursc unct amtlysts' catcgorics: and thinJ, thc treatmGnt or lingui.stic
products as transparcnt indicalors of undcrlying objccts ,or dispositions.
1

grant' as a bland descriptive category covering both these group.s. In fact


46 Discourse !and social psychology
Unfolding discourse analysis 47
More gencrally ,¡' what might a study of participants' discourse tell us about anti, and the position espoused is pro. Following standard attitude theory,
phenomena tra~itionally understood in terms of attitudes? The time has we w~uld treat th1s speaker as possessing a specific attitude. Jf they had
come to get do~n to the nitty-gritty of accounts and perform our own to fill m Marsh's questionnaire they might endorse the 'sympathetic' end
ana\y.sis. of the scale - or so the traditional account would have it.
Yet, when we look at more of this sequence, the simplicity starts
to fall away. Here is the entire turn of talk from which Extract One was
Discourses of i~nmigration taken.

_, --(iJ,n the remaind~r of this c~apter .we will indicate how a disco~rse analyst /¡ 2. Respondent: I'm ~ot an~i the": at ?11 you.know, I, ifthey're willing to get
r:.:..~)('-"'),r'hJghL go- abou~ rese~rch10g a_!t.Jtu_?~~~s.Qn~!.t\t~t~_Q_fª~~nes such_as J on and be ltk.e us, but 1f they re JUSt gomg to come here, just to be able
_...A:-:-r!'"'\ 'coloul'eél ¡_mm¡grants . Wc willClosely examtne sorne accounts produced to use our socmJ welfares and sruff Jikc that, then why don 't they stay ho e?
-rf~-al~-~~--~~g~nTzed environment than Marsh's survey, but which, none- ~~~:~. m
theless, are eas\ly recognizable as evaluative expressions concerning race
and immigratidn. All the accounts we shall analyze have been extracted There ar: a number. of interesting features here which immcdiately question
from open-end~d interviews with \Vhitc, middle-class New Zealanders. our ~rst mterp:etatmn. To begin with, the 'pro immigrants' claim is made
These intervie~s discussed gcnerally 'controversia!' issues in New Zealand contmgent on rmmigrants cxhibiting a willingness 'to be like us'. Thus
society. we can no longer read it as an unqualified expression of sympathy.
Thc goals ar!aur analysis wi\1 obviously differ from thosc detcrmining ~oreover, the w1.1~le statemcnt is organized within a complex linguistic
traditional attiiude research. Broadly speaking, discourse analysts are structure of condrttonals and contrasts. This can be shown clcarly if we
interested in the different ways in which texts are organized, and the rearrange the extract.
conscqucnccs dr
using so me organizations rather than others. So our aim
will be to loo k iat the different forms takcn by evaluative discourse about 2b. A 1 If [they'rc willing to gct on and be like us]
A2 then [I'm not anti them]
minority grou~s, and the effects of these forms. At the same time, the bu!
analysis wi\1 trY
to avoid thc thrcc problems we identified as endemic in
Bi if [they'rc just going ... to use our social wclfares]
traditional attitude rcsearch, namely presupposing the cxistcnce of thc 82 thcn [why don't thcy stay home]
'attitudinal object', making translations from unexplicated participants'
discourse to urlcxplicated analysts' discourse, and treating utterances as In technical tenns, contrast structures are now revealed. Taken in isolation,
indicators of tl{c prcsencc ar enduring, undcrlying attitudcs. We sha\1 try the co~sequences of. t~ese kinds of contrast structures are not easy to
to show why the conccpt of an enduring attitudc is thcoretica\ly redundan l. ascertam. However, lt 1s notable that studies of the way public speakers,
1

such as Marga_ret Thatcher and other politicians, elicit applause ha ve found


that co_nstruct!Ons. of th1s sort are very effective in producing audience
Context apprecmt10n (Atkinson, 1984a; Heritage and Greatbach, 1986). And it
":"Y well be that this kind of construction is commonplace in everyday
Pcrhaps the nist thing which becomes apparent when embarking on this dJscourse because 1t helps package the message to make it more convincing .
task is the shccr complcxity of working with extended scquences of talk . A further feuture of Extruct Two also highlights its persuasive orienta-
rather than thcibrief isolated utterances which make up responses to attitude tton. It draws upon what Pomerantz (1 986) has called the extreme case '11·
questionnaire~. Take the following intervicw extract for example. formulation. For example, if someone is asked why they carry a gun and
they respond, 'everybody carries a gun', they are providing an effective
J. Resp01ul~nt. I'm not anti them at all you know (Benton: 26),_ warrant. Gun-toting is depicted notas a notable or restricted activity but
as nonnative, something shared by everybody. Extreme case formulations
We do not hav:e any troublc in reading this as a relatively positive statement take ~h~t~ver evaluative dimension is being adopted to its extreme limits.
of the speakei-'s position on 'therh' - in this case, in the New Zealand Thus If 1t !S a question of numbers, then it is 'everybody' or 'only one'
context, 'Poly~csian immigrants'. In attitude terms, the 'object of thought' thmgs are 'very' or 'terribly' bad etc. In the second part of Extract Tw~
is 'Polynesian: immigrants', thc 'dimcnsion of judgment' lies from pro to the speaker produces un extreme case formulation of this type: 'if they
48 Disco/IrSe and social psychology discou ~rse
Unfolding 1 analysis 49

are just going to come here just to be able to use our social security th!!!g_as_~!:l_n___attitude' oran enduring, underlying state iexpressed in tal k
and stuff. The repeated use of the word 'just' paints a picture of people and behaviour. This position becomes extremely difficuli to maintain when
whose sale purpose in coming to New Zealand is the collection of social we look at the variations which appear in participanL~' accounts.
security, a selfish motive much more blameworthy than, say, coming to
provide essentiallabour but being Jaid off dueto economic recession. By
representing it in this extreme way the criticisms are made to appear more Variability
justifiable.
Finally, if we look back to the first part of Extract Two- '1 am not The following example is typical of lhc sort of variation in accounts which
anti them at all' - we can see that this operates as a disclaimer. Hewitt has now been documented in a wide swathe of different ~inds of discourse.
and Stokes (1975) define a disclaimer as a verbal device which is used These two extracts are taken from subsequent pages of the interview 1

to w~Jtpotentially ojmQxipus.attributions. Thus, if someone says '1 transcript.


am no sexist but. .. ' they are aware that what they are about to say may
sound sexist, but are trying to head off such an attribution. In this case, 3. ~esp~nde~/1. Wh,at .r would Ji .. rather see is that, su re, b~ring them ['Polyne-
the speaker is disclaiming possible attributions of racism consequent on smn tmmtgrants] mto New Zcaland, right, try and tr~in them in a skill,
the suggestion that immigrants should 'stay home'. and encourage them to go back again (Pond: 17).
Now, these interpretations of Extract Two are tentative. They are not 4. Respon~ent. 1 think that if we encouragcd more Poly~csiuns and Maoris
based on a systematic study of many instances but on a detailed reading lo be s.kllled pcople they would wunt to stay here, th4y're not um as uh
of a single one. But they suggest two things. First, that even a small amount nomndJc as New Zcalanders are (Imen 1iewer. Hnhn.) sO I think that would
of additional information about context can throw into quesfrOñ·wnat~ at be better (Pond: 18). !

first, appears to be a reasonable interpretation of á person's utteáiñce.


Second, that discourse has an action orientation; it is constructed in such The comradiction is sta.rk. 1~ Extract Three the respond~nt states that thcy
a way that particular tasks- in this case blamíng and disclaiming responsi- would hke Polynes1an lmm¡grants to be trained in New Zealand and then
bility for the obnoxious effects of this blaming- are facilitated. This is, to return to the ~acific Island.s. In Extract Four thc resbondcnt claims it
of course, just what is expected in the light of developments described would be better tf Polynesiuns were encourugcd to beComc skilled and
in Chapter One. then stay in New Zealand. What are we to make of thi$ variability? Thc
These points ha ve important implications for attitude scale research. If problem ts parttcularly acute for the attitude rcscarchcir beca use of thc
the person filling in the scale is viewed as merely describing or expressing co?flict.b~twccn versions. An attempt to recover the per$on's 'underlying
their attitude, things seem quite clear-cut. Yet, if we start to view their attttude ts not going to get very far. 1

response as a discursive act, which it always is, things become murkier, The discourse analyst's repense is rather different from the attitude
because there is a great deal of scope to perform·different kinds of acts researchcr. We d~ not intend to u~e the discourse as a p~thway to entítics
when filling in the scale. For example, a person might fill in the scale or phenomena lymg 'beyond' the text. Discourseanalysisdoes not take
to perform the task of disclaiming by marking the 'sympathetic' pole; or for granted that accounts reOect underlyiágattifüdes orÍdispoSili~ns.and
they might perform the task of blaming by marking thc 'unsympathetic' thcrefore we do not expect that an individual 's discourse lvill be consistent
pole. They might hesitate bccausc they see themselves as sympathetic and and cóher7n~. Ra~er, the fo~us is ~n t~e discourse itself. h~w it is organized
unsympathetic at the same time - '!'m not anti bu t. .. '. Two people putting and what tt JS domg. Orderltncss tn dtscourse will be vie~ved as a product
the same mark on the scale could well be doing very different things with of the arder! y fimctions to which discourse is put. !
'\.
If we rcturn to thc uccounts quoted ubove, and prmdde a bit more of •.
their discourse. If the opinion pollster is coordinating an interview rather
than requiring paper and pen responses the person might offer the whole thc context, we can illustrate how a functional analysi~ might begin.
utterancc to the pollster and how it emerges, in tcrms ofthe category scales,
will depend on the pollsters current mcthod of scaling. 5. hlli'rt•iell•er. [Dol ynu think that, suy, immigration frornithc Pacilic lslamls
shou!ú be cncoumgcd [ J to a much lurgcr cxtcnt thJn it is'! lt's fuirly
One way we could proceed, given this line of argument, is to suggest rcstncted at thc mmnent.
that attitude mcasurement might survivc in its present form if it bccamc RL•s¡umdenl. Y_cs. Um, lthink t~1cre's some prnhlcms in, in cncnuruging
a more subtle business, more scnsitivc to thc Ui!Tcrcnt ucts pcrformcd. ~J~a.l .too much, IS that thcy come m uncducalctl ahout ou~ ways. and Jthink
We should note, howcver, thut this continues to assume that therc is such a H s 1mportant thcy unúcrstand what they're coming H~. 1, what 1 would
.------
50 Discourse a!1d social psychology Unfolding discourse analysis 51

¡¡ .. rather se~ is that, su re, bring thcm into New Zcalan~, right, try an_d is based upon this. If the object is not the same for different people thcrc
train them in': a skill, and encouragc thcm to go back agam becuuse thc1r is no sense in comparing attitudes and the notion ceases to ha ve utility.
dcpendcnce Ón us will be lesser. I mean [ ] while the people b~ck thcre However, when we cometo look at the detail of people's accounts this
are depende~t on thc peaple being here earni~g money to ~end 1t back, I separation becomes virtually impossible to sustain. Far from the objec¡
mean that's a very very negative way of lookmg at somethmg. [ ~ people of thought being a simp~J!!c~.ª.<Ü'..P':~~!'.fl_l_!!_gti_t:Y.L the object is formulated/\
really' should!be trying, they should be trying to help their own natJOn first
and constructe~)n __ c;Ij~~-ourse. in the course of.ctOillg~e·valuatlO~- /
(Pand: 17-18). "iaJCellle. (óíiowing extract, for example, which is part of an ans~er
6. Polyncsians, !they are doing jobs_ now tha~ white people woubldn't do. S~
1

to a question about Polynesian crime.


in many scct9rs of of thc commumty or or hfe, um, :ve would. e very mue
ata Joss without them, I think. U m, what 1 would J¡ke to see IS more ~ffort
bcing made io train them iota skills, skilled jobs, because we are w1thout 7. Respondenr. Then again, it's a problem oftheir racial integration. They've
gota big racial minority coming in now and so they've got to get used to
ski\led peopl:e and a lot of our skilled people, white people, ha ve left. the
the way of life and, er, perhaps rape is accepted ovcr in Samoa and Polynesia,
counlry to gd Lo other placcs. 1 think that if wc encouragcd more Poly~esmns
but not in Auckland. They've got to learn that. And the problem's that a
and Maoris ~o be skilled people they would want to stay here, the~ re not
lot of pcople coming in with mental disease 1 think it is, because there is
as uh nomddic as New Zealandcrs are (lnterviewer. Haha.) so 1 thmk that
a lot of interbreeding in those islan .. islands. And that brings a big, high
w~u1d' be b~tter (Pond: 18). incrcase of retards and then peoplc who come o ver here, retards perhaps
and they/1
Now we ha ve a ¡bit more of the context available we can see that the lnten,iewer. 11 and that causes problems?
qucstion of Poly~esians rclurning to lhe lslands. is related to ~~ifferent Re.\pondem. And that's prctty general I know (Johnston: 20-1).
issue in each extfact. In Extract Five returmng lS related to the problem
of 'dependence' .! The speaker expresses concern that if Polynesians stay In this passage the speaker is notjust giving his views abollt 'Polynesian
in New Zealand they will become dependent upon the1r mcomes to support immigranL'', he is formulating the very nature of the Polynesian immigrant.
pcople in the Pa~iflc Islands. The. speaker sugg~sts it wo.uld, be better 1f That is, he is not working with a ~cutral description of an object and then
they returned w~th skills to contnbute to .thet: own natiOn . . saying how he feels about it; he is constructing a version of the object.
In Exlract Six,i on the other hand, returmng 1s related lo proble~s Wilh _lt is in this way cyaluat~W.l~y_¡:.>!_. H1s version of the object carnes
thc New Zealand workforcc. Thc speaker suggcsts that Polyncstnns are offhis evaluation. Polynesian immigrants are floridly depicted as a group
doing dirty jobs ictisliked by 'whitcs', so lheir leaving would precipitate who are involved in rapes and are carriers of 'mental disease'. lt is implied
economic probl~ms. Moreover, she goes on to cla.lm, the em.¡gratlon of they are from a culture which cannot control its desires properly, something
skillcd whites has lcft a hale in the Jabour market wh1ch Polynesmns should they will have to learn to do befare settling in New Zealand.
be trained to fiÍl. So the speakcr's two different versions of whcther A central fea tu re in this speaker's construction of' Polynesian immigrants'l 0 (;"--·
Polynesians ougllt lo stay or not can be seen to flow Jogically and naturally is his use of broad explanatory principies. He_dgesnol .rest at merely · .J'\.
'_\j from the formulritions in the surrounding text. 1t ts, of course, onl~-~rysJble -.~scribl~gp~!!llmena;-:_he-explain~ theiiitoo. Speciflcally he accounts for ~/, ~,,.,
--.... to adj_~~~-,9_1~~-· ~ .~~_sp_¡;¡_n._?_e __t~ _a ~C!p_i~ _~ccordiñg lo _theco~1~~t. However, the prevalence of mental d1sease m terms of s1mple farmyard genetics; r.ti ,,
tliiskind of adNstment tends to be overlooked by the attJtude researcher it is a consequence of 'interbreeding'. This drawing together of description, -1./ ·;'-
who would cxpcEttlie-spcaker to be able to articulate on a dccontcxtua]\z~d evaluation and broad systems of explanation is dramatically illustrated in
·¡scalc a static cory.stanl attiludc regarding w. hethcr Polyncs1ans should stay the next extract from anothcr speaker.
v} \ or rewrn. 1f it ~s not static and co~stant then much of the pmnt of th1.S
' 'y, kind or mcasunrment technology c\isap~ears. ; ·::<_ '.__ ·. __ S. lmerviewer. Yeah, so [crimcl is pnrtly sort of immig., its rclated to
1
immigration?
.<. . Responden!. Ycah, wc don't, seeing thcm coming through, offthe aircraft
al night, hall' of them can't speak English. U m, if they can't speak English
Consri wrion thcy're not going to be able to get ajob, they're going to go and be in their
littlc communities and they'rc not going to be able to contributc anything.
In traditional attitucte theory, the attitude is considered to be scparate from And they're going to get frustrated because they're going to get bored. And
thc 'object al" tJjoughl'. The enlire logic of altitude measurement, where they'rc going to. you know, therc's nothing for them todo so the kids are
a scale ¡5 used tP compare different people's attltudes to the same obJect, going to start hanging around in the s!recls. At home Mum and Dad can'¡
52 Discotlrse and social psychology Unfo/ding discourse ana/ysis 53
speak English and so the kids can't speak English. They go to school and accounting (Billig, 1987; Reeves, 1983; Wetherell ahd Potter, 1986,
suddenly they are confromcd with English- 'wc can't speak that, and what f-;;¡;ihcofiimg'). Theories of stereotypes presume that pe\\plc carry around
do wc do'?'- nothing. And so by the time thcy get lo fif1ccn thcy just drop
these biased images and reproduce them on demand,i while discoursc
out. They huve had it up to hcre with school and it's not thc school's fault.
They have brilliant lives, thcy have brilliant livcs back in, family lives, analysts have noted how the same person can produ¿e quite different
back in the Islands and that is wherc thcy 5hould be (Joncs: 16). stereotypical categorizations depending on the demands of the situation
(Wethcrell et al., 1986).
There are many fascinating features of this passage as a rhctorical
construction. But the central thing to note is the way the final claim -
that potential immigrants should stay in the lslands - is warranted through Beyond altitudes
the use of an elaborate psychological and sociological story starting with
a charged image of Polynesians slipping off the plane at night, going Earlier in this chapter we illustrated three problcms elaguingtraditional
through language difficulties encountered by immigrants, which is taken attitude research using Marsh's study ofattitudes to 'co!Qured immigránts'
to cause unemployment, and, eventually, the children's alienation from asan examplc. Thc problcms wcrc: thc status of the 'Qbjcct' which thc
school. All this is contrasted with the 'brilliant family lives' available if attitude assesses; the dubious translation from parti~ipants' terms to
they had not come to New Zealand. Throughout the passage we see a analysts' categorics; and the assumption that attitudes ar~ enduring entities
,~f t ~mplex inrertwining of description, expianahon affd -eValj.iatiOif which gcncrate equivalcnt responses from occasion to o4casion. Wc ha ve
we
Cruc¡J.lfY' whit we
find when eXUmine naturally occurring discourse .seen how discoursc analysis deais With these problerris. Wc will now
detail is that the distinctio~ between '~bject or thought' and position quickly revicw sorne of thc responses attitude theorists ri1ight make to thc
('- a~'du_:n~n.swn of JUdg~_:~t Jbecomes vtrtually tmposstble to make. It variation demonstrated in this analysis. '
pc_,J t ~'seems ~f(Ústinction is an artifact of the way attitude scalcs are put Marsh adopted a simple one~dimcnsional scalc in _!his study: othcr
..,cl.l·'·~-': together: all respondents are supposedly reacting to the same object of workers have produced multidimensional scales which are intended to
:· ... • •'."' thoGght. Y et, as we can see from the extracts discussed in this chapter, measure many different aspects of racist altitudes (Mc~onahay, 1985).
'·- d {l.?.. . . . ·~meness of wording does not necessariiy mean that respondents will However, while this path to increasing methodological sqphistication may
v understand the terms or formulate the object of thought in an identical provide a mensure with a number of different facets, lt does not avoid
way. We have seen how different respondents formulate 'Polynesian the general problems raised above. The variability in p~ople's discourse
immigrants', and how the same respondent might reformulate this notion cannot be explained mere! y as a product of a more comp:lex multi~faceted
on different occasions. If a reseªr~.h~_[J~.~UY .!Vishes lo get to grips with attitudinal structure which a more complex scale canj asscss, bccausc
of
racism then a vital part iheir activity must b~-¡¡:¡.;-investigation of how the view~_ex,pr!!S§ed vary so _radicailyfr:om_occasion tb __ occasion. It is
., r -;; ctes.fí:i¡)tiDnañcte·,q,lánatíon_."Ef! n;~shed.togetbera~ct how differcnt kiii_ds impossible to argue that the claim, Polynesian immigrdtion is des_i.rable
·..) -- · Q[_explanations assume different!<lm!~_o[o_bJectso' supply the soc1al_"'_orld and the claim that it is undesirabie are mere! y facets !of one complcx
with varymg objects. attitude. The notion of enduring attitudes, even multid~mensional ones,
One way out of this dilemma might be to argue that Extracts Seven and simply cannot deal with this. i

Eight simply illustrate the use of stereotypes of 'Polynesian immigrants' A social psychologist familiar with attitude research ~ight argue that
by people with 'prejudiced altitudes'. However, this interpretation relies there is yet another way the attitude theorist can cope¡ with variability
on the idea thatthe speakers are using a distorted picture (the stereotype) and maintain the basic concept of attitude. There is, \or example, the
of un unproblematic objectively identifiable category ofpeople (Polynesian complex theory generated by Fishbein and Azjen (1Q75) which tries
immigrants). Unfortunately categories are as complicated as 'objects of lo explain why altitudes frequently fail to predict behaviouf. It is commonly
thought' (as we will illustrate in Chapter Six), and psychologists do not found in attitude research that people will say one t~ing, or express
ha ve access to the true, factual nature of Polynesian immigrants to allow one kind of attitude, but then will behave in a way whicih is inconsistent
them to assess the distortion in participants' versions (Potter and Litton, with this attitude (Wicker, 1969). This kind of variadon differs from .
1985: 82-4). the variation iliscourse analysts note as we are suggcstingi.that what people/
Rather than loo k at racism in terms of distorted perception, we are more say or their attitude will not be static either but just as ~ariable as their
concerned with looking at the racist consequences of different forms of behaviour. Nonetheless, we should not ignore Fishb~Cin and Azjen's
·------------------( . . ·-. --
contribution.
'
54 Discours~ and social psycho/ogy
Unfolding discourse analysis 55
Azjcn ( 1975) arguc that thc path b:twcen at~itudes a_nd
'

l{..--Fishbcin anctj
; behaviour is indircct and muddled. When pcople dec¡de to act tn a certam
wayAheluJeci!ion is only partly influenceci~Y:~hejr_reléY.ant.attitudeS; it is
In response to these difficulties, the discourse approach shifts the focus
from a search for underlying entities - attitudes - which generate talk
and behaviour to a detailed examination of how evaluative expressions
also influencec! by theirjudgeméñts ·af the normative desir~!J_il_ity ()f the
are produced in discourse. Two central and novel questions become
action (what w:ill other people think?) and so··an. Tne--éxpression ol' an
dramatized. How is participants' language constructed, and what are the
attitude in a pa~icularbeliavTóiíranarffiwilrálways be modified by many
consequences of different types of construction? Whether at the end of
other variables. Presumably the same argument could be extended to
this examination space is found for sorne modified notion of attitudes is,
discursive actsi. The question is, can a theory of this kind explain the
as yet, unclear (cf. Lalljee et al., 1984). All we have done in this chapter
phenomena dis~ourse analysis demonstrates? Do the New Zealanders we
is indicate how an analyst might begin to address these questions. However,
interviewed act~ally ha ve underlying consisten! attitudes, although they say
we hope we ha ve given an initial demonstration of sorne of the limitations
many contrastirig inconsistent things in practice? This is difficult to resol ve. of traditional research and the promise of discourse analysis.
How can we teb, for example, which expression is the 'genuine' one, and
which is merelY 'normative'? There is always the danger ofrnassive post
hoc interpretati~on to preserve the notion of attitude. Given enough modify-
ing variables hÜge flexibility in response can be explained! although there
must come a Point when it is no longer useful to continue stressing the
underlying attit).lde. Ou.r view is that a much more powerful explª@!\9~n..C~l!!l­
be given if thc reseafcher looks at the organization of discourse in rclation
to function and!context. This kind ofanalysis usefully supplants attempts to
decodc gener~l afunctional decontcxtual principies such as attitude.
In summar~. a brícf anaiy!-;is of some extracts from interviews has
highlightcd th~ imporlam;c ora numbcr or phcnomcna which ha ve bccn
rclatively neglected in traditional attitude research. We stressed first
thc importune~ of examining context. Contextua! information gives the
rescarC!lCr·-·a--h1Uch--ful'lcr··uhderstanding of the detailcd and delicate
organization o:f accounts. In addition, an understanding of thi.s organiza-
lían clarifies the action orientation of talk and its involvement m acts such
as blaming a~d disclaiming.
Thc second !phenomenon we illustrated was variability. A high degree
of variation in! accounts is a central prediction of the discourse approach:
widely differeht kinds of accounts will be produced todo different things.
On the other hand, considerable consistency must be predicted ifpartici~
pants are producing their language in the light of sets of attitudes which
are stable acrdss different contexts. Y~~y of the kind seen in detailed
studies of dis¿ourse is thus a considerable embarrassment to traditional
attitude theorles.
The third phenomena we noted was the construction of the attitudinal
object in discc}urse. The customary view is that attitudes are about di?ti~ct
entities. Attitlldes to immigrants, for instance, shouldconcer~ 3:11 extstmg
out-there-in-the-world group of people. Yet when we examined·actual
1

discourse thisi simple 'word and object' view ofattitudes became unwork-
able. It is cledr that the attitudinal objcct can be constituted in alternative
ways, and th6 person 's evaluation is directed at these specific formula-
tions rathcr than some abstrae! and idealized object.
MakinJ.t ~·u/es \\'ork 57

distinction in Chapter One when we discusscd Chomsky'~ (1965) contrnst


between the speaker or hearer's kno1vledge of languagc (dompctcncc) und
3 the actual way peoplc use language in concrete situations! (performance).
~cco~d~ng to this model, if we have properly identif¡ed a spcakcr's
Making rules work lmgUisttc competence we ha ve a means of explaining howi the speaker can
produce grammatical sentences - although this by no m!eans provides a
complete explanation of their speech. ;
During the last fifteen years, as we noted in previous chapters, there has
Harré duplicates this argument for social life in gene~al. He suggests
been considerable debate about the adequacy of the experimental method
that all action, whether it is performed through talk or otherwise should
and the fonn of explanation prevalent in social psychology. The critics ha ve
be explained by reference to the actor's 'social competente·. The,posses-
castigated traditional researchers for displaying a number of vices such as
attempting to ape mythical versions of what goes on in the na~ciences,
sion of this social competence sets limits on the acts onei can perform in
social situations and determines what will be accepted as the correct
explaining all social pror.esses by reference to the cognitions of individuals,
performance of these acts; although, just as with linguislic competence,
and tending to treat all human action as a consequence of causal mechanisms.
we need to know many additional things befare we hpve a complete
The most coherently articulated alternative to traditional research which ·
explanatton of the performance of acttons. As an examp)e of this notion
emerged from this debate within social psychology is known as ethogenics
wc can consider the actions involved in having a dinner Party. The hosts
(or the identification of the generative 'mechanisms' which give rise to
are usually socially competent, in Harré's sense, that is, ithey know how
behaviour) and is associated with Rom Harré and his colleagues (Harré,
lo behave in this situation, just as an English speaker ~enerally knows
1979; Harré, Clarke and De Carla, 1985; Harré and Secord, 1972; Kroger,
whether a sentence is grammatical or not. The guests, t9o. know which
1982). Jnstead of experiments, the methodological approach these workers
acts are sui.tabl·e· and which are nol. When discussing the ~inncr party thc

M
dvo~a_te relie_s largely.on.!he_in_vestigation of accounts; and ratherth¡¡n
next da y wtth tnends, guests and hosts can draw once mofe on thcir slOck
eafching for causallaws, the aim·;s··tc;·¡aerúrty·the- fuTes·a·nd conventians
of social knowledge to give an intelligible description o~ what happened
sed by social acto re; ta generate lheir behaviour.
and lO CVaiuatc thc SLICCCSS or raiJurc of thc party, i
·-In this chaptcr wc are going to concentra te nn onc central racct orcthogcnic
Thc central ctlmgcnic hypothcsis, thcrc!Orc, is that pcoplh posscss a ston.>
theory: the identilication of rules. Our aim is to highlight thc dil'liculties
of social knowledge which cnables them to both act and ~o givc accounts
which arise when ethogenists- or indeed any social researcher using this
suc~ as explanations or justifications of their action. ImpPrrant methodo-
analytic strategy - try to extrapolate sets of rules from people 's tal k which
log¡cal consequences flow from this hypothesis. Most n:otably, beca use
are thcn used lo explain their behaviour. It will become clear that the primary
the same set of cognitivc resources are thought to undetÜie both actions
problem with this form of analysis arises from its unsystematic approach
and their description/justification, it should be possible to idcntify the nature
to discourse. We will start by giving a brief general introduction to the
ethogenic approach and then willlook in detail at one study which purports
o~ specific actions by using the actor's accounts. The anal~sis of accounts
wtll reveal features of people's social competence whidh, in turn, will
to identify social rules: Marsh et al. 's (1978) 711e Rules ofDisorder; a piece
reveal the significance of their acts. Thus from listcnin~ to people tal k
of research which exarnined the closed and seerningly violent world ofBritish
about dmner parties, the anthropologist from a non-Weslern society can
football fans. This study will be contrasted with research conducted by
learn the underlying social competence and thus undérstand what is
d¡scourse analysts on scientists' use of rules when choosing between different
happening when next invited to dinner. i
theories. Throughout the chapter we will be emphasizing the fundamental
At this point one might object that surely it would lbc much more
importance of making a detailed study of participants' rule discourse.
stratghtforward to obser¡'e the behavi~ur _directly, rather ¡than follow this
rounda.bout route of gettmg peoplc to gtvc an account of it; Unfortunately,
The ethogenic perspective accordmg to Harré, this simplicity is beyond our grasp, be¿ause the actions
of human beings within a culture are far more than mere mbvements which
Linguistic comperence and social competence could be easil~ de~cribed und catalogued. Movcments an~ behaviour only
take on meanmg m thc context of spccific, and often v~ry local, social
The clearest way to approach ethogenic theory is through its use of the conventions. To. an observer it may seem obvious that twci pcople shaking
basic distinction betwecn 'competence' and 'performance'. We met this hands are grcctmg onc anothcr. But this obviousncss is· not duc to any
58 Discourse !and social psyclzology Making mies work 59

natural relation~hip between the movements involved in shaking hands regulative and interpretative rules - to produce effective and legitimate
and the act of greeting; rather it derives from the fact that both the people social activity, and to mesh their activity with other people' s actions.
shaking hands ahd the observer understand the same social conventions.
It is these share d conventions which allow them to see the movements
1

of hand shak.ing as an act of greeting. Rules ro arder disorder


The local natllre of our conventions can be seen when we think of
the problem of iilentifying greeting conventions in very different cultures. The implications of these ideas will become more vivid if we describe
Thc observer mi~ht well think that two Maoris from New Zealand rubbing a specific rcsearch projcct. We will look at Marsh, Rosser and Harré's
noses together {verc expressing intimacy - yet those familiar with the (1978) study of the social world of British soccer fans, one of the most
conventions of Maorí culture would know that they were simply greeting elaborated examples of ethogenic research which illustrates extremely
each other. Ani untutored observer of a greeting between two British clearly the centrality of rules in this type of exp!anation of behaviour
soccer fans might well conclude that they were witnessing nota friendly (Marsh, 1982).
nct but an exch~nge of insults. The general point ethogenists are making Soc'cer fans and violence at football matches featured very prominently
herc is thnt actions have a semiological character (Swartz and Swartz, in the news media throughout the 1970s, and now in the 1980s debate
1982; Totman, 11980). That is, thc relation betwcen brote movcment:; still ragc5 about causes and solutions. For many pcople, soccer violence
arid their meanihg is not a natural one, but is dependent on the existence seems to breach civilized values, expressing almost animalistic, aggressive
of a network of social conventions which al\ow certain movements to drives. The principal goal of Marsh et al. 's research was to show that
be scen as thc ~erformancc or a particular act. As we saw in Chapter thc apparcntly unrestraincd, aggressivc behaviour of the football fans
One, when reviewing semiology, e ven the basic human movements we is, in fact, tightly structured and rule bound; and thus what seems like
regard as expr~ssions of pain can mean different things in different a threatening chaos of disorderly actions is in fact closely guided by
contexts, for exb.mple when produced in a wrestling rather than a boxing a social competence shared by the fans. If this analysis was correct,
match. For this !reason the social rescarcher is dependent on the accounts, it would suggest, controversially. that soccer violence has been radically
there is no way ~eople' s interpretations of their actions and understanding misunderstood.
of social convc~tions can be bypassed through mere description of what The starting point for the research was a body of accounts (interviews
goes on. and conversations bctwcen fans) and videos made of thc fans' behaviour
In Chomsky'~ theory linguistic compctence is taken to be a speaker's on the soccer terraces. From this material Marsh et al. built up a picture
mastery of an uÓderlying set of rules. Harré makes the same case for social of the normal events in the life of a football fan and, in particular,
competcnce. Hd argucs that ít is madc up from knowledge of sets of social confrontations between rival groups of football fans which provide the
rules which en~blc peoplc to act proficicntly and rationally and, further- arena for soccer hooliganism or football violence. Confrontations typically
more, to display that proficicncy and rationality (Harré, 1974, !977b). takc place between the fans who support the home team and those who
One of the mai~ goals of ethogenic research is thus the elucidation of the have travelled to support the opposition, or the away team. The point of
rules that makq up the social and conventional knowledge of a person, these confrontations seems to be to try and take over the other group's
subculture or sbciety. 'territory' (usually home and away fans are seated in different parts of
These rules fdll into two basic classes (Collett, 1977). There are regulative the ground) orto make the opposition fans run away when threatened,
rules which areiused to guidc behaviour down appropriate pathways. This thus acknowledging the superiority and valour of the successful group of
concept very m;uch resembles our everyday notion of rules: 'do not walk fans. Marsh et al. argued that in the various confrontations which occur
on the grass', 'remove your hat in church'. On the other hand, there are between rival groups a complex structure of rules was operating. These
interpretative rilles, which allow people to assign meaning to movements rules helped organize any conflict in such a way that few fans were actually
and events. Thb rule in games of soccer which states that kick.ing the ball hurt. Actual incident or injury. they suggest, are rare given the number
between two White posts counts as a goal is this k.ind of rule: it is not of confrontations tak.ing place, despite the picture of violence and mayhem
just guiding be~aviour, telling players how they should play soccer, it is painted in the newspapers. In their view the newspaper stories considerably
saying that ond sort of movcment of the ball is to be interpreted !15 the overestimate the extent of the violence.
act of scoring.!Harré's suggestion, then, is that people draw upon a set For example, Marsh et al. claim that fights'between rival fans are afien
of cognitive r~sourccs - their shared social competence made up of ended by the loser ceasing hostilities and look.ing downwards and away
60 Discourse and social psychology Making rules work
1 61

from his opponent (the participants in this study were overwhelmingly safety- which keeps the fans acting thc way thcy do. Thb accounts which
maJe). This display of submission would, in most instances, stop the emphasize violence and destruction make sure that w~en the fans get
aggression befare either party was hurt. In many confrontations, there involved in the action it will be exciting: they seem td be risking their
was no actual physical contact at a11; both sidcs could satisfy the require- Uves! They have a rich vocabulary for constructing stori¿s after the match
ment that their honour be maintained by a ritual chase. The fans who did about their glorious and exciting actions. Yet, at the sarrÍe time, they also
the chasing would gain the highly visible, symbolic victory of seeing the know it is really safe. They are not put off taking part ~y fear of injury.
opponents off and the fans running away could often gain face or honour Thus, paradoxically, the fans' accounts of disorder pi a~ an essential and
because of the patently tactical and risky nature of their retreat. Indeed, purposeful part in the maintenance of their local social 9rder; they make
it seemed likely that often the point was not to catch the flecing fans at up a functional rhetoric (Harré, 1985a) in contrast t~ the fans' more
a11, because doing so might result in an actual fight. Rather, chasing was descriptive vocabulary, which stresses arder. ;
more an expressive than practica! activity (Harré, 1979)- it demonstrated For Marsh et al., then, the fans' accounts have a crJcial place in the
willingness to fight without the damaging consequences which might result analysis of their activities in line with ethogenic res~arch principies.
from actua11y fighting. Through these accounts the researchcrs are nblc to rcVeal thc structurc
Although Marsh et al. could deduce this orderly rule-guided view of of rules and thus the social competence which regulates a~d givcs meaning
events from the fans' accounts, these accounts also at times conflrmed to the fans' action. These researchers concur with discburse analysts in
the opposing newspaper view of soccer hooliganism, namely that a great finding it analytically fruitful to pay clase attention to the variabilitv in
deal of actual violence does take place, many people do get hurt, and soccer participants' discourse or the divergencies in thc accoJnts givcn by. thc
matches are inherently dangerous places which sensible people should fans. Yet despite the shared focus on people's languagc u~e, and emphasis
avoid. Sometí mes these two sorts of accounts appeared closely intertwined on variability, there are important differences between thiS ethogcnic study
in the fans' accounts. and the perspective a discourse analyst would adopt for d~ta ofthese kinds.
In particular, there are damaging inconsistencies in the ~vay discourse is
Questioner. What do you do when you pul thc boot in? interpreted in this study, as we shall now attempt tn d:emonstrate.
Fan A. You kicks 'cm in thc hcad don't you?.
Strong boots with metal toc-caps on and tbat. 1

Questioner. And what happens thcn? Problems with ethogenics


[Quizzicat \ook]
Qucstioncr. Wcll what happcm; to thc guy you'vc kickcd? i

Fan A. He's dead. One arca in which discoursc analysts and ethogcnic rJscarchcrs are in
Fan B. Nah- he's all right- usually anyway. accord is in emphasizing the range of purposes lo w~ích language is
(t 978: 83) put. It is used for explaining, justifying, cajolingJ persu?cting; to present -p1:U,_ lo c.,~ a~
the speaker as worthy and competent or to accusc another. As we noted
In this extract from Marsh et al. 's data Fan A fortnulates a picture ofviolent in Chapter Two, the problem for any analyst who is concerned with
1

and dangerous struggle, while Fan B interrupts with a much more placid people's accounts is that they do not come ready labelled with the goals
picture. The researchers, however, treat these two versions in fans' they are designed to achieve. Quite the reverse, unfohunately. Often
accounts very differently. They regard the placid, rule-bound picture as accounts are constructed speciflcally to keep their aims h\dden or, equally
a more or less literal version of what goes on, as indicating the social often, the consequences of sorne accounts may not be irfimediately clear
competence, while they maintain that the violent, disorderly version to participants. The kind of literal and explicit speech a~ts which Austin
presents a very inaccurate representation of footba11 supporting. The focused on- '1 hereby accuse you of lying'- are very much the exception
fans know that it is not really that dangerous; indeed, the authors suggest rather than the norm. This does not necessarily mean that listeners who
that they are simply reproducing in part of their talk the dramatic and are part of an interaction have difficulty in detecting the u~shot of accounts,
exaggerated story telling of the mass media evident in headlines such as but their embedded, inexplicit nature presents the anPiyst with great
'Smash these thugs', 'Ciobber boys on the rampage'. problems. 1

At this point Marsh et al. make a very subtle and intcresting analytic The soccer fan researchers treat different parts of th~ fans' discourse
suggestion. They claim ihat it is the combination of the two kinds of in very different ways. For them, sorne parts reveal ~hat is actua/ly
accounts - one stressing disorder and violence, one stressing arder and happening on the football terraces, while others are broadl* funclional: they
62 Discoursei and social psychology Making rules work 63

J
serve to create frisson ofcxcitement and danger. Now, the authors claim triangul.ation. With abservation, the accuracy of accounts is checked by
that they are n?t trying to determine the truth of accounts, for the~e is campanng them to the researchers' observatians; the correct accounts are
no one true literal account of an event; any event can be charactenzed those concurring with these observations. In psychology, ecological
in ~umc'rous difrcrent ways. Neverthcless, they treat the accounts of arder researchers have placed the most stress on this observational method
as indications o~ thc gemdne structure of the fans' activities (although they (Wicker, 1979). Triangulation suggests that a variety ofdifferent accounts
do not take thetn as directly truthful). This portian of the fans' discourse from different sources should be collected to allow the researcher to
is used to reve1al a social reality beyond itself. Marsh et al. state: determine the correct version of events. This approach has been popular
! in areas of micro-sociology (Bogdan and Taylor, 1975; Denzin, 1978).
[Thc] appare~tly disordercd evcnts on the footbnll terraccs [ 1can be scen as The first thing to note about observation is that if it were sufficient to
conforming tb u vcry distinct and orderly system of roles, rules nnd shared simply observe the actions and events taking place there would be no need
meanings. A~tion is ncithcr chnotic nor scnseless but rather is structured and for ~h~ extr~m~ly complicated and time-consuming process of collecting
reasoned. (1978: 97) partlctpants d1scourse. Psychologists could just watch what went on and
!
ha ve do?e wi.th it. However, this forgets the point made earlier concerning
In contrast, thelathcr set of accounts, which emphasizes senseless violence, the semtolog¡cal character of action: the connection bctween movements
is treated as a fhetoric which is more important for what it does than what and particular actions depends on knowledge of shared social conventions.
it says. These! are conspimtorial accounts, derived largely from press This idea is fundamental to ethogenic theory, and thus observation is not
reports, whicl' the fans use to retain an exciting sense of danger and as an approach open to Harré and his fellow researchers.
a resource for ,~alorous display. The authors contrast what the fans really
The second point to note about observation is the assumption that actions
'know' about ~vhat goes· on with their rhetorical 'construction' of events.
! have one, intrinsic, unitary meaning. Yet even ethologists, who have
based a whole science of animal behaviour on observational research of
Sincc fnns ·~now' that this is not thc case - thcy are aware and can tcll you
v~rious ki?ds, ~ate that .visu~lly identical behaviours may be serving quite
that fcw pcorltc gct hurt cvcn wbcn things 'gel out of hand'- thcy must conspi~c
to construct disordcr. And bccausc there is an eusy rhetoric al hand- lhe rhctonc dtffercnt lunctmns, whde vtsually dissimilar behaviours muy be serving
of the media ¡which insists that cvents at footba\1 matches are infact disordered,
1 the same functions (Hinde, 1975). When we are dealing with humans
the conspiraCy is un casy onc to conduct. (1978: 97) their linguistic and interpretative abilities make things many times mor~
1
complicatcd. The same movements may be interpreted as performing
The crucial pfoblem, however, is that Marsh et al. offer no method or different actions by different people, or the nature of the movements
criteria for m~king this division iota genuine and rhetorical accounts, n~r may be reassessed and reinterpreted over time (Gergen, 1982; Schutz,
do they cond~ct an analysis which shows that ~ne set of accaunts 1s 1972). An event may be a demon~tration to sorne people and a riot
important for !its acule description of the fans' socml warld ~nd .th.e .oth~r to others; a person may initially view a comment as a joke but later
far the largcl)' imaginary sense af danger it produces. Thetr dtvtsmn !S come to view it as an insult. These phenomena are a pervasive and
plausible- bu~ then sois the alternativc. We could easi.ly i":aginc t~e f~ns important part of social life - they cannot be circumvented by observational
drawing on th:e vacabulary of rule use and arder to ratmna~tze and JUSttfy research.
their part in what is in reality a violent, senseless and potentmlly dangerous Triangulation or the use of 'multiple partial indicators' (Denzin, 1978)
activity! The ipoint is not that this alternative is any more likely to be has been suggested as the other way of defusing the problem of variability
correct, it is !that Marsh et al. ha ve not provided a principled way of m accounts. Indeed, Harré himself suggests something like this when
deciding bet\yeen them. describing the soccer research (Harré et al., 1985: 87-8). The term
triangulation is, of course, derived from navigation. In the B movie,
the exact position of the damaged bomber limping back from Germany
1

Obsen,ation fmd tricmgulatioll is always found by triangulation. The direction of its radio signal is
assessed by two geographically separated receivers, allowing an exact fix
At this point !traditional researchcrs generally offcr one of.two sol~tions to be made on its position. In the same way, it is thought that the researcher
for overcomi;ng the problem of deciding which of the vanous avatlable can use discourse collected from different sources to 'home in' on the
accounts shotild be takcn literally. They suggest either observation or facts of the matter and thus show up sorne accounts as dislorted or
. '
rhetorical.
¡

64 Discourse and social psychology Maki11g ~ules work 65


1
1'tJc,!:l.ioo This is an a_¡¿¡llilling solution. However, as we shall see in this and luter possible to obtain records of naturally occurring inter~ctions such as
chapters, when discursive data from a number of sources are examin~d conferences and workshops. By collecting material ip this way the
in fine detail the information does not allow the researcher to home m researcher is able to amass what is basically a slicc th~ough the social
on the correct version of events. The detailed studies of discourse reviewed life of one discrete real m of human activity. '
in this book overwhelmingly indicate that the ciase study of documents With respect lo the theme ofthis chapter, science is a place where rules
and accounts from different sources and settings compounds rather than abound. At one leve!, there are the more or less cxplici~ practica! rules
reduces the variabiliry between participants' claims and descriptions. and guidelines for experimentation, analysis of data and wfiting up results.
Indeed, in the course of attempting to triangulate the result is more afien At another leve!, the philosophical and sociologicallitera:rurc concerning
than not 'homing out', resulting in the proliferation of more and more the operation of science often suggests that what disting~ishes scientific
inconsistent versions (Gilbert and Mulkay, 1984). discussions are their adhercnce to generalized norms; For cxamplc,
So we are forced back to the fundamental quesrion. If observation and there is a norm or rule that scicntists should assess all kriowledge claims
triangulation are not the solution, how should we respond to t~e variation sceptically and independently of the status of the claimer !,(Merton, 1973;
between accounts? As we indicated in Chapter Two, the solutlon adopted Zuckcrman, 1977). Therc are in addition particular rules lar thc sciection
by discourse analysts is not to try and resolve the variation between of theories like always select the simpler, or more !estable of a pair of
accounts but to make that variation a way into analysis. In the ca·se of theories (Kuhn, 1977; Popper, 1963). Justas Marsh et ai. claim that the
rules, our approach is to look al how participants' rule accounts are distinctive culture of socccr fans is managed and reproctuScct by following
constructed and organized to try and reveal what is achieved by different a limited set of rules, so Popper, for example, suggests [he same is true
accounts. We do not ha ve access to Marsh et al. 's data, and so cannot for science. Its distinctive culture is produced through co~formity to rules
undertake this cxercise on the discourse of soccer fans. Discourse analyses such as formulate testable theories and try to refute them (Mulkay and
of rules, however, ha ve been conducted on the language of scientists, and Gilberl, 1981). !

it is to these that we now turn. Discourse analysts ha ve conducted a number of studies on rules and
theory choice in the natural sciences (Mulkay, 1979n; Mul~ay and Gilbert,
1981, 1985) but to make things more intcrcsting wc will! conccntrate on
Discourse analysis and the study of scientists' rules a study of rules in the discoursc of researchers who afe much closcr
to home, namely psychologists (Potter, 1984). The dat~ for this study
Why science? consisted of the transcripts of discussions ata majar fivc-day international
psychology conference concerned with fundamental theo~etical issues. It
It might seem odd that discourse analysts ha ve taken scientific language was a particular!y fertile sctting for the study of rules bec~usc, at confcr-
as one of their central tapie areas (Brannigan, 1981; Bazerman, 1984; ences, scientists are continually describing and justifying their
'
acts of theory
Gilbert and Mulkay, 1982; Myers, 1985). You might think that science choice undcr thc critica! scrutiny of experts within their bwn field, who
is an abstract, technical and precise realm, and as such can tell us little may respond by supporting, modifying or chnllenging the propriety of
ofvalue about the social and psychological processes operating in language. choices. Here scientists try to negotiate a more or less ~table scientific
Yet it is exactly these features which make scientists' discourse such reality, and it is cxactly in this kind of situation that wc oJght to see them
an attractive research tapie. If interesting discourse processes can be identifying the rules which, supposedly, guide and sustain iheir theoretical
found even in this rarefied environment it is extreme1y likely that they commitments. ·
will also be found, probably in more exaggerated forms, in everyday We will concentrate here on the rule or criterion for theohr choice which
conversation, newspaper ~;tories and in al! the other kinds of tal k we will specifies testability. Roughly stated, this is the rule that sbientists should
encounter. Science is a useful hard case where discourse analysis can hone always adopt theories which are open to empírica! test a~d, conversely,
its claims. they should never select theories which are untestable. ~s well as being
Therc are other good reasons for studying scicntists' discourse. For one a rule dccmcd fundamental by many philosophers and !sociologist.s of
thing, a large proPortíon of scientists' discussion and writing is encoded scicnce (Crane, 1980; Popper, 1963; Quine and Ullian, 1970), references
in highly cxplicit forms and is readily available for research. It is fairly to it appeared rcpeatedly in thc data transcript.
easy to collcct scicntists' discourse without disturbing their social fabric;
journals and books are full of publicly avuilable accounts and it is o !'ten
66 Discourse land social psyclwlogy Making rules work 67
!
!
rules. Just as sorne of the footbaJI fans' accounts suggested different
Versions of the lrestability rule
versions ofthe way rules guide activity on the terraces, so it is with these
!
The first step ii\ the analysis of the conference data was to select frorn scientists' accounts. In Extract One the testability rule acts as a potent
the transcript ad references, however obligue, to testability or the re!ated constraint ori the choices of scientists; yet, in Extract Two, the very
notions of refuthbility and falsifiability. It becarne irnrnediately apparent possibility of applying testability as a potent, c!ear-cut rule of choice is
from this exercise that there was no consensus in this scientific grouping denied because of the ever-present option to reinterpret the nature of
with respect td either the nature or the importance of testability. In potentially falsifying data. In the case of the soccer fans, of course, Marsh
fact, the confetence participants provide a whole range of different et al. found that not only would different fans draw upon two inconsistent
versions. For ~xample, in the following extract we see a psycholo~ist versions of rule operation but the same fan would describe rules variably
stressing the crucial irnportance of testability. (In aJI the extracts wh1ch in his or her talk. As we examine further extracts we will see that this
follow pseudo~yms have been adopted to protect the identity of those is also true of our psychologists. There was not only variation between
involved.) !
the psychologists' accounts of testability, there was also variation within
psychologists' accounts.
'
1. fA] meanl of tcsting your explanation- l. J of course that is thc absolute, The iJiustration of variation between accounts is analyticaJiy useful
cardinal rJature of scicnlific work, scicntific explanation. And thut's not because it reminds us of the dangers of taking accounts at facc value as
just pointirg to somcthing that you say is important.. lt has thc ~o\low~ng accurate descriptions. Ifwe had examined either one ofthese extmcts alone,
critica! value: it allows us to make progress; we can d1scard theones wh1ch we might ha ve taken itas a description of the role played by the criterion
have prov~d usc\css. (Norton: 3) of testability in the scicntific spccialty we know as psychology. However,
This is a very s~rong
account of the role of testability. The speaker claims
when we examine such contradictory extracts together the foil y of this
approach beco mes clear. Testability cannot be both a clear-cut, potent
that a theory which cannot be tested is simply not scientific; for testability
critcrion and a rule which can be interpreted to fit any set of data.
is not mcrcly ih1portant but esscntial. Progrcss happcns through tcsting
Ncvcrthclcss, thc idcntification of thcsc sorts of variations is only a
explanations ahd rejecting those which fail these tests.
preliminary stage in discourse analysis. On its own, it provides no way
The next acdount ís very different. The possibility of any clear test of
of inferring thc sorts of functions perforrned by different passages of
thcoríes is qudstioncd bccausc or thc way data can be reintcrpreted.
discoursc. To gel lo a !'unclionallevcl of analysis mueh more attention must
be paid to the patterning or organization of different versions and the way
2. \Ve can nJgotiate the meaning of any particular observt~tion in virtually t~ny
direction.i So that, for examp\c, we tal k about aggression, and yct it's not they are constructed. To achieve this end we Jookcd in this study at the
clcar wheh wc e ver ha vean instancc of aggression. [ ] Virually any activity, varying accounts given of tcstability when speakers are depicting the
1 suspeet,l that 1 cngage in nny duy, could be looked <Jt as t~n instance of criterion with respect to their own work and then work they disagree with.
t~ggressioft in somc form. [ ] it depends on the set ~f social agreements
of what things are called. And those agreements, 1t seems .to me, are
ncgotit~b\C ovcr time, t~nd thcy can be dist~greecl upon and wnthcd round
in t~ny ca~e. So that, in effect, any theory can be sustained so long as you 17ie organi:z.ation of variation
have <1 cdpab\e negotiator of reality. (Leary: 278-9)
! The following extract illustrates a phenornenon which appeared repeatedly
In this view, theories can ncver be clearly tested against observations
1
in thc conference transcript we analyzed. H is interesting and illuminating
beca use the m~aning of any observation is dependent on conventions which because the speaker characterizes testability very differently at different
can be interprbted in different ways. The scope for tlexibility is so great stages in his presentation. Because rather more detaíled interpretations
that a 'capablih negotiator of reality' can indefinitely shape observatmns than we ha ve rnade up to now are required, we will number the sentences
into Jine with 1theories. Testability is thus virtuaJiy irrelevant.toscientific so that the separa te parts of the extract can be referred to accurately. Thus
..
progress, bec;ause what is seen as a test will depend entirely on the (3-4) is a rcference to sorne features of sentences 3 and 4 of the extract.
ncgotiated carvcntions for interpreting observations. . .
Thesc two i1lustrntive cxtracts are comparable to the dtfferent kmds of 3. (1) 1 woulcl like to return to this gamc of 1 am more refutable than you.
uccounts give¿ by Marsh et ni. 's socccr fans. They dcmonstrate that within (2) I said refiaab/e, not reputable, although you can be forgiven for
a given social! group there is no conscnsus about the constrainíng role of rnisundcrstunding that. (3) [lt~ughter] [ ] (4) U m, thcy al!, cvcryone pretty
68 Discourse and social psyclwlogy .
Makim: lmle.\· work
'
69

wcll has paid lip scrvicc lo rcfutution. (5) It is inlcrcsling to noticc thc strut, and pcr~onal intcrcsls beco me irrclcvant. ln thLs wuy, thc ~1ccount providcs
the dífferent strategíes that people, er, havc used. (6) U m, síncc Norton strong legitimation for his scicntilic claims. 1

spoke such a lot it is a pity he is not here. (7) lt's, I would like to draw On the othcr hand, Young also dcpicts the rule of tcstqbility as open to
attention to his, which is very clearly to say: [chungcs tone of voice] '!cave
Norton's strategic manipulation. Norton is said to inconsi~tently invoke the
me a\one, 1 am doing very well in my own srnall corner' - you know -
'go und die in your own'. (8) [luughter] (9) Although he suys thut evcry- rule to support his own theory at the expense of others, pioducing theories
body has a duty to [create large-scale thcories] he is not trying to produce, so sma\1 in scale that they are not likely to be overthrownlby thc cvidence.
um, Jarger, more umbitious models which are, of course, much easier to, The nature ofthe rule is thus seen very differcntly in this case; testability is
er, refute, er, than thc Jess ambig, ambítious ones. [ ] no longer a clear-cut and impersonal constraint on the sciertist's
'
actions but
(lO) The good guys are the ones who are Jeavíng hostages to fonune, who part of a rhetorical display. By depicting Norton 's theorptical claims and
really are laying themsclves open to being knocked down in all possiblc actions in this fashion, the account undermines them. i
ways. (11) The good guys are thc pcople who are putting up thc bíggest As with earlier extracts, if we examined this passagej of Young's tal k
possible theory that they can imagine und, um, hoping that, um, ít won't in isolation we might well interpret itas an unproblema~ic description of
be knockcd down too quíckly. [ ] (12) Among these people I would pul the way the testability rule has been applied to two diffcre9t theories. When
Squire, um, Chester - 1 apologize if 1 huve misscd anybody clsc out - 1
we compare this account to others, however, the inconsi~tencies are such
can't remember all the things that have happened in the, in this meeting.
(13) But 1 can, of course, remember what I said, and I am one ofthe good that one cannot conclude that Young is simply correcqy describing the
guys. (14) [laughter] (Young: 270-2). difference between his own and Norton 's theories. For exdmple, in the next
cxtract we will examine, Norton's theory is prcsented aS' unambiguously
In this extract, the same speaker, Young, draws upon two rather different open to testing and Squire's as resistant tC? this testing -lthe very reverse
versions of theory choice, resembling the different versions documented of Extract Three! '
in Extracts One and Two above. In the first version- in, roughly, !-ientences But befare going on to discuss thc implications of thik next !-itep in thc
1-9- thc criterion of testability is said to be used strategicully (5); it is functional analysis of rule discourse it is worth bríetly pointing out onc
1

depicted as being inconsistently invoked in support of Norton's theory of thc proccdurcs that discourse analysts ha ve used to cbnfirm claims of
at thc expense of othcrs (7-9). In contnJ~t. in rclution to thc ~pcakcrs' thi!i kind. (A more dctailcd discu!i~ion can be found inl Chaptcr Eight).
own theory, the criterion is depicted as an effectivc constraint an action. This procedurc is bcst dcscribcd as conlirmation thro4gh cxccption.
Refutation is removed from the speaker's sphere of social control; al! he '
can do is hope that his theory will not be undcrmined by thc cvidence
- it is a 'hostage to fortune' which will be rejectcd or retaincd by virtue Conjirmation rlzrrmgll cxception
ofthe data alone (12-13). Unlike Norton, he is unable to influence whether
or not his theory will be found wanting. In our analysis ofthe rule oftestability we seem to ha ve (·ound a regulurly
It is worth pointing out that this kind of combinution of strong criticism occurring pattern of accounting. The testability rule is ;not used equally
and self-praise is trick-y to manage- it can easily be hcard as persecution wid1 al! thcories, one's own theory tends to be described 9s unproblematic-
and boasting (McKinlay and Potter, 1987; Pomerantz 1978; 1984). In this ally testable while other scientists are portrayed as strategically attempting
extract we see Young ski1fully using humour to dcflect any potentially to avoid this rule. We hypothesized that this regularity tan be explained
negative attributions of this kind (Mulkay and Gilbert, !982a). in terms of the functioning of the accounts for the criticis~ of others versus
So, what is the significance of these two different versions? What is self-justification. However, what of exceptions to the rdgular pattern? If
Young doing with these two ways of formulating the rule? To address the proposed functional hypothesis is correct it ought t~o make sense of
this question we can loo k at the way Young presents his own and Norton 's both thc pattern regular! y found in the data and the exc'eptions. That is,
agency. On the one hand, the rule of testability is depicted as operating the analyst must predict that there will be special featurd of the organiza-
on Young •s own theory quite independenrly of his agency. In effect, he tion of the exceptions which allow them to fulfil the req~ired
'
function in
stands back and waits to see if the data overthrow the theory which is sorne other way. If these predictions do not work out,! of course, then
a hostage to fortune. The rule of testability is applied impersonal\ y and the analyst will need to start reassessing the adequacy of ~eir explanation.
allaws clear-cut decisions to be made. Young thus represents his own In the study ofthe testability rule in psychologists' conference discourse
position, albeit humorously, us fully conforming to the canons of empiricul wc found jusi onc cxception to thc pattcrn of own the~ry dcscribcd as
science where the data are the most crucial consideration in theory choice testable, other theory described as resisting testing. Tl~e exception can
!
i
70 Discourse Qnd social psychology Making rules wark 71
!

be found in Extr~ct Four. Unusually, in this passage, thc speaker fashions theories ought to be refutable is thereby made equivalent to the extremely
an account whicr depicts his own thcory as actually less testablc than implausible claim that certain classic and profound theorics are not properly
ccrtain otl1cr theo:ries. Although it is a rather long extract. it repays detailed scientific. And by characterizing his own theory as a member of this class
examination as it is a superb rhetorical construction. Squirc equates it with theories which are usually seen as the very pinnacle
of scicntific achievemcnt and, at the sume time, indicates that its Jack of
i
4, Could Ijustisay a quick word about fulsilication, because lthink it is really refutability is an asset rather than a fault.
u rather impbrtant tapie. And lthink part ofthc troubles that arise in trying When we look at the detailed structure ofthis exceptional extract, then, ,C:¡?(~
to gel u haridlc on it, as a pmcticnl working scicntist, is thut it is quite we can see it is performing the sume justificatory function as the previous¿ -·er-
di!Tcrcnt, it ims a quite diffcrcnt charactcr dcpcnding upon thc typc of theol)' one. The speaker is depicting his own theory as the better, more powerful,
íl that you ar~ operating with. If you are operating with a theory which has more scientific one. It inverts the idea that better theories should be more
·_ 1 a wcll-established dcductive structure then the, er, the refutatory process
testable, but warrants this inversion by offering 'lessons' from the history
is, rclativel~ spcaking, logical [ ]. Now that's okay, and that's l think the
structure o~ what we have bcen calling mini-thcories, i.e. theories which of science. The very fact that the speaker feels it necessary to provide
are sufticicntly small-scalc to be articulatable in that deductive form. [ ] this detailed warrant shows that he is orientating to the expectation that
But I think lhe case Lhat's much more intercsting is the sort of thcorizing better theories should be more testable. Thus a e! ose examination of this
at thc leveliof which I was talking of this morning: molecular theory of exceptional passage provides further confirmation of the operation of the
gases; cvolutionary theory; [Squire's own theory]; those kind of theories more standard pattern of discourse.
which contÓin thc [specified] structurc [ ]. And it may be that a really Squire's difficulty in sustaining this non-standard version of testability
powerful théory contains within itself enough, as it werc, potential material,
as indced th:e molecular theory of gases did, to go on with, through a great
can be seen when he is faced with a theory which he feels has been
den! of truditional refutatory proccdures and still survive as the theory in undermined by empirical findings. Here his non-standard view would make
the field. A!nct Darwinian theory has also bcen through the same kind of it hard to criticize the theory and so Squire reverts to the more familiar
gamc. Nm~ thc theory 1 was trying to outline this morning has just that pattern.
character; t)ull is, it is a theory for providing a conceptual systcm. But it
does havc -1 a\though Young, of course, sees itas having pseudopodia which 5. The survival by n lhousand qualificntions ah, is, er is what goes on. And
are puta falr wt~y into the, into the refutable world- at lcast 1 would also ccrtainly [the specified] thcory is a beautiful exnmple of that. The more
claim for it that it has a certain mensure of clasticity and thut iris not goi11g
1

thc thcory got rcfuted the more distinctions that wcre madc. Now. of course,
' .
to be too dam el/.\'\' ro rcfutt!. (SqUJrc: 273-4)
! •
that is u tcchniquc which you can use. But it has, ns it werc, nothing to
do with thc logic of thc case. And one would then query thc motives of
The first thing tolnote about this passage is that it is an almost total inversion the people engaged. (Squire: 276)
ofthe previous extract we considercd. Like Extract Three, it cmploys two
very different idmges of the way the tcstability rule opcrates: as a clear- When Squire 's own theory flexibly avoids refutation that is a good thing;
cut, Jogical proqess which effectively constrains theory choice, and as a it is only doing what great theories ha ve done in the past. When the same
flexible, negoti<ible process where no data on their own can force choice. procedure is followed with a theory Squire opposes, he suggests that this
Howcvcr, thc atiribution ofthc two imagcs is rcversed. Opponents' thcory is notan acceptable, rational process, but a situation where motives should
choices are depicted as resulting from the impersonal, data-drivcn rule, be questioned.
whilc the spea~er characterizes his own choice as negotiable and less
depcndcnt on eXperimental findings: in effect, \ess tcstable.
When we spe)l it out in this bald way, it is hard to reconcile this with Testability, soccer violence and the analysis of rules
our suggestion that these scientists are depicting their own theories as
Wc ha ve now indicated how a discaurse analyst might approach the tapie
testable to justil)j them. Perhaps our analysis is confused or, worse. perhaps
of rule use. The analyst would start by examining the manifold ways rules
it is wrong! A ciPse examination of the passage suggests another intei]Jret?,-
are formulated in participants' discourse and then go on to loo k at the
tion. We can s~e that Squire fashions a wide-ranging historien! i1ccbunt
of the nature of theories in science lo display a lack of testability as a organization of these differcnt accounts. With respect to the rule that
theories should be testable we found an important asymmetry between
positive fcature br ccrtain theories. He does this by characterizing classic
thcories from tlie natural scicnces, such as the molecular theory of gases accounts of the rule with respect to the speakcrs' own theories and those
of their opponents and critics. This asymmetry was understood in terms of
and evolutionar~ thcory, as not bcing clearly refutable. Thc chargc that all
A1akinc~ mies work
'
72 Discourse and social psyclwlogy 73 . , ·
i
i • J'.<;_~'
. ~o('- /L" !·N.~•·
¡
le Cz ~
the functions served by the different accounts in buttressing speakers' own oftheir actíons as governed by the code to achicvc a wh:ole runge ofgoals
positions and making opponents' positions problematic. In addition, thesc (Heritage, 1984). The scientists tlexibly drew on the notion oftestability 6c c,. v[Oi.¡ Ji
,¡., '''C'"
findings ha ve important implications for thc philosophy and sociology or to construct a coherent and convincing version of the \Ya y thcir own and ""'"¡·,1 _/ko
(!te;' {,t.... / · " -
science, but that is another story (Potter, 1984). others' theories were chosen. The degree of flexibi'lity possible was / C., 1 ¡__ f./ '··:+
graphically shown in Extract Four where, in an imp:ressive rhetorical lli:OJ'~ ,. 1.../ o<:¡.¡ ~f;u:,z.
0 11 1
Although Marsh et al. reached the stage of showing variability in .~ ~ '>" 1
accounts of the rule-bound nature of life on the terraces, they did not go construction, Squire inverted the evaluation af test~bility which was Pr?Jt(rc-J..r:. ..f · ti._,;,.( V íli~t")fJ.-.·
on to loo k at the organization of these different accounts. They speculated standard elsewhere. In each of thesc studies, then, rules *re se en to opera te p1)C,¿.4 ??
about functions but did not attempt to reveal them through systematic notas templates or mechanical guides, as part of a 'co~petence', but as ,
analysis. Instead they introduced the idea that sorne accounts are simply symbolic resources used to perform specilic tasks. (- f<¡ Í' J:{c_( ¡,¿,_ ,,bs-/<o/, <"- ¡ 'd; tr v
rhetoric for generating excitement in arder to resolve thc difficulty of The general problem for researchers such as Harré, wro wish to cxplain ,1vr- /.!t['h! 'k ~f &~~~ ;,~...-<'1-.
providing a coherent explanation of hooliganism at soccer matches when behaviour through the control of rules, is that there ám be no neutral,
faced with the variable and contradictory naturc of fans' accounts. Their disinterested statemcnt ofrules and how they are applied. Harré sees rules
resolution works by dismissing one subset of accounts as fantastical whilc as part of a person 's social compctence, but this comp~tence can only be
treating another as providing genuine information about what the fans revealed through a study of performance data, namelyj tul k. And, givcn
actually do. the reflexive and contextualized nature of rule talk shown¡ in Wieder's work
The variation which discourse analysts ha ve found in rule accounts of and analyses ofscientists' discourse, any extrapolations from perfonnancc
scientific theory choice is very similar to that found in soccer fans' to competence require extreme caution. The danger is that the researcher
discourse. Sorne accounts stress the constraining force of rules and others will fall into their subjects' trap and adopt the factual yiew of the world
claim the rules ha ve little practica! effect. However, instead of assuming they alTer. To prevenl this, we need the much more comprehensive .study
that one of thcse vcrsions is correct and attcmpting to choose between of rule tal k providcd by discoursc ¡tnalysis.
them, discourse analysts ha ve tried to demonstrute that both serve important
functions and thus explain them in this manner. Of course, without a full
study of football fans' tal k we cannot say that this will necessarily be true
of the soccer discourse, but it is at the very least a plausible altcrnativc
to the explanation provided.
The findings from these studies of scientists' rule use are quite consistent
with Wieder's rcsearch on the convict codc dcscribcd in Chapter One.
Both pieces of research emphusize the reflexive dimension of rule talk:
talk is not just about actions and events but it is also a potent working
part of these things. The realistic model which would treat this tal k as
basically a descriptive medium is flawed. It would be impossible to
integrate the highly divergent versions we saw in Extracts One and Two
into a single set of facts, or to sensibly accommodate the many subtle
differences between Extracts Three and Four. lt is only. when we start
to look at the functions to which the talk is put that we can begin to fully ,fvl'<, 1.-i<-¡llu-, ¡.) ¿;], eL;, ¡.t( ,.{¿;,;
understand what is happening in social life. The scientists are using the
rule of testability justas the inmates are using 'the Code' - to persuade,
justify and criticize.
0 (D re&
Both studies also indicate how participants draw upon rules in an
occasioned fashion. That is, the meaning of the rule is expressed in a way
appropriate to the speaker's context. lt is not the case that rules somehow
carry implicit instructions for their precise and proper application; the way
they are applied is, in fact, as much depcndent on people's constructivc
use of languagc. Thc inmutes in Wieder's study creatively dcpictcd ccrtain

1 IV 01Ülr Vl~..!A'' C~<¡¡ c,;t'lt,.c'r 1 v.Ld!'h-•17)+


1
Accowus in sequence 75

of the normal. As Austin put it, excuses 'help us penetrate the blinding
veil of case and obviousness that hides the mechanisms of the natural
4 successful act' (1961). Sorne light might be thrown on the way social arder
1 is produced in a society. The study of fractures and their repair should
Accounts m sequence J'r'>,vpviJ(hc( illuminate how social frarneworks emerge.
1 / - (--1\JI;Lf!..Q'\.v./
i•
The second reason for studying excuses is their significance for under-
standing the causes of action. Austin argued, for example, that a majar
So far in this book we ha ve used the notion of accounts to refer generally breakthrough might be possible with the perennial philosophical problem
to any passage dr talk or writing. However, therc is a more restricted of free will if it were refocused on the kinds of things which people treat
technical sense 6f this term marking out the discourse produced when as interfering with their actions. He suggested that rather than discuss free
people are explalning actions which are unusual, bizarre or in sorne way will in abstract terms we should look at the procedures people use when
reprehensible (S9ott and Lyman, 1968). lf you a_re caught doing son;ething accormting for freedom and constraint in those cases where breakdowns
which appears odd - perhaps someone walks mto your room while you occur and excuses or justifications are required. We are all familiar with
are wearing yoJr underwear on your head - you will want to accormt the apocryphal defendant in the dock who explains their crime by saying
for your behavio~r; you might, for examplc, explai.n that you ~re cxploring '1 am justa victim of a broken horne, Your Honour'. This defendant is
possibilities for Saturday's fancy dress ball. lf th1s account IS successful providing an excuse or an account but his statement could also be taken
your behaviour fVi!l seem rather less odd. as evidence for the causes of his behaviour. Austin advocated a systematic
Thesc kinds df accounts are intrinsically interesting because they are conceptual investigation of tal k of this kind, rathcr than naive acceptance
such a habitual feature af languagc use; they are notable, too, because of its import.
social psycholo~ists have recently paid sorne attention to this tapie arca Both of these points are mirrored in the way ethnomethodologists ha ve
(Semin and Ma'pstead, 1983; Snyder et al., 1983). We thus have an organized their research. Garfinkel (1967), notoriously, sent his students
opportunity to cValuatc the achicvements of currcnt cxpenmcntal mcthods out on cxercises which involved disturbing the social fabric in an attempt
tfo ufindedrsta~d i~ bette ~.kFor inhstance(, ht~ey wedrebasked t? ?arter_in shops ?~-~~~-~_ ;.~.·,
1 11
and social psychological theories when they are turned on a central area
of language funttion. These achievements can be compared with those or IXe pnce Items 1 ·e toot paste t lS COU 1 e SUrpriSillg1y effective) ve. ,
cmerging from the discipline of conversation analysis which takes a more or to imagine that they were strangers in their own home. The idea is
naturalistic oriedtation to the study of discourse and accounts. This chapter not unlike the goals of high energy physicists when they shoot protons
will hence servd the dual function of describing sorne rnainstream social at matter and study the tracks ofthe breakdown particles. Or imagine taking
psychologica\ re~earch while introducing an increasingly important analytic different parts out of a car enginc - the air filter, say, or the horn - to
perspective on ~anguage. see the cffect on its running. In addition, ethnomethodologists have
commonly used the strategy of taking issues typically construed from a
i
philosophical perspective and looking at their importance for participants
Accounts and social psychology themselves. For them also, the problem of free will should be translated
¡
into the question of how people depict constraints on their actions when
Austin 's work i~ undoubtedly the best place to begin any examination of giving accounts and excuses, how they formulªte freedom or constraint,
accounts. We s~ould particular! y note his influential article, 'A Plea for and what these formulations achieve. /J•.o--

Excuses', whic~ appeared in 1961, as this paper acted as a basic starting Austin 's paper contained a wealth of interesting ideas about the language
point for many'¡of our contemporary ideas about this subject. of excuses. However, his most important legacy to social psychology was
the clear distinction he made between two main species of accounts:
1

justifications and excuses. Austin observed that when someone is accused


Austi11 's plea ¡Jr excuses of acting in a bad, untoward or shocking way, these two alternative
strategies are open to thc person: they can provide either a justification
Austin claimed that there are two important reasons why we should study or give an excuse.
excuses. First, hs excuses occur in situations where there has been sorne Excuses are accounts which admit the relevant act was bad in sorne way,
fai!ure or breakbown in normal conduct, their study may clarify the nature but claim performance was influenced or Caused by sorne externa! agency.
!
!

76 Discourse and social psycho/ogy Accmmts 111 sequence 77


!

Thus you might excuse late arrival ata meeting by noting that the alarm dock Disclaimers are the third new category. As we notcd ~in
Chapter Two,
failed. Justifications, on the other hand, do not involve the denial of responsi- these are pre-accounts which attempt to ward off antlcipated negative
bility, instead they claim certain actions are in fact good, sensible or at least attributions in advance of an actor statement. To recapi(uJate, the typical
permissible in the circumstance. A person might justify stealing by the sug- form for a disclaimer might be: '! am no sexist but .. !. ' foJlowed by a
gestion that a JI property is theft and il is a good thing to redistribute wealth. negative oomment about women. People use disclaimets when they are
about todo or say something which is Jikely to be interpreted as coming
from someone with a particular identity, in this ca~e a sexist. The
Accounts and typologies disclaimer attempts to prevent the listener interpreting !the talk in terms
of this noxious identity by acknowledging the possible ihterpretation and
Austin's basic distinction betweenjustifications and excuses was later taken then denying it (Hewitt and Stokes, 1975). ¡

up by two sociologists, Marvin Scott and Stanford Lyman, in their 1968 Semin and Manstead .integrated these various genera¡ of accounts into
paper on accounts. Scott and Lyman made a number of interesting although a sort of super-typology mtended to cover all possible excuses and justifica-
often rather speculative points, but their main goal was to produce a tion. A shortened version of the typology is reproducdd in Figure 4.1.
typology of accounts. Hence they idenlified different spccies of excuses This is an extremely useful exercise, but, as they ~hemselves note,
such as the appeal lo accident ('! tripped up'), to mental elements ('! taxonomies need to be supplemented by rescarch on the use and effective-
forgot'), to natural drives ('! was so hungry, I couldn't help myself) and ness of accounts.
scapegoating ('Kevin caused meto do it'). In addition, Scott and Lyman
foJlowed Sykes and Matza (1957) in identifying a number of different
techniques of justification: denial of injury ('!can'! see what the fuss is Accoums and experiments
about, it's only a scratch'), denial ofvictim ('she deserved everything she
got') and appeallo loyalties ('! owed it to Brian'). They stressed that both We will briefly describe just two experimental studies on the effects of
justifications and excuses are highly conventional acts: they take certain accounts, because our concern is more with the general !style of research
standard forms and are couched in certain socially approved vocabularics than the detailed findings, and lhe great difficultics facirlg any rcsearchcr
(Milis, 1940). The point is that people are nol inventing these accounts who attempts to sensibly capture this subtle form ofl' discourse in an
anew but drawing from a range of pre-existing resources. experimental manipulation. The studies selected illustrate two common
The next developmenl we should note in the study of accounts arase trends in this line of rescarch: the use of written storiek- 'vignettes' ~
from a more social psychological perspective and is associated with the and the attempt to look at accounts in more realistic settings via field work.
work of Gün Semin and Tony Manstead (1983) who provided an extremely Schlenker and Darby (1981) presentcd people with alwritten scenario
useful integrative overview of the field. They, too, emphasized lhe conven- or vignette conceming bumping into others in a public platc and then asked
tional nature of accounts and saw the generation of a comprehensive and them to imagine they were the central character. The scenario was
systematic typology as an importan! task. Their typology added three new manipulated to show the offcnder - the person doing t(~e bumping - as
calegories to excuses and justifications: apologies (Goffman, 1971; having either high or low responsibility and the consequences as either
Schlenker, 1980; Tedeschi and Riess, 1981), requesrs (Goffman, 1971) serious or trivial. The subjects' task was to select the appl-opriate response
and disclaimers (Hewitt and Stokes, 1975). from a list including apologies, excuses, justificatibns and simple
Apologies are the kind of accounts which neither mitiga te responsibility acknowledg?ments. 1

for the suspect act nor attempt to justify it. Instead the apology Not surpnsmgly, Schlenkcr and Darby found that apologies were the
acknowledges that an offence has taken place and implies that thc transgl'_es- most favoured response and as responsibility for the lncident and the
sion will not recur. Requests are accounts used befare the act occurs in damaging consequences incrcased the apology becamc mbre elaborate and
an ·attempt to licence what might be perceived as a violation. Like s_g1f-blaming. Merely saying 'pardon me' was satisfactory for a slight
apologies, they admit the potential offence. For instance, if you simply ,,.- bump, whilc a more extended apology was given for ~ painful knock.
take a cigarette from someone else's pack this could be highly offensive Another study stagcd in a Taranta metro station, in vOlved a briefcase
- but it can be transformed by the use of a rcquesl: 'do you mind if I being kicked over (Ungar 1981). The offender who didithc kicking was
cadge one of your cigarettes?' In fact, it is not easy to turn down requests eithcr high or low status (thcy wore diffcrent sorts ol] clothes in each
of this kind without seeming miserly or unfriendly. condition) and there were three different Jevels o!' exc+e: the oiTendcr
1

78 Discourse hnd social psyclwlogy Accounts in sequence 79


!
l:A l(.tli•VOI..'I, ¡
FIGURE 4.1 either simply walked away after kicking the case over or said 'you shouldn't
A synthctic typology of accounts have left it there, someone might have tripped over it', or a confederate
A Excuses
1

1 -a person acting the role of an ordinary bystander- said, 'she shouldn't


Al D~nial of inr4m ('1 did not intcnd to produce these results') have left it there, someone might ha ve tripped over it'.
1

Accidcnt ¡
Unforsccn i:::onsequences duc to: Following the staged incident, someone nearby who had witnessed the
Jnck of ~Ólowlcdgc, skill, motivntion or environment event was selected and asked about the extent ofthe offender's blame for
Jdcntity of ltnrget person mistnken . , the incident. Any disparaging remarks were noted down as an index of
A2 Denial of ¡•o!ition ('i did not wunt to pcrform thts act l
derogation. Ungar found the offender was blamed less when an excuse
Physical duscs
temporn~ (e.g. fatigue) was provided but that the status of the offender did not effect the amount
semi·pcrfnanent (c.g. _r~rnl~sis).. of blame, although it did cut down the number of derogatory remarks.
Psychological causes ongmatmg m. As we indicated in Chapter Two, there has been a great deal of discussion
sdf (e.gj insunity)
othcrs (e.g. cocrcion)
in social psychology concerning the general adequacy of the kind of
Luck uf aJthority ('1 would likc to hclp, but 1 do not hnve the experimental methods used in these two studies. It is not our intention
uuthorit~') to rehearse the various positions in this controversy here. Indeed, in the
AJ Denial of (/!-/~·ncy light of recent philosophy and sociology of science, we would suggest
Mbtnken i dcntity ('lt wusn't me, huncst')
1

Amnesia <f¡
can't rcmcmber unything about it') there are no perfect methods and the use of any technique involves a
Joint prodUction ('lt wasn't unly me who did it') complex balancing of pros and cons (Barnes, 1982; Chalmers, 1980;
A4 Appeal 10 mlrigating circumstrmces ('1 um not entircly to blame') Collins, 1985). Nonetheless there are sorne points which should be made
Seapegoming - bchnviour wus a response lo othcrs
S¡¡d tales f- arrangcmcnt of facts highlighting dismnl pust specifically in relation to language function. These concem the naturalistic
status of the research, the anonymity of participants and the way the context
1,
was conceptualized.
B Excuses 1
Bl C/aim tlwt Jj}ixt lw.~ ht!f!tz mürepre.u•nted
1
Dcnial oflinjurv (no lmrm done) flé: .,~t/tft.J.t-\, !.,'E/JOJí'A !f'll.
Minimiza/ion ~f injul}' (conscquenccs only trivinlly harmful)
Drawbacks to experiments on accounts
B2 Appeul 10 pfilrciple. of retribr.llirm . . . • .. . • . . . • . . . .
Rcciprod~y (victun dcservmg u~ .m~ury bccausc o~ h.r:fl11.:r ,u.::r~~~)
Dcrogatio'n {victim dcscrving o! lllJUTY bccausc o! lm/hcr qualuu.:s) The problem of experimental naturalism, or ecological validity as it is ;:.:--l"E:·.
BJ Social com¡Jarümr . often known (Brunswik, 1955), is-a thomy one. While sorne workers have 1 -:::'fr::~~_,-tc~
(Othcrs t~b samc nr worsc but g.o unnoticct!. unpunishcd or cvcn prmsctl)
expresscd worries about the possibility of adequately duplicating 'real {J(-!
84 Appt•al 10 h\'g/wr autlwrity
Powcrfullpcrson{s) comnmndcd world' phenomena in thc laboratory (Harré and Secord, 1972: 44-66),
Highcr stlrtus pcrson(s) curmnantlcd others ha ve argued that the quality of experiments does not depend in any
tnstitutio~al rules stipul:ncd way on successfully mimicking aspects of social life. From this point of
B5 Self-fulftlm~tzl . .
vicw the crucial issue is the skilful operationalization of variables which
Sclf-nmirltcnancc (catharsis, psychulogrcal or physrc:rl hcalth)
1

Sclf-dcl'c1opmcnt (personal growth, mind cxpnnsinn) might appcar in very differenl forms outside the laboratory (Turner, 1981).
Conscicn~c {¡rctcd in accord¡¡ncc wilh) The difficulty with experiments on accounts lies at a rather more basic
AppL"nl /1! {rritrciph• of utilitarianism
]](¡ leve!.
Lnw ¡rmlir1rdcr
1
Scl!'-rclia ncc Thc problem with rcscarch using vignettes, as Schlenkcr and Darby
Bcncfrts butwdgh harm themselves admit, is that data are essentially dependent on participants'
B7 j·alue.~
.·1ppr·al to . . . .. theorics of how they would behavc in thc circumstanccs detailed in thc
Political j(c.g. dcmucracy, sm:ml!sm. natmnal!srn)
Moml {lg. Joyahy, frccdom, justicc, cquality)
vignctte. The dependent mensure is not what the participants actual/y do
Rcligiou1 (e.g. churity, ]ove, faith in dcity) · · or say in a real situation but what they imagine they would do or say.
88 Appeal 10 fu.•t.'li for facework . .. , And there may be al! the difference in the world between being confronted
Facc maintcnancc {'lf ¡ \mdn't nctcd likc thnt 1 would havc lost crcdrbtlrty )
with a situation in a vignette and in real Ufe.
RepututiPn building ('1 t!id th¡¡t bccnusc 1 wuntcd to louk tough')
The same sort of problem arises with Ungar's study. However, here
Sorrra: Semin ¡m~ Munstcar.l, 1983: 9\-2. Slightly shoncncd.
the crucial imaginings belong to the researchers and confederatcs when they

!

80 Discourse and social psychology Ac.-cowus i1~ seque11ce 81


!

construct a naturalistic scenario for their experiment in the real world.


1

such as Saussure, Austin, George Mead and Wittgens tein, Sacks had
Are participants responding simply to thc bald provision of an excuse (the published relatively little befare his premature death. MJch of his impact
identified dependent variable) orto sorne more subtle feature of the way has thus been due to his lccturc notes which wcre copi¿d and privatcly
the confederates of lhe experiment design their conduct? The problem when circulated. 1

• !
dealing with accoums is that any wooden style of prcsentation or bad acting The pnncipal locus of this work is thc cvcnts of evetyday, mundane
might be construed as a sign ofthe inauthenticity ofthe excuse giver. That convcrsation. Everyday convcrsation may be simple chduer but ít Í!'i by
is, it might be taken to indicate they do not really mean whatthey say. no means trivial. In terms ofthe historical origins of langua1ge, convenmtion
Furthermore, as we will see la ter in the chaptcr, participams pay clase prcdates lhe developmcnt of written forms of communica~ion. In develop-
attention to the delay in responding whcn accounts are rclcvant to cvcnts mental tcrn~s, the ab~lity lo t.akc part in convcrsational c~changcs .sccm.s
lhat ha ve just happened and lhis is lhe sort of variable which is diftlcult lo be a eructa! parl al bccommg a person (Garvey, 1984;1 Slwtler, 1984 ).
to properly control in cantrived field experimcnts. Moreover, conversatmn 1s all-pcrvasJve. When wc are not work.ing
Thc second problem with thesc studics of accounts i!'i n classic one in sleeping or watching tclevisirí what we are doing mo.litly ik nattcring awa;
social psychological cxperiments - the participant.s are preUominantly to onc anothcr. Sacks arguetl (anU a similar point is madb by Moscovici,
strangers to one anothcr. Exactly why thi.s is crucial for the study of 1984a - sce Chaplcr Sevcn) thal lhis tal k has an cssentid! part to play in
accounL'\ becomes clear when wc considcr the clase associalion \Vith issues thc crcatíon and maintcnance of our social worlds, and \~ould be a good
of po\iteness and formality. lf you don't account for sorne itcm of mis- canJidatc for the primary psychological reality, if onc \~ere looking for
conduct people may well Lhink you are rude; in formal sctlings whcn such a thing (Bcallic, 1983). i

strangers are present accounts may be particularly elaboratc and Conversation analysis i.s very much an outgrowth of ethnpmethodological
ceremonial. Both thc form and content of accounts may be very diffcrcnt research, and shares a number of íts main conccrns (Heritage, [ 984).
when produced between friends or between acquaintances and strangers However, its analyticaJ approach is rathcr different. It concbntratcs on what
(Brown and Levinson, 1978). are, at first sight, the minutiae of naturally occurring con~ersations repre-
The third problem concems context. The experiments described above senled in verbatim transcript. Typically, analysts will stJdy a numbcr of
abstract particular linguistic fonns- excuses and justifications- from their instances of one phenomenon and try to clucidate its systeri,atic properties.
contexts of use and treat them as basic variables. That is, accounts are As studies have accumulated it has become possible ta describe the
viewed outside the specific sequences of tal k in which they are typícally elaborate and detailed architecture of conversation (LeJinson, 1983).
produced and examined in terms of their relationship to predefined This pcrspective gives usa vcry different path to undcrsdncting accounts.
outcomes. This neglects both the detailed organization of accounts and lnstead of attempting to experimental\ y relate them to broa~ variables su eh
their position in particular ongoing sequences of discourse. As we will as the severity of the violation, or status, accounts are e~amined as thcy
see, il is precise! y these phenomena, overlooked in traditional experimental -'- occur, in context, cmbedded in conversational sequenceL Convcrsation
studies, which are vitally important in understanding the nature and use analysls argue. lhallo ll~d out how. accounts work, they p1ust be studicd
of accounts. These problems suggest thal il may be useful to take an m lerms of thetr sequent1al posilton m lalk. It ts necessary for us lo beco me
alternative analytic approach to the study of accounts, and conversation familiar with sorne of these basic sequential properties sd we can see the
analysis is one alternativc which has grown increasingly important in the point _of the analysis conducted in this tradition. For thelnext rew pagcs
\ast few years. we w¡JI go on a detour to illuminatc the primary convetsation analytic
notions of 'adjacency pairing' and 'preference structure 'j Only then will
wc rcturn, bcttcr cquippcd, lo the tapie of accaunts.
Conversatioll analysis and accoums
¡-firlf¡:-d 1.!:oL-• -!¡_;- i>/cr'
Conversation analysis is a youlhful discipline focused on people 's tal k. Mi!!.~~!9' pairs
It is concerncd with how the contributions of different speakers are meshed
together in conversations and the way different types of actions- blamings, Conversation analysts have observed that many of the more general
greelings, excuses- are produced and managed. lts initial development properties of dialogue und the way it is meshed together chn be explained
owes much to the late Harvey Sacks and his co\laborators Emanuel in terms of a .small sct of principies which govcrn thcl changeovcr of
Schegloff and Gail Jcfferson. Like olher notable lwentielh-century figures, speakers (Sacks el al., 1974). Changeovers are rcmarkably orderly with
!
i
82 Discourse!and social psychology Accormts in sequence 83

spcakcrs prccticdng thc cnds of uttcranccs in such a way that thcre is usually 2. A: Is there somcthing bothcring you or not?
( 1.0)
not mueh overl~p of tal k or many gaps betwecn utteranccs. Typically,
A: Ycs or no
there are more lgaps within a speaker's turn of tal k than between turns. ( 1.5)
Within this genhal turn organization conversational analysts ha ve notcd A: Eh?
a furthcr struct~ral fcaturc known as adjacency pairing which is relcvant 8: No.
to the specific hcts performed in the discourse. (Atkinson and Drew, 1979: 52)
At first glante the phenomenon of adjaccncy pairing seems hardly
startling. Typicbl examples of adjacency pairs are questions and answers, The fact that A rcpeats the question is evidence that an answer is expectcd
greetings and r~turn greetings offers and acceptances, assessments and (and the truncated form of the repeat shows that A does not think B has
second assessmbnts. These ha~e a first part produced by one speaker and merely not heard). Indeed, the question is repeated twice until an answer
a second part prbctuced by another. They are also typed, so that a particular is (grudgi¡l_glY.) forthcoming. It is possible to point to many of these i'b .
first - a questi:an, say - requires a second from a restricted range ~f phenomena which show the conditional relevance of adjacency pairs -1;¿-t_s~¡~_,,~
utterances- most likely an answer (Schegloff and Sacks, 1973). 1t JS (Atkinson and Drew, 1979) and suggest this is a central organizational "l:*o.:.:.:-
tempting to thirlk thatthis is exactly the sort of truism which gives social principle of conversation. '
scientists a badi name. But these rather simple initial observations serve The usefulness of this notion seems rather blunted by the observation
as a base allowihg conversation analysts to describe complex organizations that there are often quite a wide range of potential second parts available
(Levinson, 1983). For example, if the first part of the pair is a question,
of talk which rlre far from obvious.
The first think to note about adjacency pairs is that the adjacency should apart from an answer the second part might be a protest of ignorance
not be interpretect strictly. The second part of the pair is not always exactly ('haven't a clue'), a re-route ('ask Dick'), a refusal ('!'m not commenting')
adjacent to the rlrst. For instance, insertion sequences are commonly found. ora challenge to presuppositions (' 1 think yo u are barking up the wrong
Thus in the fodowing extract certmnllrings have to be sorted out befare tree'). On the face of it, this means the adjacency pair is a rather looser
organization than it first appeared. But, therc is a furthcr organizational
thc question isl answered.
principie at work which allows us to arder these different possibilities,
1
known as preference strucrure.
l. Ql. Kevin. What's on next?
Q2. Jand. On this chilnnel or Four?
A2. Kevin. Four. Preference structure
Al. Jand. Ah, it's that thing on the Sandinistas.
Prcfercncc structurc is thc idea that thc second purts of adjacency pairs
i 'fi .
Bcf'orc Janc anSwcrs Kcvin's qucstion shc chccks on its spcct te meamng~ can be rankcd into responses which are preferred and responses which
a sccond qucsÜon/answer adjacency pair is inserted into thc first. As are dispreferred. Thus the preferred response to questions is an expected
evcryone knoJs, scquences of this kind can gct quite complicated as, at answer .while the disprefcrred response is a non-answcr or unexpected
times a considerable number of things ha ve to be dealt with befare the answer. The preferred response to an invitation is an acceptance and the
secon'ct part is '¡offered. dispreferred a refusal. lt should immediately be emphasized that the notion
So adjacencyi he re does not mean the second part will immediately follow of preference is not meant to describe psychological desires or dispositions.
thc first; insteatl conversation analysts suggest the relation is one of condi- A speaker may be personally very keen to refuse an invitation but this
tional relevande (Schegloff, 1968). That is, given that the first part of an does not stop the refusal being, in normative terms, a dispreferred act.
adjacency pair ¡ has been uttered, the second part is immediately r:levam
1

The concept of preference is used to indicate a normarive ranking of


and expected ~ although its actual produetion may depend on sortmg out different responses exhibited in the organization of tal k.
a variety of otner things. The first and second parts are not bound together The following question/answer sequence is typical of those where the
by a strict rul~; the uttering of the first sets up norn;ative expectations second part of the puir is preferred (the large bracket indicates the point
to which spea~ers must attend (Goffman, 1981; Hentage, 1984). where one speaker's tal k starts to overlap the tal k of another speaker).
The operatiÚn of these expectations can be seen in sequences ltke the
following (theifigures in brackets are pauses timed in seconds- see the 3. Hanna: ls Wednesday dcfinitely early closing [ day?
Appcndix for :a full glossary of transcription conventions). Alan: Umhum
84 Discourse ami social psychology Accmtlll.\' 1 .rec¡Jtei/Ct! 85

The first part of the pair is a question and the second an expected answer; S. Dave: l. What's in Jusi ice j(Jr=All
2. (0.3)
that is, an answer related clase! y to the possibilities raised in the question.
Paul: 3. Its (.) ah (.) its about
There are a number of features of the way the answer is del_ivercd which 4. (\.0)
are characteristic of preferred responses. First, the response is produced 5. Wel\
with a mínimum of delay, in fact in this case Alan's answer is given 6. you've got na:h I'm not going to tell you
in slight averlap. Second, ít is a brief response, which means the answcr 7. (\.0)
component of the turn is delivered straight away - the speaker d.oes 8. 'cos you wo:n'L belie:ve the reason I likc he film
not add any extraneous material befare introducing the answer. Thtrd, !

there is no qua!ification or h~ging_in the answer - it is clear-cut and Again we see Paul's response is initiated only after a delay (2-4), it is
positive. V{íA' Ctr marked as a dispreferred second by his use of 'well' (S), his refusal to &" t~·fi+í,av,L
We can contrast this with a sequence involving a dispreferred second. answer is a hesitant and softened one (6). and finally he dffers an account
The following comes directly after Extract Three .(for clarity, Alan's of the non-answering (8). This extract al so shows evidende of the conven-
response has been split into different components). tional, expécted character ofaccounts. Another interestiAg feature ofthis
extract is the full one second delay after the refusal to ~nswer in which
none of the other speakers cut in. This suggests they 4re expecting an
4. Hanna: l. There's no point in going then at al\
account to be offered, as indeed it is. A pause of one Second may not
2. (1.0)
Alan: 3. Well sound likc a long time when thinking about it in the !abstract, but ns
4. not uh I rTiean convcrsation analysts ha ve demonstrated, il can sounctllike a yawning -C.CJ/c /
5. the majar shops will probably be open c:!!_ª_sm in the fast tlow of conversation. ------~ -p¡¡-Tp~¡ .¡r
1 ·

The fact that even a vcry short de!ay is a reliablc tliscriminator of


Here AJan disagrees with Hanna 's assessment about the shopping trip (the preferred from dispreferred seconds means that spea~ers can use thc
preferred second part of this pair would, of course, ha ve been agreement). accurrence of delay as a cue that a dispreferred second is dbout to be given
1

Again, thcre are characteristic fea tu res to thc dclivery. First, thcre is a (Davidson, 1984; Lcvinson, 1983). Witncss thc follnw ing cxtract.
delay component (2). Alan only starts to respond after a pause of about
a second. Then the response is prefaced by the term 'well' (3). This both 6. A: l. C'mon down he:re=it's oka:y i
¡
marks thc disprcferred status ofthc response and incrcases the delay befare 2. (0.2)
the disagreement is broached. Third, when disagreement is formulated A: 3. 1 got Ion¡¡ stuff=I got bc:cr en stuff ' '¡

(4) it is done so rather weakly- contrast 'not uh 1 mean' with the 'no' (Davidson, \985: 105)
or 'certainly not' which Alan could have said. i

At this point we can return to the central tapie of this chapter: accounts. The speaker makes an invilation ( 1) and then therc is sil en ce for a fifth
There is a fourth feature which distinguishes preferred and dispreferred of a second (2). This very short del ay is taken by the spe~ker as evidence
seconds; dispreferred seconds almost invariably include an account. This the response will be a refusal - that is, a dispreferred Second - and he
extract is no exception. Aianjustiftes his disagreement by providing infor- then provides further information which makes the invitation more attrac-
mation about the quality of the shopping (5). That is, rather than simply tive: when the listener arrives they will find lots ofstuffcf'ood?) and beer,
do the disagreeing and lea ve it at that, he accowlls for his disagreement with emphasis on the beer. 1

with Hanna. In general, accounts are a recurrent and routine component The demonstratmn that speakers both closely monitor the silences in
of dispreferred seconds. The detailed content of dispreferred seconds may conversation, and use their occurence to make inferenc~s about the acts
be extreme! y varied but their organization is highly regular. In most cases being performed, has important consequences for the exp1erimental social
an account is included, usually in the same position as Extract Four. psychological work on accounts. In studies where subjedts are assessing
Take another example. In this case the first part is a question and the the adequacy of excuses and justifications from writtenl vignettes (such
dispreferred response is the refusal to answer the question (note that = as Schlenker and Darby, 1981) it is not possible to convh details ofthe
indicates words run together, (.) indicates a short but noticeable pause, : manner of delivery. Yet if this is crucial to the interpretaban of accounts
indicates extension of vowel sounds). The speakers are talking about a - and the conversational analytic work we ha ve examincid he re suggcsts
film called Justice for Al/. it is- doubt is cast on the generality of the findings. MÓreover, in tield

1

Accounts in sequence 87
86 Discourse land social psychology
¡
knowledge drawn on in the account is hard to dispute: the person inviting
experiments (sJeh as Ungar, 1981) it may be the sty1e of de1ivery that
ts unhkely to know about the recipient's future commitments.
is mediating the effectiveness of the account; yet this is virtual! y impossible
1

This kind of account, then, is carefully designed to achieve a number


to examine in rbsearch of this kind.
We have scen!, then, that in everyday conversation accounts commonly of different goals; it does the refusal while avoiding as far as possible
appcar in ccrtai/1 kinds of lUrns, namcly in thc sccond parts of adjacency any negative or critica! consequences. Of course, the speaker is vcry
pairs where th e second part is disprefcrred. Conversants use their
1
unlikely to be thinking about these things or spending time on complex
knowledge of th e delivery fea tu res ofthese turns- knowledge about deJa y
1
planning - they are probably just using the conventional refusal formal
which comes na tu rally. Yet clase attention to the considerations outlined
for cxamplc- tÓ makc judgements about the kind of act being pcrformed.
above helps us understand why such a refusal formal might ha ve evolved
Howevcr, convbrsatio;l analysts have pushed on bcyond this to study thc
into a taken-for-granted social convention.
relhtion betweeb the specific type of account which is offered and the
Conversation analysts have extended this work on accounts in everyday
particular adjadency pair in which it appears.
tal k to make sense of more formal situations. We will take a detailed loo k
i at one example: Max Atkinson and Paul Drew's (1979) research on how
1 accusations and defences are managed in courts of law, an example which
Offers ami acclmnts reveals a great deal about social innuencc processes at work in courts.
1

John Heritage (!984) has provided a useful discussion ofaccounts presented


in the context dr offers and invitations. When somcone makes an offer, Accounts in court
to Jend yo u sor!,ething perhaps, or to invite you to a party, the preferred
response is an acceptance. As we have seen, this response is preferred
1
When we know how speaker changeover is achieved, how conversation on
in the sense thdt it will typically be given with little or no deJa y, it will a tapie is sustained and how corrections are managed in everyday mundane
be positive and will be produced early in the turn of talk. However, if settings we have more of a handle on things happening in formal institu-
you rcfusc thd offer or invitation you wi\1 be making a dispreferrcd tions: when a teacher is controlling a class (Mehan, 1979), when positions
response, and you are likely todo this in the standard way with delay, are formulated in a TV news interview (Heritage, 1984), when a scientist
marking ancl atl account. In principie you could use all kinds of different graciously accepts the Nobel Prize (Mu1kay, 1984a) or when a 'charis-
¡_~~ whcnl making this response. Yct H:rit<Jgc no_t~d th<Jt accounts matic' politician such as Margaret Thatcher attempts to solicit applause
in this case overwhelmingly cluster around the 1ssue ofabil1ty (Drew, 1984; from an audience (Atkinson, 1984a). In each of these cases everyday
Levinson, J98B; Merritt, 1976). Thus you don't turn down the party conversational practices are refined and modified to fit the requirements of
invitation becaÓse the host's friends are so boring (at least not to the host's the situation. Atkinson and Drew's work is a good illustration of what can
face) but beca~sc you ha ve another, more pressing, arrangement. be achieved here. They effectively demonstrate how an understanding of
The followiJo is a typical refusal of an invitation. infonnal, unconstrained conversation can aid understanding of interaction
lo in a more formal situation.
7. Mark: We w're wondcring if you wantcd to come over Saturday, f'r
cÚnner. Questions and accusations, answers and rebuttals v"--rrír'lA~cov'.' CD:O: tt!l ¡.,_-)
(0.4)
Jnne: Wcll (.) .hh it'd be grcat but wc promiscd Caro\ already.
' Atkinson and Drew began by noting two basic ways in which the examina-
tion of witnesses in court differs from the conduct of ordinary conversation.
The invitation ls refused using the standard dispreferred format. It contains
First, the turn arder is fixed - on the whole, who speaks next is not
an account whibh makes refercnce toan earlier commitment. The advantage
managed locally on a turn by turn basis, but is pre-ordained: counsel
of an account like this is its 'no blame' quality. It avoids the implication
speaks, then witness, then counsel and so on. Second, certain kinds of
that the invitadon is unwanted or unattractive. Furthermore, as lt invokes
turn are preconstituted. In witness examinations it is nonnally the case
new informadon (the commitment to Carol) it does not imply that the
that turns should be question and answer pairs with the counsel asking
person givinglthe invitation is being unthinking. Since there is no way
the questions ~nd the witness answering. Moreover, each tum oftalk should
the invitcr coÚ!d know of this commitment tlle·y can not be accused of
be, at least minimally, a question or an answer.
making an inv!tation which they knew would be turned down. Finally, the
88 Discourse and social psychology Accotmts il¡ sequence 89
!

Within these restrictions, however, talk is managed in basieally the 9. C: \Vas therc firing in Sandy Row thut night? i

same way as in everyday conversation. Participants use thcir turns W· Not to my knowlcdgc /
to do different kinds of things, to bring off different sorts of goals. C: Wi\1 you loo k at 00.55 hours: 'Automatic liring in Sundy Row'.
Being a court of law, one of the goals of the prosecuting counsel is W: No, that is not corrccl. 1

! (1979: 70)
to make accusations and challenges. Counsel does this by deploying
sorne sequences of questions rather than others. Witnesses, on thc other
The important thing to note about this cxtract is thc orderlaf the counsel 's
hand, often wish to rebut accusations, and thus they offer particular
questions. The first elicits information from the witness \\'hile the second
types of answerS.LPut simply, the counsel constructs accusations out
question provides apparently discrepan/ information. The effect of this
of sequences of questions and the witness constructs rebuttals out of
second question is to cha/lenge the testimony of the wit~ess. This is not
sequences of answers..;j
simply Atkinson and Drew's theoretical interpretation, but! how the witness
We are all familiar with the conversational phenomenon where one 1

sees things too; his reply displays his understanding that the sequence is
general form of tal k is used to achieve a different purpose. For example,
a challenge. Jnstcad of noting the lag book entry and wt\iting for furthcr
wc may couch a request in the form of a statcmcnt (see Chaptcr Twu),
qucstions he issues an explicit rebuttal. By questioningl thc correctness
or an invitation in the form of a question as in the following case.
of the lag book entry the witness attempts lo undermine the force of the
challenge to his evidcnce. 1

8. A: Why don't you come and sec me somc [times One of thc fcatures of thc rcstriction of turns to qucstibns and answers
8: l would likc to is that it is very difficult to allocale. blame directly. lndbed, it normally
(Atkinson and Drcw, 1979: 50) takes two (as in Extract Nine) or more turns to achlcve a blaming.
1 -

This opens the possibility of the witness deflecting blame befare it


has bcen propcrly assigncd. The counscl may then havb to recover thc
Although A's utterance takes the syntactic formal' a question, the response
is not one which fits a question, i.e. an answer, it is an acceptance. The
deflection to rcstart the blaming sequence and so on. a/ al so means that
the witness has a choice of where to produce their ctef~ence. They can
recipient, B, treats A 's utterance as an invitation.
give it befare the blaming is completed or after. Dcspit~ this diffcrcnce
In courts, this packaging of acts into question and answcr sequcnces
both kinds of defence tend to orientate to the samc Óbjectives: first,
is a delicate matter. For if the counsel is seen to stray too far beyond the
to avoid allocating blamc to oneself; sccond, and IeJs obviously, to
mínima! requirement of asking questions they may be reprimanded by the
minimizc disagreement with the information provided ~y the counscl in
judge. On the other hand, if witnesses make statcments which are Seen
the prior question. 1
to exceed mere answers to the question they may be struck from the record.
One of Atkinson and Drew's most interesting findings ~oncerns thc dif-
Thus accusations, denials etc. have to be carcfully managed within the
ference bctwcen thesc two kinds of dcfcncc as witncssc~ use contrasting
broad question/answer adjacency pair.
stratcgies to achievc the two objectives.
So, how do counsels produce accusations in the course of constrained
questioning? The following extract comes from the proceedings of the
Scarman Tribunal on the 1969 disturbances in Northern Ireland. This
tribunal was set up to investigate police practíces after a disturbance in De.fences made befare blaming is complered
which Protestants invaded a Catholic area, buming and ]ooting. The police
were subsequently accused of being too lenient with Protestant offenders As we ha ve notcd, thc general qucstion/answcr constraint mcans that the
and even abetting their activities, in line with general uneasiness about counscl has to build up to blamings and accusations o ver a lnumbcr or turns.
the role of the largely Protestan! Royal Ulster Constabulary in keeping In the Scarman Tribunal the blameworthy issuc was thdfailure of RUC
arder in Northern Ireland. As with the other extracts we will use from officers todo certain things (arrest rioters, protect properiy), so a blaming
Atkinson and Drew, our first extract records the examination of Royal is produced when the counsel makes this failure noticcable. '
However, the
Ulster Constabulary (RUC) officers by counsel and foeuses particular] y build up to making the failure to act noticeable provideslthe witness with
on why they did not intervene to quell rioting and assault (C is counsel, an opportunity to produce a defence befare the accusa1ion of failure to
W is witness, the 00.55 hours referred to is an entry in a police station act is completed. 1

lag book). The general form of these defences plays down the imparance or severity
!

90 DiscoursJ and social psyclwlogy Accounts in sequence 91


1

ofthe events inlqucstion. The witness manages to imply that action against JI. C: Whnl did you do nt thnt point?
rioters was unnccessury. This kind of dcfcncc is illustrated in the next W: 1 wns not in a vcry good position to do nnything. Wc wcrc under
extract.
1

1
gunfirc nt the time.

The rebuttal is the first part of the response to the counsel's completed
1

JO. C: YoJ saw this ncwspaper shop bcing petral bombcd on thc front of
Divis Street? accusation. The aihi is to counter the presupposition of the prior question
W: Ye~. that the witness slzould and could have done something; for example,
C: Ho~ many petral bombs were thrown into it? stopped the crowd throwing petral bombs. It does not, however, disagree
\V: Onl~ a couplc. 1 felt that thc window was already brokcn and that with the presupposition that this action miglzt have been relevant.
thcrc was part of it burning and this was a rckindling of the flames.
The second part of the defensive response is the account, supplying an
(1979: 137)
excuse for inaetivity. It supplants the idea that the witness did not want
1

to do anything or was not trying to do anything by suggesting a powerful


The counsel hJs not yet asked the witness why he did nothing to stop the eonstraint. Who is going to try and stop a crowd throwing petral bombs
bombing - butl this accusation is anticipated and the witness produces a when they are being shot at? In Semin and Manstead's terms, this form
description of t~e scene which minimizes the need for action. We can note of account is a denial of volition- the witness intended todo sOmething,
the use of 'only' to qualify the number of petral bombs used and the but was thwarted by the physieal danger from gunfire.
description sugkesting damage had already been done, with the implication In these cases it seems the effectiveness of the account depends on
it was not worth bothering about. The overall upshot of the defence is rnaking the constraints on action as strong as possible. Given that the
that thcrc was ria point in further action as nothing could be done to rescue witness did nothing, the greater the eonstraint the less the blame. If we
thc shop. In te~ms of Semin and Manstead's typology, defenees of this examine thc extract below we sec an extended passage of talk from the
kind wauld be~t be classified as a species of justification, namely a claim witness which gives a powerful picture of causal constraint.
that an effectllas been misrepresented ('ee Figure 4.1). These kinds of
1

claims, of cou'rse, need to be extreme! y earefully managed: 'what's a v,v /'1<i--¡¡;i'¡i;t=-.r.!C:;


12. C: Did you know theJ!!QQ were .liE-ble.. to follow you on this occnsion?
couple of petrbl bombs more or less' might begin to sound like black W: What 1 knew and what actually took place are two quite different
humour. 1
1

things. 1 was powerless to prevent quite a Jot ofthings that did happen.
In addition tb giving an account whieh underplays the damage, these In fnct we were just a drop in the ocean with so few in number. I
defences can detlect attention away from matters where blame could realized ccrtainly that there would be sorne of them following behind.
potentiallv be rillocated. In this case, the witness concentrates on the issue 1 could not do anything to prevent that. 1 saw damagc causcd and
of property, ig~oring the issue of whether he should ha ve tried to arrest tried to prevent it. I realized it was bashing my head against a brick
wall really; wc wcrc complere(v ineffecrual.
the people thrdwing petral bombs. In this way, he attempts to select the
(1979: 164- emphasis added)
ground from fhich to eonduct the defenee.
This, then, is the first type of defence, occurring befare the counsel
displays to the! court the RUC man's failure to take action. The second We see the witness constructing a version of events using a number
majar type of defence occurs after the blaming has been eompleted, and of rhetorical devices to maximize the impression of causal constraint.
involves a verY diffcrent type of account. We have already come across the use of 'extreme case' formulations
(Pomerantz, 1986) in Chapter Two. This aceount joins severa! of these
together, as well as drawing on metaphorical constructions such as 'drop
1

in the ocean' and 'bashing my head against a briek wall'. Overall, the
Defences mad~ cifter blaming is completed witness undermines the blaming by displaying himself as having the best
1
of intentions, someone who tried as hard as he could, but was unable to
In Atkinson antl Drew's data, defences following completed blamings had do anything beeause of the weight of constraint.
two components- a rebuttal andan account, and these components could One way of Iooking at constraint in psyehology is to Iook at the acwal
either be prodpced in a single turn or in separate turns. The following eonstraints and limits on people, while a second very common procedure
extract (actually a continuation of Extraet Ten) is a typical example of is to loo k at people 's attribwions of constraint, That is, to loo k at the beliefs
this kind of d~fence.
!
1

92 Discourse and social psychology Accounrs inlseque11ce 93


¡

people express about causal processes when they give explanations (Heider, large extent, when we study social influence, the truth br falsity of an
1958; Jones and Davis, 1965; Kelley, 1967). In this chapter, and indeed account rnay be Jess relevant than thc constructivc proced~rcs adopted in
in the book as a whale, we have described a third important approach. the courtroom.
This conccntrates on the construction of versions of constraint in discourse.
Atk.inson and Drew's work has illustrated the connection between the
version of the world put together in cross-examination and the accounting Discourse and accounts
work it was intended to perform. In this case, they have shown how
descriptions are manufactured to achieve rebuttals. This chapter has described two contrasting approaches tq accounts from
At first exarnination, this k.ind of conversation analysis concerned with social psychology and conversalion analysis. The social psychological
turn tak.ing, adjacency pairs and preference structures may seem to ha ve approach has conccntrated to date on the dcvelopment 9r typologies of
little to offer the social psychologist. Social psychology is generally thought accounts and the identification of relationships between different variables,
to be about broader questions, and the study of excuses or accounts and such as the type of account, severity of transgression ~nd status. The
sequential structure seems to ha ve more to do with etiquette than social conversation analytic approach works frorn the assurnption that phenomena
1

issues. In this respect Atk.inson and Drew's research provides a salutory like accounts will be best understood through looking at l~eir positioning
lesson. It points to sorne exciting new lines of research and dernonstrates in sequences of discourse performing differenl kinds of act. Thus this
the relevance of conversation analysis. approach views accounts as conversational events produc~d in the sccond
Over the years the study of social influence has formed one of the main part of an adjacency pair when a dispreferred response is being supplied,
research areas of social psychology and recently this work has been for instance, when an offer is rejected or an accusation! challenged ..
extended to the study of influence and court juries. Generally research This research has clarified how accounts can be developed to deal wtth
has focused on relatively macro-variables such as the status and physical a number of differenl kinds of problems: turning down un offer without
attractiveness of defendants (e.g. Kulka and Kessler, 1978; Sigall and being offensive or challenging an accusation without gdnerating strong
Ostrove, 1975). The impression a jury member forms of a witness will disagreement. When understood as conventional proced~res adopted for
also, however, be based on their talk when examined, yct thejury member problems of this kind it is possible lo make sense of both lhe broad
will probably be unable to articulate how they arrive at their impression , categories of account giving and the fine details of indiJiidual accounts.
and may well be unaware of the cumulative effects of the procedures Conversation analysis h?s a.lso. show~ ha~ we can ~ove 1 fr~rn every.day
investigated by Atk.inson and Drew. Conversation analytic research in settings to more formal, mstJtutJOnal sttuauons where mterestmg questtons
courtrooms demonstrates how we might begin to investigate the k.ind of of control and power are raised. !

influences on juries which are so delicate that they cannot be grossly The findings ofthis tradition raise apparently insurmoun~able difficulties
characterized (Drew, 1985a). for the experimental social psychological approach to accbunts. Evidence
Atk.inson and Drew's research also calls into question two commonplace concerning the important role of detailed linguistic featurds such as del ay.
guiding principies in court research: firsl lhat witnesses are generally stress and the selection of particular words casts doubt a ni research which
passive and counsels generally active, and second, that the crucial issue is based on written vignettes and on experiments which gdnerate accounts
is the truth or falsity of witness' testimony. As we ha ve seen, witnesses, in the field using confederates. This is not to say that theselproblems could
allhough highly constrained, are nol necessarily passive victims of not in principie be surmounted, or that there is no value in experimental
counsels, but capable of actively developing and packaging their own work on accounts; there may be sorne questions which a fe very dil'ficult
versions of events. Furthennore, although fact and fiction are vital marters to answer without experimental techniques. Howevcr, th~ importancc of
for courts to establish, lhey may not be the most useful or the most vital sorne of these phenomena would not ha ve become so clear without the
categories for lhe researcher. The RUC constable's testimony, regardless systematic analysis of discourse conducted by convefsation analysts
of whether it will be seen as true or false, constructs a version of events. coupled with a theoretical perspective which treats accourits in the context
Even if false, it is an effective, persuasive defence. As we argued in of language function. !

Chapter Two, certain sorts of descriptions should not be ruled out of social More generally, in conversation analysis, we can see some of the broader
psychological analysis because they appear to be factual descriptions of theoretical aims described in Chapter Two lleshed out in a detailed,
events. Factual descriptions are also constructions, they are also doing analy~_ically bascd, rcscarch programmc. Thc general c{mccrn with thc
social psychological work and are equally worthy of our altention. Toa action ~,pricntation of tal k is now seen in tcrms of thc dftailcd study of
!
1

94 Discoufse and social psychology


1

adjacency pairs (in effect, paired actions) and the way people mesh pairs
together to Perform further actions. Most importantly, conversation
analysis emp~asizes the vital na tu re of understanding each conversational S
turn in term~ of the sequence in which it is embedded. The utility of
sequcntial anblysis should now be fully cvident. Speaking subjects

In the previous chapter we were concerned with discourse in relation to


accounts of breakdowns or disturbing actions. We will now move on to
the broader tapie of discourse methods for characterizing people: the
language of the self and 'subject'. This is a crucial area for social
psychology given that the discipline usually takes the individual as its main
unit of analysis, and defining the shape and nature of this individual is
seen as involving sorne model or theory of the self.
The selfhas been a controvcrsial arca within social psychology. At one
stage it scemed impossible to study scientifically something which has such
confusing and diverse manifestations and which appears to be so prívate
and hidden within the individual; this conflict was particularly centred
on the disputes between humanist and behaviourist approaches (Matson,
1973; Wann, 1964). In recent years argument has continued to flourish
but it has takcn on a new form as social psychologists have come to question
thc terms of the debate itsclf and the vcry notion of thc individual or the
subject which is al the hcart of psychology (Gergen and Davis, 1985;
Harré, 1983; Henriques et al., 1984; Sampson, 1983; Shotter, 1984).
Our aim in this chaplcr is lo describe sorne of thc traditional models
of the sclf developed in social psychology and then follow the new twists
in the dispute, producing a critique of the principal assumptions underlying
this traditional model building. We shall describe the challenges emerging
from recent language-based or 'social constructivist' approaches which
take a similar line to the discourse approach, and we will want to argue
that, in the last analysis, any sociopsychological image of the self, in fact
the very possibility of a self concept, is inextricably dependent on the
Iinguistic practices used in everyday life to make sense of our own and
others' actions.

Traditional images of thc self

The key assumption behind all the traditional models we wish to describe
is that the self is an entity and, like any other entity or natural physical
object, it can be described definitively, once and foral!. In other words,
it is assumed that the self has one true nature or set of characteristics waiting Y
to be discovered and once discovered a correct description of these
charactcristics will follow. Thcorizing about thc sclf is thus thought to
!
1

96 Discourse ami social psycho/ngy Speakin~ subjects 97


'
be rather like thcorizing about a chemical. Much of the structure of the thc Pickwicks, for instancc, ofDickem;' novels, as ifthey limply are thcir
chemical may be hidden and mysterious; different research groups may dispositions or the su m of thcir traits: honcst, lazy, Machidvell ian, earnest
for a time offer competing descriptions, but given proper investigative or whatever. These traits alonc provide sufficient motive Pnd explanation
methods, the true verídica! description of the chemical will ultimately for thcir actions. Uriah Heep will continue to be underhand and obsequious
emerge. whatever the situation as that is his disposition, justas Eysknck 's extravert
This assumption is one a discourse approach will want to qucstion but will continue lo be the life and so u! of the party. 1

for the moment Jet us examine sorne ofthe competing theories which ha ve In no sensc could the honest soul or trait theory person be sccn as acting
been proposed and their different claims as to what counts as a proper out a part of 'managing' the impression given to others, they are entirely
description of this object- the se! f. Our review will build on an earlier synonymous with their disposition and identif)' completbiy with it. The
discussion of this topic (Potter et al., 1984: ch. 8), and will focus on the extravert is not putting on a facade, they cannot help bbt be extravert,
metaphorical elements of these models rather than the technical details it is their disposition. For this reason it would be meaningl ess total k about
1

of the theories. seeing them as proposing a set of 'character types'. this kind of selfhaving an identity crisis (Rorty, 1976). Hbnest souls ha ve
only one identíty, not many, and there is no distance or sdparation within
the self to produce the possibility of this kind of self-cÓnnict. It is not
Trait theo!J'.' the self as an 'honest soul' within the capacity of Eysenck's extravert, for ínstance, tÓ wonder if they
are being 'authentic' to their 'true' self, they always ar~ that self. This
Trait theory has been one of the most influential models in social speculation is just not possible if the person is conceptualized in this way.
psychology. This theory regards the se! fas a 'personality'. That is, it sees Trait theory people cannot be 'game players'. they arel 'honest' souls,
the person as consisting of measurable personality traits, abilities and conveying, in Trilling's terms, a 'sentiment of being', oran aura of solidity
attributes. Traits muy be eithcr superficial or deeply rootcd, constitutionally and unrcllectiveness. 1

inherent or the result ofchild rearing (Cattell, 1966; Eysenck, 1953). A Critics of trait theory und thosc who wish to o!Tet un alternativc
person's behaviour or actions are thought to be largely determined by thc description of thc self ha ve noted the wcakncsses and liJnitations of this
combination oftraiL'i thcy posscs.o.;. The."ic truit.'i will nutwcigh thc inllucncc 1

conccptualization. Tmit thcmy is highly a.'incial in its app[ nach, it ignores


of thc immediate situation or the context surrounding the person. lt is this thc inconsistency in human bchaviour (Mischcl, 1968) Givcn cnough
image of the self which underlies the concept of the personality test or personality test data, trait thcorists can comment extensively and toa high
personality inventory with its aim of measuring individual differences in leve! of sophistication on thc traits individuals posscss. Hbwever, in their
specific traits such as extraversion or introversion and recording the precise own tcrms, this effort is wasted if, as it seems, lhese traits ticar little relation
patterning and combination of traits within a particular individual. to how people perform in non-test situations. The majori\y of people, for
Asan example ofthis approach, consider the following person descrip- example, may come out as partly cxtravert and partly introv~rt on personality
tion offered by Eysenck and Eysenck (1964): tests. Whcther introvcrsion or cxtraversion dominatcs dcpcnds on thcir
perception ofthe situation. They cease to be predictable ~nd trait-Jriven.
The typical extravert is sociable, likes parties, has many friends, needs to havc Furthermore, according to the opponents of trait theod, people are not
people lo talk to, and does not like reading or studying by himselL He craves just 'personalities', they are social creaturcs, they ha ve different parts to
excitement, takcs chanccs, often sticks his neck out, acts on thc spur of the
play in society which require differcnt maniti!stations oft~e self: mothers,
momenl, and is gene rally an impulsive individual. He is fond of practica! jokes,
always has a ready answer, and gencrally likes change; he is care·frec, casy· daughters, doctors, style setters etc. This is the line that has been taken
going, optimistic, and likes to 'laugh and be merry'. He prcfers to keep moving by a second very influential description of the self in so¿ial psychology.
and doing things, tends to be aggressive and lose his temper quickly; altogcthcr namely role theory. The social situation anda person'sl position within
his feelings are not undcr tight control, and he is not always a rcliable person. a society, not their charactcr, is seen as channelling b~haviour.
(Eysenck and Eysenck, 1964: 8).
'

What kind of assumptions are being made he re about the nature of people'? Role theol)'.' the theatrical image of rhe self
Trait theory adopts what the literary critic Lionel Trilling (1974) has called
!
the 'honest soul • approach to the self. People are viewed as if they are Role theory perceives a tension between two 'facts': the fact of individual
like the minar characters in a Victorian novel, the Doras, thc Uriah Heeps, self-expression and thc fact of society and social determini~m. The concepl

1
1

98 Discout..re and social psyclzo/ogy Speaking subjects 99


!
ofrolc is a means ofrcconciliation. Roles are dcfincd as sets ofactivities, unself-conscious in response and thus sincere to thcir disposition, whereas
qualities and ~tyles of behaviour that are associated with social positions the role player is capable of putting on a facade, hiding behind a mask,
(Dahrendorf,¡1973, provides a classic description). Social positions and acting out a pre-ordained script. This aspect of self-experience,
exist indepentlently of any particular individual; they are impersonal or inherent in this image of the self, is very well expressed in sorne of the
supra-individltai, and include, among others, occupational, religious, novelist David Lodge's characters.
recreational ahct kin categories. Modero, complex, differentiated societies
require the sabia! positions of wives, husbands, criminals, priests, dentists,
It is obvious, from his stiff, upright posture, and fulsomc gratitude to the air
politicians, fo'r instance. People who fill these social pasitions are expected
stcwnrdcss scrving him a glass of orange juice, that Philip Swallow, flying
to act in the hppropriate way, they Jearn to play a role. Through being westward, is unaccustomcd to uir travel [ ] The sang froid of his fellow
fixed in a cerlain social position in a structured society, a person acquires passengcrs is u constunt source of wonderment to him, and he observes their
a self anda rdrm of self-expression. Thus what determines a person's self deportmcnt carefully. Flying for Philip Swullow is essentially a dramatic
and their pedonality is the social positions they occupy; dispositions are performance, and he approaches it like a gume amateur actor determined to
varied and sdcially manufactured. hold his own in the company of word-perfect professionals. To speak the truth,
According fa role theorists, people conform to their roles not just because he approaches most of life's challenges in the same spirit. He is a mimetic
man ... (1978: 8-10)
of obvious sahetions and rewards for doing so (e.g. prison sentences and
higher salarie!) but beca use of the more subtle mechanisms of socialization.
Children Jearh to see themselves as others see them (Mead, 1934), the Philip Swallow is not presented as possessing a personality in the way
self is a 'Jooking glass selr (Cooley, 1902), it comes to reflect social i the trait theory person possesses his or her personality, he simply possesses
expectations. lndividuals
'
leam to 'refer' to their social groups and through 1
the ability to put on the masks required of him, or reflect the expectations
a process of ·~acial comparison' adjust their identities (Festinger, 1954). of his society. The attributes belong to the roles, a1though after life-Jong
1

The girl and the boy Jearn their appropriate sex-role, for instance, and practice at playing the roles, they will become habitually assumed, and
beco me femiÍline and masculinc adults. the individual will gain an equal predictability.
A very diffbrent kind of self-experience is obviously being posited here. With role and trait theories therefore, we have two competing versions
People ceasejto become 'natural' characters, they become performers, of what makes up the self and the individual and how they are best
~acial charac crs; thc dramatic and thc theatrical begin to provide more described. These two accounts are generally seen as mutually exclusive
appropriate n\etaphors (Goffman, 1959). Like the actor on the stage, a options in line with the model building assumption outlined earlier in which
person's actidns are not cxpressive of a unique personality, but expressive the self, like other objects, can only have one correct description. The
of thcir role, land most individuals are interchangeable when it comes to final image of the self we shall consider adds yet a third, supposedly
role playing.¡ As Gombrowicz notes: definitive, version ofwhat the selfis und thus how we experience ourselves.

A human ticing does not externalize himself direct\y und immediutc\y in


conformity With bis own nuturc; he invuriably docs so by way of somc dcfinite Humanistic theories: the romantic image of the se/f
form; und t11Ul form, style, way of speaking and responding, docs not derive
solely from him, but are imposcd on him from without - and the sume man
can exprcss himsclf somctimcs wisc\y, sometimcs foolishly, bloodthirstily or From a humanistic point of view, the classic role theory account of the
angclically, mnturcly or immnturcly, according to thc form, thC style presentcd sclf prescnts a morally offensive image ar people. The role player is
to him by t)1c outside world . .. (1979) portrayed as a social dope, a victim of social circumstance. Dahrendorf
1 argues, people know themselves and their friends as richer characters than
The role-pla~er self thus has two characteristics denied the honest soul. this, people believe they ha ve a sense of control and agency,·a sense of
First, a kind of social insincerity and, second, a multiple set of possibly creating new social forrns as well as being created by them. The role player
discordant id¿ntitics. The possibility of individual fragmcntation is intro- is perhaps but a paJe imitation of human potentialities (Hamlyn, 1974).
duced, differÓnt roles may require different selves, the person is divided Role theory places the self firmly within the public real m but inherent
bctwccn manY responses and may cxpcricncc a conflict aliento the tnadel within its l'ormulation is thc possibility of a distinction between public and
of the person assumed by trait theory.
1

private se! ves. There is the social self described by role theory, but there
lnsincerity ~trises from bcing aware of society. The honest soul is seen as is also a truc or real background self which chooses lo act out the roles
1
i

100 Discourse and social psJ•clwlogy Speaking ¡.1"11/Jjects 1O1

and which monitors the success or failure of the perfonnance. The humanist What is being asswned? 1

notes, there is room for a kind of double consciousness, justas the actor
on the stage has a double consciousness. There is the role being acted As we pointed out at the beginning, each ofthese models ofthe se!f claims
to be the only val id self-portrait; cither beca use it supposediy cncapsulates
out and there is the actor's real self, which guides the performance and,
in sorne sense, remains removed or alienated from the part being played. sorne phenomenological or experiential truth (humanistic ltheory), is sup-
The humanistic tradition in psychology exploits this tension and potential
ported by psychometric research (trait theory), or appears to be the most
sociologically coherent analytic too! for the social scientistl(role theory). It
self-estrangement (Perls, 1971; Maslow, 1968; Rogers, 1961). It has
argued that both trait type theories and role theories wrongly represent is taken for granted that through paticnt investigation, whether it be the
human nature and that psychological research and therapy should focus phenomenologiql stripping away of !ayers or the obserlvation of large
numbers ofpeople, the true nature and substance ofthc self~ ill be revca!cd.
1
on this authentic true self. An individual's life is seen as a process of
searching to establish this true self, as a quest for self-fulfilment and self- This core assumption is a versan of the realist princ pie. McCabe's
actualization. The role player mentality and the sense ofalienation which summary of the features of this principie for film and li erary narrativo
characterize modero life indicate that this quest has been interrupted and discourses applies equally wel! to psychological discou1se.
is incompletc.
The narrativc discourse cD.nnot be mistaken in its idcntificutions because the narra-
The following extract from Fritz Perls nicely il!ustrates the assumptions tivc discourse is not prcsent as discourse- as urticulation. The uriqucstioned nature
at work here: ofthc narrativc disc~ursc cntails that thc only problcn~ rcalitylpl~s~s is to go ami
look and scc what 77ungs thcrc are [ ] Thc rcal1s not artJculatcd- 11 !S. ( 1974: 12)
In Gcstalt therapy, we are working for something else. We are herc to promotc 1

thc growth proccss and dcvclop thc human potcntial. Wc do not tal k of instant Thc psychological self-images we ha ve bccn considcring ~imilarly do not
joy, instant sensory awarencss, inslant cure. Thc growth proccss ís a proccss call attention to themsclvcs as constructiom; or discursivc 1Irticulations but
which takcs lime. In therapy wc not only havc to gcttbrough thc rolc-playing.
We also have to flll in the hales in the personality to make the person whole present themselves as rcprcsentations of thc real objectf .
Other assumptions are implicated here (Sampson, !983D. It is taken for
and complete again [ ] you have to invcst in yoursclf and it takes time to
grow. (1971) grantec.J that this sclf, thc objcct to be discnvcrcd, ii
thc CL'IIIre nf'
experience, an initiator of action, a coherent whole, sep~rate from other
There seem to be two, probably contradictory, images within this tradition. distinct selves. These aspects are crucial to psychology's conception of
First, there is the concept of a pre-existent, authentic self which muy be the individual person and the human subject. They makd it possible, for
inhibited and thus require rediscovery, and, second, there is a self which example, to contrast the individual with sociery, as natural pairs in a
is not pre-existent but needs to be willed into existence and self-consciously balanced dichotomy. The individual can be posed as thd causal unit the
created from the wreckage of the cultural debris cluttering up the psyche.
1

researcher retreats to when explaining social phenomena, and represented


The !atter self implies hard work, patient and careful self-analysis, as the possible origin point for society and, indeed, of any I!now!edge about
exercising and trying out the new psychical elements which will make the world (Henriques et al., 1984). It is presupposed that thé selfis a unitary
up the more fulfilled authentic self. While the first kind of self more entity, a relatively autonomous agent in the world and th~s a so urce and
frequently stresses spontaneity and simply being. It is assumed the conven- explanation of the kind of phenomena which interest the ~sychologist. In
tions of social and pub!ic life forma veneer over an older, deeper, more fact this assumption comes to define what the psychologist should study
basic self we all share in common; this veneer can be stripped away to and the prominence given to the laws of motivation, for !instance, or the
reveal the healthy whole self underneath. cognitive structure of the person. Even for the role thebrist, the self is
We ha ve now considered three separate accounts of the self, presented seen as bounded in this sense, and as the origin point o!¡ f experience.
as character types rather than empirically based theories. There are
many other versions of the self available in psychology but in different
ways trait thcory, role theory und humanistic theories ha ve been the Towards a new conception of the suhject
most central to the development of the discipline. Befare going on and
examining the challenges to the general framework underlying these Attacks on these fundamental propositions havc emer~ed
from many
models, let us f1rst considcr in more dctail thc assumptions bchind this diffcrent sources. Thcy havc reccntly becn co-ordinatctl, howcvcr, by
kind of model building. Kenneth Gergen (1985) under the general rubric of 'social bonstructivism', (~ \

1
1

102 Discou~se and social psychology Speaking subjects 103


1

thus providing a summary of the main points of disagrecrnent. In this universal features of personhood. Psychology may simply e!aborate instead
section we wi)I consider the critica! points raised by Gergen and others upon the conventional ways people are described in this particular society.
and the growth in alternativo !anguage-based approaches to the se!f. It Psychologists may think in terms of traits, roles and autonomous selves
is worth noting in advance that, unlike sorne other developments reviewed beca use these are part of our culture's common sense about the self (Potter
in previous ch apters, the movcrnent to·a discursive model of the self has
1

1
et al., 1984: ch. 8). If it is the case that psychologica! mode!s of the self
been more af~ccted by trends in semiology and post-structuralism than rellect not the true nature of the object, as supposed, but our social history,
ethnomethodology, conversation ana!ysis and the study of speech acts. the rea!ist principie would be undermined. The researcher should be
For this rcasoh this perspective is strong on theory and is more abstract encouraged to begin focusing on the multiplicity of self-constructions and
and philosoph/cal!y complcx with f'cw analyscs of the !!!!)!Land ~_k.ind their social and interpersonal functions.
encountered in other chapters. In the final section of the chapter we will Various careful analyses of indigenous psychologies or the psychological
return to this ~ort of analysis but for the moment the less concrete will theories of non-Western people (Hee!as and Lock, 1981), of the history
take precedenbe. of the self-concept in the West (Henriques et al, 1984; Lyons, 1978),
The main object of the critica! movement has been to displace attcntion and discussions of the overlaps in the images found in social science
from the self-ds-entity and focus it on the methods of constructing the se! f. and in other arenas such as literature (Potter et al., 1984; Trilling,
That is, the q¿cstion becames not what is thc truc nature of the self. but 1974; Wetherell, 1983, 1986) have given weight to this claim. We will
how is the se!!f talked about, how is it theorized in discourse? This is a shortly consider one type of non-Western indigenous psychology to
move which ib welcomed by the discourse analyst. It is suggested that demonstrate this strand of the urgument and the divergencies possible in
mcthods of mbking sense are the key to any kind of exolunation of the self-understanding.
1
se ¡l.·, as peop 1lf ' S Sense O f thCffiSC1ves !S
. m. fact 8 CongJOmcratC
'lh""'f'lh,;,¡- O f these The other majar strand in this new approach to the self is to set up new
.•
1~ methods, produced through talk and theorizing. There is not 'one' self prescriptions for future study. The social psychologist is now asked to
/. waiting to be lctiscovered or uncovered but a multitudc of sclves found study something different, not the person behaving in an 'environment',
. in the diffcrcnt kinds of linguistic practiccs articulated now, in the past, but the language practices and discourses prevalent in different contexts
historically arld cross-cultural!y. 'Considered from this point of view', (Harré, 1983; Shotter, 1984). Gergen describes the background to the new
claims Harré,l 'to be a self is not to be a certain kind of being but to be research framework as follows:
in possession of a ccrtain k.ind of theory' (1985b: 262).
From this perspectivo, the trait theory, role theory and humanistic Thc languagc of person descriptian (behaviarallanguage) is not Jinkcd in any
character typcS discussed in the last section become, not competing mode!s, systematic way to spatiotemporal configurations of the body, Thus it appears
but different 8ossiblc methods of making scnse that someone might draw the rules for 'what counts as what' are inherently arnbiguous, cantinuausly
cvolving, und free to vary with the predilections of those who use thcm r ]
upon to describe themselves. These character types thus cense to be seen Whether an act is dcfincd as cnvy, nirtation, or angcr floats on a sea of social
as mutually e~clusive options; a person could, at one moment, depict interchange. Intcrpretation may be suggcsted, fastened upon, and abandaned
themselves in ¡trait terms, at another moment in role terms or even blend as social relationships unfold across time. (1985: 6)
two or three of thesc types together in their self construction (Potter et
1 al., 1984: 45(7). It is meaningless to ask which of these models is As we noted in the discussions of semiology in Chapter One and rules
1 the correct description of the self, they are simply equivalent ways of in Chapter Three, any human act does not determine the interpretations
' making scnse 1af the self with their own context, relativc advantages and placed an it, varying interpretations and constructians of the action might
disadvantages. 1
emerge. Gergen argues that, if this is the case, then the task of the social
There are kcveral di fferent strands combined in this constructive psychologist is to study the variation of interpretation in descriptions of
approach to t1\e self which we need to tease out. A majar strand is ·the people. What is crucial is the discursive functions served by particular
argument thatl psychological models of the self are inevitably cultural! y interpretatians.
and historicall~ contingent, dependent on certain kinds of social practices. Gergen is perhaps overstressing the degree of open-endedness and llexi-
The assumptibns on which modern Western psychology is based, the bility in this process in arder to emphasize the general point he wishes to
concept of th.\ self as the centre of experience, for instance, whiéh we make. As the examples from conversatian analysis in Chapter Three
discussed abare, may be peculiar to this period of history and this type demonstrate, there is considerable constraint and regularity in the procedures
of society. What psychology discovers, therefore, may not be the time!ess peaple use for articulating ar making sense af an actian, and successfully
!
i

104 Discourse and social psychology Speakinb subjects 105


1

'bringing off their version of events in a persuasive manner. We would According to Harré, Maori self-experience is ro6ted in a wider
want to argue that not every interpretation or construction of the self will cosmology which enlivens the material world, so that •ehch harmless and
be acceptable in a specific situation. But the imporlant point is that the quiescent thing is charged: in our tcrms it has potenti~l energy' (1983:
social psychologist abandons the attempt to describe once and foral! human 91). The person and the physical wor1d are thus represented as a set of
acts and their meanings, subduing and struggling against indexicality, and force fields, where the forces are interreiated with thd activities of the
instead attempts to describe the linguistic practices which produce an act gods or spiritual ancestors (atua) of a tribal group. 1

as a certain type of event. The social status and position given to any individual within the social
This type of research is not an id le or dilettante activity either. As many group is dependent on their birth and genea1ogy, and a person beco mes
cultural analysts have demonstrated, it is a critica!, political and potentially invested with a particular kind of power, mana, as a re~ult of their birth
emancipatory activity. The methods of conceptualizing the self in volved in circumstances. The positive successful achievements dr
a high-ranking
different linguistic practices ha ve vital consequences for the positioning of person in war or intergroup negotiation is hence attribtited to his strong
people in society; they are not neutral or without impact, they produce mano, given by the gods, which is made manifest i~ actions. Mona,
senses of the self which may be negative, destructive, oppressive, as well however, is a variable or inconstant force rnatching the aontingencies and
as senses which might change and liberate (Coward, 1984; Henriques variability of everyday lite. It could wane if ritual observ~nces, particular! y
et al., 1984; Parker, forthcoming; Williamson, 1978). in relation to the spiritual ancestors, are not observed, and thus the
In the following sections we will enlarge on the two majar strands of previously successful chief could sudden1y fail in battle a~ his power drains
this new approach to the self summarizcd hcrc, demonstrating through out of him beca use of sorne misdcmeanour of which he n~ay note ven ha ve
the work of particular theorists how the challenge against traditional models been aware. 1

ofthe selfhas been mounted and, with more precision, what an alternative If one views the world in this kind of wuy, with the individual seen
conception might involve. We willlook first at the argument that concep- as the si te of varicd and variable externa! forces (and +anais only one
tions ofthe selfare cultural! y relative, and then at the claim that to research ofthese possible forces which inhabit the individual), then different kinds
the self we need to study language.
1

of self-experience become possib1e. Specifically, indiv idua1s can cense


to represent themselves as the centre and origili of their actions, a
conception which has been taken to be vital to Wester~ concepts of the
1

Jndigenous psyclwlogies: the Moori self se! f. The individual Maorí does not 0\1'11 experiences such as the emotions
of fear, anger, lave, grief; rather they are visitations governed by the
While there may be sorne universals in how·people across different cultures unseen world of powers and forces, just as an inanima e object, a stonc
conceive of the self (Lock, 1981), so that, in fact, sorne basics of self- or pebble, can be invested with magical taboo powers so that touching
organization may be prerequisites for even a minimally organized society; it places the offender in danger. As Smith points out,l
it is clear cultures systematically differ in their theories of the person and
self-practices. These differences are exceedingly difficu1t to ana1yse and
The Maori individual was an amalgam of various independenl organs of
decipher. It is a perilous activity comparing across cultures. But this research experience, and it would appear from the description oflthese [ ] that to a
is fascinating beca use at its best it makes our own cultural practices appear signilicant extent these organs reacted to externa! stimuli ir¡dependently of thc
strange and unnatural, as ifthey could be other than they are. This is perhaps 'se!r. Thus Maori experience campa red wilh our own Was impersonal and
the first step in successful discourse analysis, the suspension ofbelief in what objective. Because t.he 'self was nol in control ofexperíence,la man's experience
one normal! y takes for granted, as we begin to think about how a practice was not felt to be integral to him; it happened in him but not ofhim. A Muori
is constructed and what it assumes rather than seeing itas a mere reflection indiviJual was nol so much thc cxpcricnccr of his cxpcric1\cc as thc obscrvcr
of it. (Smith, 1981: 152)
of an unprob1ematic reality. The relativity of our own discourse of se1ves 1

West~rner
and people is rnade explicit when other cultures are studied.
One of the most interesting cultures that has been studied from this lt is possible, given these descriptions, for the to begin to
perspective is the Maorí culture of Aotearoa or New Zealand. The account imagine a different kind of mental life, almost a different world of
ofthe Maori selfwhich follows is taken from descriptions given by Harré sensational experience, and through this imagining to bbgin to de-centre
( 1983) and Smith (1981) which in turn acknowledge the anthropological our own conventicos for self-expression. As Smith is hware, however,
work conducted by E1sdon Bcst (1922). there is thc risk that wc are still dealing in idealizations. A comparison of
106 Discowlse and social psyclzolagy Speaking subjects 107
1

the conceptioÁs af the self in two cultures requires a reification and ourselves and to describe the self as split when introspecting on self-
homagenizatidn of those conceptions in arder to compare at all. We are experience; it thus explains one kind of self-construction. This feature of
also missing th e Maori reading of our conventions to set beside our reading
1

language makes sense, too, of sorne of the ways the self has been under-
of theirs. 1
stood in social psychology.
Ironically, Jcry little is known about thc ordinary construction of the We saw that a central feature of role theory formulations of the self
self in our kinds of society. The Maori version contrasts with most aspects is the possible division of people into public and private se! ves so that
of idealized 'festero psychological models of the self but is it so alien there is a social self, a me, andan 1 that is aware and can monitor this
to the way people m thts soctety represent themselves tn everyday conversa- social self, evaluating it pretty muchas one might cornment on the actions
tion? Peoplc frequently talk about being taken over by emotions, about of a friend. 1 can, in other words, see how this me appears to other people,
being the viclim of their nerves; that is, as out of control of the1r without feeling 1 am complete!y identified with this particular me. 1 can
experience. Tb properly discuss and compare Western images of the self always create other types of me. This is, of course, also the starting point
with the Maoh images in this way we first need to mvestigate our own for humanist theories and the search to define the authentic self from this
indigenous psychologies with the care and attention anthropologists have plethora of divided se! ves.
paid to the foleign and the strange. But we know enough at this stage to Harré argues that it would be a mistake to take this commonplace
support lhe clkim lhat different cultures and historien! periods produce way of thinking about the self at face value and, for instance, erect a
different sclf-practices. philosophy or metaphysics of the self which might give a real-world
ontological status to these I's and me 's, seeing them as actual constituent
entities. And it would be a mistake to follow the traditional psychological
The grammJcal and metaphorical self path and attribute a descriptive phenomenological significance to the
possibilities of self-division, thus theorizing about the split ego or different
As we noled,¡the second main strand in thi7 new approach to the self is layers of consciousness. The I and the me or the subdivided self, aceording
thc crnphasis wlaccd on languagc as thc mcd1um for self-constru~tlon and to Harré, is simply a fea tu re of grammar and the constraints and potential-
thus for rese~rch on the sclf. Harré has described the human mfant as ities of self-description afforded by that grammar. lt simply reflects one
an 'apprcnticé person' who has to acquire the 'linguistic resources for way we ha ve for accounting for ourselves linguistically.
acls of self-d~scriplion' (1985b: 259). The infanl is in much thc same Harré offcrs sorne cxamples of how this prescntational grammatical
position, theréforc, as the social analyst of the self. In ~ach case ~ body device works in practice, such as 'you are deceiving yourselr, 'I made
of knowledgclneeds to be acquired about how the self zs conventtonally myself do it', 'llalked myself into it' (1977c: 330-4). In each of these
articulatcd. what is originally gained unawarcs has to be 'self-consciously' examples the speaker makes use of the subject/object property of language
rctricvcd anc{v und rc-cxamincd. to create two se! ves: a rccalcitrant, dilatory, unwilling selfand a powerful,
Language is lhe medium for this knowledge for both lhe child and controlling self. It is easy to see how this mundane linguistic practice could
thc social sdcntist. As Harré says, the child is acquiring 'linguistic be confused with a genuine split into sub-agents in the head. It is also
rcsourccs'. sdlf-cxpcricncc is formed as the child lcarns the grammar of clear lhat lhis abjlity lo divide lhe self in language is a very useful
our language hnct perfects communication. They also learn the :~~versa­ accounting strategy, through it one can disavow responsibility, manage
tionally accep[able ways of presenting oneself as a person. Credtbthty and accusations, claim credit and so on.
the power to dersuade others of what one wants them to accept only come This, lhen is one approach advocated by the new conception of the self
lhrough the u~e of sorne routes to self-expression, and what is credible which is emerging. To look at how the self is constructed, the social
and conventidnally acceptable in one context may beco me unsatisfactory, scientist should research the grammatical matrix and everyday language
ineffective cofnmunication in another. The pressure to be accountable and usage. In addition, we should study metaphor and analogy, these too can
intelligible t~ others sustains and gives power to certain communal cast light on why we understand ourselves as we do.
organization \Jf self-experience (Gergen, 1987; Shotter, 1984) ..... Shotter (1985), following Lakoff and Johnson (1980), has drawn atten-
One of the ¡most familiar organizations in our grammar IS the ab1hty tion to sorne of our guiding metaphors for the mind. We think of the mind,
lo split the pe1son into subject and object. A person can refe; to themselves for examp!e, as a container (things are 'held' in mind); as a machine (minds
as bolh I and me. According to Harré (1977c; see also Harre, forthcommg) which fail to 'function' properly) and also spatially so that there are 'deep'
this grammati?al fea tu re makes perfect sense of our willingness to subdivide Ievels of consciousness and 'higher' mental processes. Structural or
'
'
1

108 Discourse ami social psychology Speaking subjects 109


!

architectural rnetaphors and strearn or water irnages are arnong the rnost He also argues that this pressure explains the way lhe jself disciplines
pervasive in literary and psychological discourse. Thus it is common to s.uch as psychology and the philosophy of the mind ha ve¡ developed over
think of the rnind as divided into regions or spaces, sorne of which rnay b=. 1

be 'partitioned' off so that the person can think of themselves, if necessary,


as breaking down mental 'barriers' - a metaphor repeatedly employed In effcct, ns wnrrnnts nrc dcvclopcd, disputed, and elaboratcd in dcfcnsc, the
by humanistic psychologists. There is also the furniture of the mind: the result is n rich and varicgatcd languagc of the sclf along wilhlsets of supponing
self may become cluttered up with old-fashioned and out-dated apparatus. institutions and practices. What wc takc to be the dimensions of sclf in the
prcsent era may thus be viewed as thc accumulatcd nrmamcn~arium of centurics
The irnage of the self as a strearn of consciousness has a prestigious of debate. They are symbolic rcsourccs, ns it wcre, for nlaking claims in a
history in psychology. William James (reprinted, 1968) argued that the sea of compcting world constructions. (fonhcoming) 1

rnind was cornposed of a flood, or successive series ofthoughts and irnages.


He carne to this conclusion, he thought, as a result of careful introspection Thus psychologists and philosophers ha ve reconstructed ¡their theories of
on self-experience but the origins rnay líe elsewhere, in a discursive systern the self and introduced richer and more complex accounts prompted by
which offers cxactly this kind of mctaphorical shift. their own nced for successful sclf-preRcntation in thc facb of compctition
between philosophicnl schools nnd the desire to gnin carltrol and power.
We would want lO take a somewhat lcss individualisticl approach to the
The ideological scll' prcssurcs favouring ccrtain sc[f-discourscs over others. Thc urge to sclf-
presentation seems too narrow a basis on which to buil~ un analysis or
One of the dangers of lhis kind of grammatical analysis is lhalthe social variability and variation in broad social sclf-constructiÓns.
dirnensions of self-discourse can be neglected. Why, for instance, do Gergcn's approach can be contrasted with a different dradition derivcd
sorne kinds of Jinguistic practiccs and sorne self-construr.:tions flourish at principally frorn sorne intcrprctations ar thc work of Fouduit ( 1970, 1972,
particular historical periods and in certain societies, only to fall out of 1981), lhe earlier work ofthe Frankfurt School (cf. Sum~son, 1983) and
use in other periods? In concentrating on grammar we can forget that recent Marxist analyses ofsubjeclivity, discourse and idedlogy (Althusser,
ccrtain constructions ofthe selfmay survive bccause thcy serve importrmt 1971; Cowurd und Ellis, 1977). Thc main thrust al' thiJ work has bccn
social functions or maintain a particular kind of society. to demonstrate how the discursive articulatio\1 of certai~ kinds of se! ves
or human subjects is inlimately involved in lhe reprodt!ctioo of certain
kinds of sociely. 1

Warranting voice In lhis lradition, people bccomc fl-red in posilion through the rangc
of linguislic practices available to them to muke sens1\. The use of a
Gergen (forthcoming) argues lhat the motive force behind the dominance particular discourse which contains a particular organiúition'
of the self
of sorne self-constructions is people's desire for 'voice' or speaking rights, not only allows one to warrant andjustify one's actions in!Gergen's scnsc,
lheir wish to ha ve lheir interpretation of evenls prevail against competing it al so rnaintains power relations and patterns of domination and subordina-
versions. The self is thus articulated in discourse in ways that will maximize tion. In constructing the self in one way, other constructiobs are excluded,
one's warrant or clairn to be heard. Sorne versions of the self will thus hence, to use a common phrnse found in this tradition,[ the creation of
come to predominate in sorne contexts. one kind of self or subjectivity in discourse also creates a particular kind
Gergen contrasts this kind of self-presentational ímpetus to self- of subjection. 'A subjectivity is produccd in discoursb as the self is
construction with clairns to voice which arise from the naked use of power,

r
subjecled lo discourse' (Parker, forthcoming). 1

where rnight is used to force through a version of events, oran economic


ímpetus, where the ability to construct in the most effective way possib1e
í is first dependent on having the money todo so (e.g. in advertising). He Discourse and power
/ is suggesting that in everyday life with one's friends, family and colleagues 1

voice is delermined by how skilfully one can use warranting conventions. Severa! concrete analyses of thc power relations involv~d in discursivc
And a vital part of warranting one's actions, rnaking thern appear self-presentations have been conducted on tapies which s~ould interest the
reasonable and justifiable, is being able lo present differenl kinds of the social psychologist (e.g. Coward, 1984, und Hollway in Henriqucs et al.,
self ap.propriately. 1984, on constructions of femininíty and masculinily; and IHeuth, 1982, on
/r oÚ-'.i ¡1
n "
r~_C;"-{J:¡,Vhf!
1
i

Jl O Disco11~se ond social psychology Speaking subjects 111


!

sexual relatidns). Rosalind Coward has demonstrated how discursive Self-construction and intergroup conjlict
practices aim~d specifically at women 's selves construct a version of what
should be plclasurable for women and thus encourages specific female The examples we shall consider here concern the construction of selves
desires. She bamines the assembly of fashionable body images, ideal that occurs when people attempt to make sense of intergroup conflict and
homes, the cobplex of encouragements and prohibitions organized around in particular when they present one version of the conflict as factual. Our
food and reciPes in women's magazines and looks at how radio discjockeys extracts will come from open-ended interviews conducted with New
and popular inu5ic construct the people who are their audience. Al! of Zealanders concerning the question of sporting links with South Africa
thcsc realms are crucial for how wom~n sce lhemsel~cs, .women are
1
and more specifically the 1981 Springbok rugby tour of New Zealand
positioned as consumers through the destres developed m th1s way, and (Wctherell and Potter, forthcoming b).
although this¡ discourse is pleasurable it is also oppressive. This tour was a highly controversia! one in New Zealand. According
to opinion polls it was not supported by thc majority of the population,
1
although there were a number of New Zealanders who felt very strongly
I see the rbpresentations of fcmalc pleasure and dcsire as producing and the tour should go ahead, and ít was strongly opposed by a large coalition
sustaining f~minine positions. Thcse positions are neither distant roles imposed of anti-apartheid campaigners who initiated protests and civil disturbance
on us from butsidc which it wou\d be casy to kick off, nor are they thc cssential
during this period. One of the scheduled matches had to be cancelled
nttributcs of femininity. Femininc positions are produced as responses to the
plcasures orifcred to us; our subjectivity and identity are fonned in the definitions because of the degree of disturbance and every game involved a demonstra·
of dcsirc ,hích cncirclc us. (1984: 16) tion of sorne kind, culminating, in the final match of the tour, in what
has bcen dcscribed as a riot. The conduct of the policc during these protests
was later called into question and became part of the debate surrounding
The grammalical perspective on the self reminds us of the contingency the events of 1981 (Shears and Gidley, 1982).
of sel f-constrl.Jction and its multi fariousness while this kind of ideological Those we interviewed can be described as onlookers ar spectators
analysis remlnds us that it is not a game which is being played here. of this conflict, in the sense that none were involved in the protest
Research int6 discourse concerns crucial elements of people's lives, not movement, although all had generally strong views about the events
only pleasure,and desire but suffering and enslavement, and the possibilities surrounding the tour, both pro and anti. In the interview they were given
for any kind of life in this society. the task of accounting for or explaining the conflict and violence to
the interviewer and articulating their position. They did so, as we will
i try to demanstrate, by drawing on particular models of human nature in
Discourse ahat vsis at work on the self
1 .
arder to blame and accuse sOrne parties while excusing and justifying the
behaviour of others. Varying constructions of the self were brought into
Although, as¡ we have seen, t.he. new social c~nstructivist. approach.t~ the play in the process. These were generally third person models in the sense
subject has dcvcloped a soph1stlcatcd theoretlcal alternatlve to tradJt10nal that other people rather than one's own self were being described. But
self-conccpt ~esearch, thcrc are few examples of dctailed empirical analysis from our discourse perspective, the same analytic principies apply, and
to back it up.¡we ha ve a strong theoretical ration~le for starting a program- discourse about others' selves is just as interesting as discourse about one 's
matic study of sclf-discoursc but thc rescarch 1tsclf has yct to be done own sclf.
in a systenu~tic l'ashion.
In the last Section of this chapter wc will attempt to illustrate with sorne
cxamp\es thb kind of empirical programme required. In line with the Violence, human nature and rariona/ control of the self
analytic methods advocated so far wc will stress function in particular.
Not only do ~ve need to be able to describe the cantent of rcprescn_tations Commonly three kinds of narrativc charactcrs were produced in accounts.
of people in 6irferent contexts or the sheer range of self-imagcs available One charactcr was developed to explain the role of the poi ice, and the
in ardinary t'alk, but we a\so nccd lo ask how these images are used and two olhcrs lo account for the actions of the protestors.
to what cnd, hnct thus what they achicve for thc speaker immeUiatcly, !ntcr- When asked ifthe police had been to blame for the violence ofthe tour,
personally, 1nd then in terms of wider social implications. many of those interviewed responded in the follawing way:
i
!

112 Discourse and social psychology Speaking lwbjects 113


!

J. With the police Jct's facc ittbey're only people, they get angry you know Protestors as a class thus tend to be divíded into those with ~enuine motives
and you quite often sce them get carrícd away which is you know. bmv and those who are not genuine. This division salves a particular kind of
quite, um, how can 1 say, forgivcuble. { ] 1 would suy it was u ~as~ of ~be accounting problcm for those intervicwcd. Most wished Ita support thc
police gcning fcd up cos they arcjust pcoplc afler al!, and 1 can qUJtc 1mag1ne right to protest but also to condcmn violent protest. Thcisubdivision of
becoming heartily sick of the whole business and lashing out at someone. protestors achieves this goal by separating what are two distinct principies
(Bradman: 15-16) into two types of people or narrative characters. 1

2. They might have madc a few mistakes. But you know policcmen ure only The concept of genuine and non-genuine motives is a fascinating one
ordinary people, they must ha ve hud a lot of provocation and 1don 't blame because of the model of the mind and rationality it prdupposes. This
them if at the last they were a bit rough. (Owcn: 16) beco mes clearer if we loo k at further descriptions of the ·Jrrong elemcnt'
3. 1 think the policc acled vcry wcll. They'rc only human, if lhcy lashcd ?ut or the non-genuine protestar. 1

and crackcd a skull occasional\y, it was, hah, only u vcry human acuon !
1

!'m su re. (Bird: 11) 7. 1 feel vcry strongly that it gavc troublc-mukcrs who wcrcn't!: rcally intcrcstcd
in thc basic morals of it an opportunity to gel in and causd troublc, lo bcut
What is interesting about these accounts - and the many similar ones in up pcoplc, to smash up property. (Bird: 7) i

our sample - is the concept of human nature, or ordinary human reactions 8. There's always groups around the country, for instancc, m~ybe a biker gang
they draw on. Human nature, the basic person, appears as flawed and or somcthing, who are anli-police, and it is a known faCt that whcn thc
brutish. Most of the time this uncivil na tu re is quiescent or restrained in Springbok Tour happened, it was an excuse for them to bé able to gct into
sorne way, but given a sufficiently extreme situation or sufficient provoca- something and do something [ ] I reckon it was an outlel for all sorts of
other groups as well who wercn 't really interestcd in tbe issue but just the
1

tion it will appear. An aggressive response to provocation is depicted as


fact of bcing ablc to be violenl. (James: 20-1)
natural or preprogranuned in human beings~ presumably, the rational social
1

agent has to Jeam at least a veneer of control, albeit an imperfect control, 9. A Jot of protcstors wcren't actually for the cause, they j~st wcnt for thc
aver this underlying aggressive potential. Of course, the police are fun of the game. (Milis: 12) 1

supposed to maintain self-control, but these accounts stress that they too 1

·.r \h1re at root 'only human'. What moti vates this character is the pleasure of violence, !the 'fun of the
\'.'1 f¡_l~~ v This character construction thus stresses the fallibility of human beings game'. It is accepted that the desire to cause trouble, to tiggress against
)' vi· and the inevitable falling away from 'higher' standards. (A dominant others is an understandable motive and sufficicnt explanatioÓ for behaviour
'~j.. \ metaphor is to speak ofbeing 'carried' away or 'swept' along.) Its adoptlon in itself. This is subtly different from the way the poi ice )vere depicted.
~ < in this context allows the poli ce to be excused and their behaviour justified Whereas the police were shown as having a natural viole nt response to 1

- the police cannot be blamed if evcryone would ha ve acted in this way provocation, thcy were seen to ncither seek it out or cnjoy it (e f. Extracts
(sce Wethercll and Potter, forthcoming b). Onc lo Three). It is this diffcrencc between the poi ice chdracter and the
A different kind of approach tended to be taken for the protcstors. Very non-genuine protestar, ofcourse, that allows the poi ice to be excused and
frequently protestors were divided into two types of people. This represen- this group of protestors to be criticized. The same empha~is on violence
tatíon scemed to be shared by both those who supportcd und npposed thcir at the 'base' of human nature is used but tbe policc violencel is understand-
aims. able and indeed forgiveablc, and the protestan;' violence is worthy ar the
strongest condcmnation. 1

4. 1 think mosl, probubly 90 pcrccnt ofthcm. wcrc thcrc for lhc right rcasons Fr;qucntly this image of protcstors becomcs typi!icd a/1d rcilictl into
a socm. 1 ca tegory, the . stJrrer
. 1
' or •trou bl e-ma k·er ' , -a catcgory w h.te h .ts
and the ones who got on ¡cJcvision and hit thc hcudlines wcrc the 1O percent
who wanted to beat up the police. (Muxwell: JO) immediately recognizable to New Zealanders and preJaient in their
discourse in all kinds of contexts. The 'stirrer' is an extreme! y use fui
5. In my opinion a !ot ofthe protestors that wcrc there wcrcn't gcnuine anti~
apurthei¡;I, in fact if you askcd thcm what apartheid meant thcy probably accounting device. Política! or other motivations behind violence can be
wou\dn't.be ablc to tell you. (Davison: 23) dismissed or ignored since the 'stirrer' is fully explained by this kind of
natural self which they possess, no other kind of account is needed.
6. 1 think thal along with somc very gcnuine protestors, thcrc wcrc sorne very
Thinking in terms of wider ideological consequences, the~efore, wc can
non-gcnuine protcstors, the people thut nre thcre just because thcrc is a
stirring going on. (Pond: 6) see how this techniquc for making sense of collective action~ might protect

1
114 Discoursl and social psychology Speaking subjects 115

and maintuin al ccrtain kind of status quo uncha!lenged. The violen! that the police take on the attributes of the non-genuine protestar and the
protestar does not nccd to be taken seriously or spcculatcd about any protestar the attributcs or thc police.
further. As the~ ha ve no genuine motives and are completely defined by
their understandab\c cnjoymcnt of violencc, encrgy necd not be invested 12. It. seemed.likc he f~~fercnce to spccific poli.ccmanl was nlmost enjoying the
VJOlencc and thal kmd of gave me <1 reve<~lmg son of si de lo it. (Rack: 14)
in listening to (hem. They are a source of violence; police violence is
merely a respobe (cf. Reicher and Potter, 1985). 13. I wouldn't condone violence, but I can sce how [ ] r can see how exas-
The other prbtestor character, the genuine protestar, who acts as a perated Ithc protest?rs] got. U m, obviously lhe lead, the Jeadcrs of the
counterpoint tolthc non-gcnuinc, has more of an honest soul appearance. protest movc~ents l1ke HART and what ha ve you didn 't envisage that son
of confrontatlon al all. (Milis: 12)
For those who support the protest movement, the genuine tend to appear
as admirable, bomrnitted, moral creatures, standing up for what they
Here.it is the police who actively choose to indulge the human propensity
belicve and co(npelled to protest by the gravity of the issue. For those
f~r. ~10lence because they enjoy it and it is the protestors who lapse from
espousing suppbrt for the tour, on the other hand, the genuine protestar
C!Vilrzed control and 'revert' in sorne way to natural human responses when
appears as a oiore bumbling figure as below:
provoked.
1

10. Whal ungercd me really was that a certain small group of New Zealanders, Overall, in this chapter we have tried to show how a discourse or
Minto, Piwutcrc, who are communists 1 belicve, lcd a lot of naive wcll~ language-based approach could revolutionize the study of the subject of
meaningiNew Zcalunders who abhor aparthcid and organized them, you psychology, thc person, the self. We have presented the theoretical
know, lo ¡jump up and down and infringe thc righLs of other New Zealanders. ar~uments ~htch undermine the traditional approach to the self and in
(Pintcr: 11) t~ts last sectJOn wc ha ve presented a tentative ana\ysis of sorne forms of
11. It was mhinly the extreme groups which took over, um and stirred pcoplc ?'s~ourse at w~rk in.~ context where the presentation of other's se! ves
up and, dh. groups likc that can, thcy can agitatc normal pcoplc who rcally ts Vital tothe dtsc.redttmg and increditing of the different groups involved
wouldn'J probably cvcn think much about it, und are tloing things thatlhcy m a conO tel. Obv10usly thcsc forms of discourse need to be cxplored much
wouldn't nonnally do and they're usually thc oncs who get caughl. (Wood: more tho.roughly and mdeed the analysis of self-discourse and person
9-10) 1 constructJOn has ~nly JUSt begun in earnest, but enough has been achievcd
to show thc functJOnal cmphasis ofthis analysis and what might be gained
In these cxtraFts the gcnuine protestar character is constructed, as it
from continuing this enterprise.
generally was iby those who wished to support the tour, as a suggestible
innocent, unaJ.,are of thc implications of their actions, the well-meaning
but cssentialiY misguided humanitarian who is easily persuaded into
supporting dubious causes.
This charact~rization like that ofthe 'stirrer' is a very effective accounting
strategy. lt arlicu1ates a self, a particular kind of personality or brand of
honest soul, a~d the individual's actions are then explained by this se! f.
The interviewée, hence, is rather like the trait theorist. People's behaviour
is explained b~ their set of traits. What the discourse analyst sees is that
in creating thi~ character the psychologist and the New Zealander are not
simply neutr~lly describing reality but constructing a version of events
which has a ~articular practica! consequence for the New Zealanders at
least: discredlting this group of people.
One of the h1ost intercsting things about the characterizations -We_have
described here is the sheer ficxibility of the resource. A particular image
1

of human na~ure is produced, for example. This image can be utilized


eithcr to blafnc or excuse, to praise or condemn. The object can be
reconstructed in various ways to offer persuasive versions. For instancc,
1

in the extrads below we can see how the accounts can be reversed so
1

. .111
eategones i/1.ISCotlrse //7
i
In this chapter we will overview the social psychalogic~\ and discaurse
approachcs in turn. Wc will stress in particular the im~lications of the
6 discourse analyses for social psychologícal work on catcigorization, and
suggest certain lessons for social research more general! y.
Categories in discourse

Social psychology and social categories


Categorization is an important and pervasive part of people 's discourse. 1

In the course of conversation everyone populates their lives with friends, '
Imagine looking ata telcvision screen in an arca of extrerrleiy poor rcccp-
doctors, Americans, extraverts, immigrants and a thesauras of other
tion. An image is there, but il is very difficultto deciph er in the sea of
1

categories of people. Pie k up any newspaper and many of the stories will
dots. Someonc familiar with the programme then points! out that this is
concern people who are described, evaluated and understood not in terms
a talk show with two personalities in a studio set. You ca1n now 'scc' the
of any unique features of their biography but through their category
image more clearly. Two vague shapes coalesce into p~ople while thc
membership: 'model reveals star's secret life', 'wife found murdered'. .
fu zzmess 1

around becomes the studio. The shapes ha ve become both more


Categorization is no less fundamental to the social scientist. Much af
integ~al and more di~tinct. The calegories (people/studiolset) clarify and
social psychology is con cerned with the attributes of social groups- males,
sJmpl¡fy; certaJn_ rcg10n.s are seco as more similar tlmn berarc ancl ccnain
palitical extremists, workíng-class adolcscents - and experimental or
conlrasts are he1ghtened. The lraditional social psycholqgical approach
survey findings from representative samples ofthese people are recurrently
assumes that processes of social categorization are ultimatBly derived from
extrapolated lo other group members. That is, people are taken lo be
pcrceptual mechanisms of this kind. 1

members of relatively enduring social categories, and in virllw of their


category membership inferences are rnade from the attributes of individuals
to the attributes of the rest of the category. Social categories are, in one way
or another, the principal building blocks in many arcas ofsocial research. Caregories and rhe physical world
As well as being a resource for both lay and scientific explanations of 1

behaviour, social categories ha ve increasingly become a tapie of rescarch A number of experiments ha ve demonstrated effects of thiJ type with non-
in themselves. Social psychologists have focused on thc cognitive pmcesses social objects. In one classic study, for example, Tajfel and Wilkes ( 1963)
underlying categorization and its consequences. What is the cognítíve showed people sels of lines with diiTerenl labels on lhen\ and asked l'or
mechanism involved in categorization? How do people break up the social an cstimate of their length. Allthe lines differed, someiwere long and
world into distincl groupings? And, given thal a category has been applied, sorne short. The crucial finding was that if' allthe shorter lines were label!ed
what effect willthis ha ve on people's perception or understanding? In this A, and the longer ones B, the difference in length betwedn the lines was
traditíon, categorization is seen as a natural phenomenon rather like exaggerated. At the same time there was a tcndency for thb lines carrying
breathing; people automatically lransform the polluting detritus of their the same !abe! to be seen as more similar. That is, if thd lines could be
. d
over-complex physical and social reality into a simplified and rcadily VJewe as members of separa/e categories - A and B - differences wcrc
1

assimilable form. accentuatcd and ifthey could be viewcd as membcrs of the same category1

In contrast to this approach, workers in the more linguistically orientated similarities were accentuated. 1

traditions of ethnomethodology and discourse analysis ha ve bcen interested It is claime~ Lhat this ldnd of categorization process is ada~tive; it clurifics
in how categories are constituted in everyday discourse and the various and systcmatJzes thc physical environmcnt. With less confUsing noisc, and
1
functions they satisfy. Instead of seeing catcgorization as a natural less complexity lo handle, lhe argument runs, the person is belter prepared
phcnomenon- somcthing whichjust happens, automatically- it is rcgarded for aclion (Tajfel, 1981, 1982; Tajfel and Forgas, 1981 )./These kinds of
as a complex and subtle social accomplishmenl. In linc wíth thc central catcgonzatJOn proccsscs havc thus been more or less explicitly glnssctl
themc ofthis boa k, this wotk emphasizes the action orientation of categoriza- in evolutionary terms. The imagc of the hunter on the savan'na, simplifying
. . 1
tions in discoursc. It asks how categories are tlexibly articulated in the , an d systemat1zmg tasty zebras from the irrelevant grass! and worrying
course ar certain sorts ar tal k and writing to accomplish particular goals, leopards is diflicult to rcsist. However. what happcns \~hcn wc mnkc
such as blamings or justifications. categorizations in the social arcnu? 1

1
i

118 Discou)se and social psychology Categories in discourse 119


1

Caregories an1 the social world


1
Categories and prototypes

One common ~nswer to this question is: nothing changes. The effect of Up to now we have described research on the perceptual consequences
perceiving pedple in terms of categories is exactly the same as perceiving of categorization. However, there is another strand of work which is much
objecrs in this ~ay. And indeed a body of research has demonstrated the same more directly concerned wíth the mechanisms underlying this process:
lcind of effects. Wilder (1978), for example, showed sorne experimental with how people 'think' in terms of categories, how, they assign a
participants a ¡videotape of four people, one of whom was offering his particular person to a specific category, for example. Social psychologists
apinions on various tapies. In one version of the video the people were have again plundered the theoretical storehouse of cognitive psychology
presented as a group (they sat at the same table and wore identifying tags), to deal with this problem.
while in anothh version they were merely shown as unrelated individuals The suggestion offered by Cantor and Mischel (1977, 1979) is that
(other differertces were, of course, held constant). The task of the partici- categories are cognitively organized around prototypes; a prototype being a
pants was to judge how similar the opinions of the others were to the typical or paradigm example. The theory of prototypes has, of course, been
speaker. In linc with the experimental hypothesis, Wilder found thatseeing developed mainly in cognitive psychology by Eleanor Rosch and her
people as a g~oup led to judgements of greater similarity between their associates (Rosch et al., 1976). Thus, moving out ofthe social arena for a
views. 1
moment, if we think of birds, the prototypical bird is something like a
In another ekperiment, Allen and Wilder (1979) randomly split a number blackbird or sparrow. This is a classic and familiar example of the category
of boys into t~o groups and asked them about the opinions and preferences 'bird' which shares most of its main features (has wings, flies, goes tweet and
of the boys ih their own and the other group. As expected, the boys so on). At the same time we are aware that there are other birds such as
perceived the¡apinions and preferences of members of their own group ostriches and penguins which are very different from the prototype and may
as similar to heir own views but different to those of the other group. be treated as borderline cases; indeed, there may well be disagreement about
Moreover, thb preferences and beliefs of the boys in the other group whether they should be treated as full members of the category at all.
were seem a~ similar to one another. So, as in the Tajfel and Wilkes Using this notion, Cantor and Mischel offer a more precise statement
experiment ah the length of lines, the opinions of people categorized of how members are assigned to categories. They suggest that the process
together in "!ilder's video, and the views of the boys in the same group is one of matching the potential member to the relevant prototype. Each
wcre seen as more like each other. Similar!y, the differences in perceived person carries around a large set of preformed, mentally encoded proto-
opinions bet\~een the twa categories of boys were exaggerated and types; if the potential member shares enough features with one of these
1

accentuated. 1 it will be included in the category. Thus, if we want to decide whether


What has really excited social psychologists about this lcind of research someone we meet should be categorized as an extravert we compare them
is the possibhity of understandíng and explaining racial stereotyping to our model instance of extravert (loud voice, goes to lots of parties etc.).
and hence disbrimination. Could it be a natural consequence, as it were, If they display enough of these features they will be so categorized.
of the way ~e have evolved to organize the world into categories? It is important to note here that Cantor and Mischel are not proposing
Could prejudlce be the unfortunate outcome of the biases which emerge this process is like following a mechanical algorithm. It may well be that
when the differences between categories are exaggerated and the dif- 110one feature is essential for inclusion. They stress that social categories
ferences witHin them downplayed? are not homogeneous entities where each member shares a specified set
A number br researchers (Hamilton, 1979; Taylor, 1981) ha ve argued of features and no others; rather they are 'fuzzy sets' in which members
exactly that.jAs Wilder pithily writes, 'categorization, per se, propels have many things in common but also many differences.
the individual down the road to bias' (1986: 293). Th1s IS a depressmg A further feature of the theory of prototypes is the claim that categories
conclusion, ¿nd one that we will argue against later, for it suggests will come in hierarchical clusters. To use Cantor and Mischel's example,
that bias andl stereo~yping a.re not ?acial a~d p~yc~ologic~l aberrations, we will not ha ve one, unitary prototype ofa 'cultured person'; rather this
induced perhaps by faulty ch1ld-rearmg or d1stortmg Ideological processes, will be split up into lower leve! categories (e.g. 'patron ofthe arts') which
but an inevit~ble product of the way our cognitivc system is organized, can be subdivided even further ('donator to art museum' and so on- see
lo res pon d t9' categonc~.
. Figure 6.1 ). Consistent prototypes will be drawn u pon only at this more
basic leve! of categorization; and hence it is here that stereotyping of
category membcrs will be most evident (Brewer et al., 1981),
1

120 Discourse and social psychology Caregories in d1~·course 121


i

FIGURE 6.1 AIJ these assumptions flow from the basic social cognition ~pproach which
The organization of sub-prototypes attempts to explain social phenomena within the framewotk of cognitive
psychology. It is cognitive processes and structures which !are seen as the

/~:~;~~
causal determinants here. The social phenornenon of stereotYping is caused
by the way our cognitive systems operate on informationl reducing and
distorting in the light of the organized structures of prototypes.
If we take a discourse-orientated approach to categorizatidn each of these
assumptions becomes rather less convincing. We can contra~t the cognitive
Patron of Man of the p~r~pectiv~ with the image of the person using their discoYrse to perform

/".""\ /\\
dJflcrent kmds of acts which we havc bccn dcvcloping throughoul this
book. From this perspcctive we predict that instead of bbing a 'victim'
of mechanical categorization processes people \Vill both ctJaw flexibly on
preformed categories and construct the sense of categorids as they tal k.

Donator to Supporter of Donator to Gourmet Counurier World


repenory eommunity un muscum truvcllcr Categorization and partícu/arizatioll 1

'

thcutrc orchcstra 1

Sm1rcc: Cantor and Mischel, 1979,


In an important critique, Michacl Billig (1985) had radic111y questioncd
the first of the assumptions identified above. He has arg¿ed against the
idea thut perception should be the primary mctaphor for prejudiced
Problems with traditional categorization research judgement and that simplifying categorizations are neces~arily udaptivc.
He sturts with a basic but telling point. The perceptual lnodel leads us
There is now a large and theoretically sophisticated body of research on the to overlook the fact that people are constantly prejudiced bgainst groups
tapie of categorization and its role in fostering stereotyping and intergroup whose members they have never met and hence we igdore the social
discrimination (see Tajfel, 1981; Wilder, 1986, for reviews from different convention component in prejudice. Furthermore, the stred on perception
perspectives). Despite providing sorne important insights and findings it is seems to make the expression of prejudice in talk a secbnctary or epi-
prone to certain difficulties. We suggest these are a consequence of phe_nom~non .. Given that many people spend considerable ti~e propounding
problematic base assumptions about the phenomenon of categorization and ractst vtews 111 casual conversation about groups they may have rarely
the failure to examine categorization as a social practice involving certain or never seen, this relegation seems entirely pointless. Lke discourse
sorts oflanguage use, Later in the chapter we will go on to examine sorne analysts (see a1so van Dijk, 1985, and Reeves, 1983), Billig suggests that
research which develops this alternative perspective. To begin with, to understand prejudice more fully we need to bring talk o\- rhetoric back
though, let us enumerate these problem assumptions. to centre stage, displacing perception. 1

l. lnevitability of biased categorization. Social psychological research In contr~s~ to the clatm that categoriz.ati~n is a natur~l and adaptive
has cometo view stereotyping as a consequence of purportedly basic and process, BIIhg suggests that both categonzatJon and the OlJposite process
adaptive cognitive processes. People's very perception is seen to be based ofparticularization (splitting categories in parts, or disting¿ishing specific
on categorizations which simplify, and thus result in distorted stereotypes. t~stances from categories) are necessary for dealing with lile world. The
2. Categories have afixed structure. The theory of prototypes suggests dtstorted emphasis on the process of categorization alone has led to what
that categories are fuzzy sets organized around prototypical instances. Billig describes as a 'bureaucratic model of thought'· ••
!
1

Categorical perception is dependent on a process of matching stimuli


to prototypes, which are further organized in hierarchical structures.
i~to
much of Lhinking is seen as a process of locking thc unfarrúlíar safe familiar
categories . .. the image of the person to emerge from this upProach r~sembles
3. Categories are preformed and enduring. People are viewed as that of a bun;nucrat sensibly ordering the messy stimulus wPrld ... Just as a
carrying around a mentally encoded set of preformed and enduring bureaucrat m1ght defend office procedures, so thcse thcorists ~Hamilton, 1979;
prototypes of category members. These will be stable and will determine Snyder, 1981; Rothbart, 19811 tal k of organization, ord~r. managemcnt,
perception and understanding. efflciency, eJe. (1985: 87-8) 1

1
!

122 Disc01¡rse and social psychology Categories in discourse 123

What is Ja¡:;ki~g in discussions of social cognition is the ability to transcend we criticized in earlier chapters. As we indicated, their model of prototypes
thc limits of familiar categorics, to develop new proccdures, and to reach suggests that people cometo identify category members, and indeed grasp
out, through ~articularization, to the unique features of individual cases. the meaning of categories vía typical or paradigm examples which al!
These abilitids are equally 'natural' to social life as those of grouping, properly socialized members ofthe group share. The category is 'anchored'
classifying aÍld categorizing. to the prototype. If people's perccption and understanding are determined
Finally, Billig
'
suggests that if we think about the reality of racist and by their prototypes in this way we should expect to find considerable
stereotyped dategorizations we can see the importance of both rigidity consistency between descriptions of category members given by people
and flexibility in prejudiced judgement. For example, if a white British from within the same subculture.
person is goihg to rigidly maintain that, say, West Indians are 'lazy and
unmotivated'i they will probably need to also constructively and flexibly
manipulate their discourse to deal with any inforrnation threatening this Prototypes of Mooris
generalizatioÓ. Perhaps they will need to generate all sorts of subcategories
of West Indihns to neutralizo potentially problematic examples of hard- We have looked at this issue ofprototype variability in a study conducted
working andl motivated West Indian behaviour. That is, they may well with Ruth McFadyen (Wetherell et al., 1986) on middle-class, white, New
rely on the very techniques of making exceptions and particularizing Zealanders' discourse about Maoris, the indigenous people colonized by
undervalued lby the categorization approach to maintain their biased, the British. In the course of semi-structured interviews, descriptions of
prejudiced response. Billig propases that instead of trying to understand Maoris carne up repeatedly, in tapies such as positive discrimination, land
prejudice in ~erms of the operation of the single process of categorization rights, and the South African rugby tour of 1981. If white New Zealanders
we should mÓve to looking at how both categorization and particularization do use stable prototypes of the catcgory Maori there ought to be a high
are managed wherever talk is manifested. Let us now go on to examine degree of consistency in the depiction of Maoris across different tapies.
the research traditions which have followed this procedure. The problem for the theory of prototypes is that this consistency was
not there, although the Ncw Zcaland situation resembles many othcr 'race
relations' situations in Britain, America, Canada etc. and categorization
Prototypes Ld variabi!ity theory is supposed to be particularly relevant to prejudice and stereotyping
issues. lndeed, we did not need to loo k very far to find examples of highly
In a number lof earlier chapters we demonstratcd how discourse analysts variable discourse about this category of people. For example, we chose
use the detailed examination of variability in accounts as a basic research four interview transcripts at random and noted all adjectives the participants
strategy. Different
'
versions and forms of tal k are the analyzable trace of applied to Maoris. The cntire list is reproduced in Figure 6.2. Out of a
the way langhage is used to bring about different ends. Varying accounts total of sixty-two adjectives, only three were the same: using the term
can be thoug~t of as the residue of the social practices through which people 'friendly'. Of course, if we repeat this exercise with more transcripts we
organize thei!r lives; practices ranging, as we have seen, from persuading find sorne increase in overlapping terms - but we also find a concomitant
others of thth correctness of a large-scale scientific theory through to increase in unique tenns. It is clear that the variability is at least as strik.ing
accounting for'
the refusal of a party invitation; from justifying racial as any consistency.
inequality in la society to choosing a turn of phrase which starts to clase As we saw in Chapter Two, a common technique social psychologists
down a teleRhone conversation. have introduced to deal with variability of this kind is the use of some~-'z'¡"~o
At the sanie time, the embarrassing presence of radical and pervasive form of gross categorization. That is, the adoption of a categorizationc.::-J.- i;•
variability ca'us into question those theoretical perspectives - whether they scheme which is broad and ambiguous enough to eliminate variability,
focus on attitudes,
'
rules, the self - which were meant to reveal an or at least to cut it down. This kind of scheme could, for example, classify
illuminating bansistency in sociallife. To be of heuristic value, theories together apparently similar adjectives such as 'dirty' and 'scruffy'. The
of this kind Jhould be able to explain actual patterns of social behaviour, problem here, however, is that consistency becomes an achievement of
and allow us! to predict patterns in the future. Failure in this central task the researcher rather than a feature of the discourse. For example, partici-
is fatal. j pants themselves may, on occasion, see important differences between the
Cantor ano Mischel 's theory of the operation of prototypes in the sense of 'dirty' and 'scruffy'; 'dirty' may be used to denote the distasteful
pcrception oñ' category membership has similar general goals to the theories and uncivilized, while 'scruffy' may refer to a laid back, streetwise

1
1

124 Discourse and social psychology Categories in d,iscourse 125


!

FIGURE 6.2 hard-working people'; 'lack of greed' and 'quite selfiJh and greedy'
Four spcukers' descriplions of Maoris (see Figure 6.3). For just about every statement made a~out Maoris an
opposing or contradictory statement was aiso offered son~ewhere clse in
Speaker Om• Music lovely the material. 1

Indcpendent Use same sort of expression


Rcsponsibility in family gane Friendly 1

Mature
Worldly FIGURE 6.3 1

Lost their culture Speaker Four Inconsistencies between respondent's descriptions of Maoris
Stirring Feel disndvnntnged
Like to congregate Slow Lnzy rnce Such hard-working pcoJle
Lovcly ethnic fecling for each other Feel they have no self-estecm They're proud Lost their pridc and digpity
Like driving Need to tnke a positivc vicw of their lifc Proud of nll their background Humble 1
Like playing with machinery Must look at selves and decide strengths Wnnt to split this society Pnssive ¡

Tend to go for jobs which are manual and weaknesses They renlly know how lo puta story They come ncross as clÚclcss thcy don't
1

Make good soldiers Fecling money will put disadvantages across, lo gel the message ncross, know how lo express their views
Haven't got sticknbility right are orntors J
What you've got is theirs Feelings inside all wrong Grenl respect of older folks Vcry good attaking udJantage of
They stea\ Simply need nol get together Lhdr old
Get cracking Less mnterinlistic and more diverting in Quite selfish nnd greed l
Speaker Two All work toge~her their culture
Pnssivc Bchnving in unacceptable way Rclaxcd fricndly munner Jll al case
Not interested in language reviva\ Wider family is importnnt to them
1

Don't take advantage of cducation Balance good feelings: body, spirit, soul
Diny Not observing own standards of hygiene Can prototype theory be sustained? 1

Scruffy Not observing own cultural belicfs


Evil-Jooking Grent difficulty opening up their heum,
One way of responding to findings of this kind might be to suggest that
1

Hot-hends tellíng how they feel


lsn't a leader Dcmnnding
different participants were drawing upon different prototypes. Variable
Expecting accounts, then, would merely be a consequence of pebpie's variable
Speaker 11zree Need a bit more insight individual prototypes. Hawever, the difficulty here is thtit variatian was
Proud Need more exercise and self control pervasive and extreme e ven within the tal k of the same ptirticipants. One
Wonderful Marvellous respondent, for instance, described Maoris as having 'aÓ innate under-
Don't want to share thcir culture Very good standing of their land' and claimed shortly a[ter that 't~eir land is just
Busy recstablishing themselvcs Courtcous
a bit of dirt to them'. 1

Trendy Smil e
No motivation Welcome Another type of response to this inconsistency might 8e to argue lhat
Takes cure of her children Friendly it is a consequence of the different levels of categoriz~tion operating
Friendly relaxed manner Hnppy (Brewer et al., 1981). In this view, contradictions are produced by people
moving between different subprototypes of Maoris - Maori children,
Maori parents, Maori leaders etc. Thus inconsistency Ónly appears if
demeanor which could be essentially positive. These significant nuances the researcher assumes participants are working with th1e broad super-
would be lost if researchers simply aggregated terms. Classifications of ordinate: Maori. 1

this k.ind should not come from the researcher's commonsense semantics, The first thing to note about this view is the danger of m~king the theory
but from a detailed analysis of participants' own use of discourse. so flexible and post hoc that it can predict virtual! y anyt~ing. We reach
E ven gross categorization of this kind would be unable to handle much the point where a prototype is posited to explain every single variation
of the variation found in this material. It is not easy to find sensible in accounts of Maoris, and the theory beco mes emplrically empty.
classifications which encompass contradictory adjectives such as 'they're However, even if we remain with the limited splitting into subtypes
proud' and 'they've lost their pride and dignity'; 'lazy race' and 'such indicated above, the problem of variability returns to ha~nt the theory.

1
1

126 Discoutse and social psychology Categories in discourse 12 7

When our data were analysed in terms of the sorts of subtypes where just how radically his view of categories differed from the standard social
prototype thedrists might expect consistency we again found variability. psychological formulation.
In our respondents' discourse Maori parents, for example, are both 'very
fond of theil children' and they 'aren't interested in their kids';
Maori leaders! are 'oftcn too radical' and 'you don't scc thcm agitating'. Hotrodders and teenage drivers
Of course il might wcll be possible for a prototype theorist to do sorne
judicious splitting into sub-subcategories, and to further emphasize Sacks starts with the question of why American kids generate so many
individual diffcrences,
'
and thereby provide sorne sort of account of these new typologies of cars. Why, he asks, are the existing terms not good
data. But this is going to get very clumsy indeed. Our claim is that discourse enough? Sacks notes how different category terms are enmeshed in, and
'
analysts are able - -more workable explanation of the
to provide a much can be used to exemplify, different world views. For a contemporary
variations wh!ich occur in participant's discourse. By highlighting the example we only ha veto think of 'gay' and 'faggot' as competing descrip-
functions of \alk we can explain why a category of people should be tive tenns for 'homosexuals' (\Vatson and Weinberg, 1982). The example
described in dne way on one occasion and in a different way on another. Sacks uses are the competing descriptions, 'teenage driver' and 'hotrodder'
Rather than dbmonstrate this in detail with the above data (see Wetherell (remember, this is 1965). The former term is used by adults and reflects
et al., 1986, rbr examples) we will turn to ethnomethodologically inspired an adult view of teenagers who drive cars. The latter term, however,
work on socia1 categories, which was the first loo k at social categorization reflects an alternative, non-adult, world view; a world which revolves
from this gen'eral perspective. around customizing cars and racing them in the street, much to the dis-
pleasure of the police.
Sacks illustrates the way these competing descriptions are managed using
Motives and mernbership categories a piece of recorded dialogue. To understand this exchange we need to
know that a Bonneville is the sort of souped-up car driven by a 'hotrodder',
Givcn that cthnomcthodology's central concern is with the way pcople while a Pontiac station wagon is an everyday, straight from Detroit, motor,
make sense dr their world - their ordinary reasoning - and given that the sort that might be driven by a 'teenager'. The categories of cars are
ethnomethoddlogists take a strongly functional view of talk, it is not linked to the categories of people. (The transcription is slighly simplified
surprising that they took an carly interest in people's social categorizations. and edited from the original.)
For these rese~rchers, this was an arca of discourse fundamentally in volved
l. Ken: In that Bonneville of mine, 1 could tnke thnt thing out, an' if !'ve gana
in the way orlctinary people make sense of social structure and provide
tie, nnd a sweater on, an' 1 Iook clean nincty nine percent ofthe time a guy
cohcrcnce lo their social worlds. At the same time, the ethnomethodological
could pull up tome, thc samc car, samc colour, same year the whole bit,
conccrn with lthe active accomplishment of social phenomena and inter-
action sensiti~ed them to the possibility that categories might be more than roar up on his pipes, and he's innn dirty grubby tcc shirt, an' the guy
[policeman] will pick thc guy up in the dirty grubby tee shirt befare he'JJ pick
simplifying perceptual sunglasses but deliberate constructions fitted for
me up.
many tasks. ¡'
(2.0)
Harold Garfinkel himself too k this approach in a detailed and fascinating
( ): hheh
case study oriAgnes', a would be transsexual (Garfinkel. 1967). Garfinkel Ken: J'st just for uh
adopted the +ew that membership of the gender catcgories 'male' and Al: (But) not many people gct picked up in a Pontiac statio~
'female' is not merely a factual matter of biology but, in social psycho- wagon.
logical terms) negotiated achievement. The study documents the complex (Sacks, 1979: 7)
and subtle procedures Agnes had to learn so that shelhe could successfully
1

bring herselflhimself
' off as a woman. Ken is claiming that he can go out, smartly dressed, in what we learn from -::e
Harvey Sa¿ks - whose work we introduced in Chapter Four - took Al 's interruption is his, in fact, non-customized Pontiac, get into drag races
the study of dltegories in another direction. He was particular! y concerned and then - because of his clothes and car not get picked up by the
with the kin~s of inferences people rnake when category terms are police. Sacks suggests that Ken is seen to be trying to ha ve his cake and
used and, as la consequence of this, the ends categories work towards. eat it. He is claiming to be a hotrodder; while at the same time remaining
In a paper edited from a lecture given in 1965 (Sacks, 1979) we can see laya! to the safe adult version ofthe world. Hence Al 1 s sarcastic deflation:

1
128 Discourse and social psychology Categories in discourse 129
1

why would the poli ce pick someone up in a Pontiac like that? One reason for the category 'father' or the category 'activist' has already Uppeared in the
• 1
generating typologies of cars, then, is tomark alternati ve social categories. d IScourse. 1

That is, the selection from the category of cars 'Bonneville'/'Pontiac' Sacks further points out that certain categories have¡ what he calls
displays membership ofthe category 'hotrodder' versus 'teenager'. Al has 'duplicativo organization'. What he means by this is that th¡; category is not
picked Ken up on, in effect, an illegitimate claim to be a member of the merely a boundary toan independent set of individuals; rkther, the indi-
category 'hotrodder'. What we are seeing here, according to Sacks, is the viduals may also be seen having sorne kind of interperson~l organization.
policing of a category and, ultimately, the preservation of an alternative We can contrast categories such as 'family' and 'gang' which are duplica-
version of the social world (see al so Cuff and Hustler, 1982). tively organized and those such as 'redheads' or 'Scots' which may not be.
The former may ha ve 'in-group loyalty' and 'stick together' L the latter may
not have these features (Watson, 1978). It is important tolnote that ínter-
Membership categorizarion and ordinary reasoning personal features such as in-group loyalty are occasioned rather than
necessary. That is, they are available to be established in ¡the course of a
In his later work, Sacks (1972, 1974) went on to pay detailed attention lo particular account rather than intrinsically linked to thc c~tegories.
people's use of categories in their ordinary reasoning. On this tapie, his The overall point is that people draw upon knowledge of the organization
best-known contribution concerned membership categories and categoriza- of categories as a resource for producing economical and ihtelligible con-
tion devices. Although this began with sorne ofthe most rnundane features versation. As well as relying on this kind of knowledge, Sacks
'
notes that
of our category discourse, its airo was to use these to illuminate the details people also draw on conventional knowledge about the acrivi(ies of members
and subtleties of complex social acts, such as blamings and mitigations. of categories. For example, one of the things we conventionally expect
A membership category is simply a commonsense equivalence class for members ofthe category 'baby' todo is cry. Crying can th~s be seen asan
the identification ofpeople in talk, e.g. 'teenagers', 'Maoris'. A member- activity bound to the category 'baby'. Likewise, giving mbdical help and
ship categorization device groups together membership categories. Thus writing prescriptions are category-bound activities belongidg to doctors. In
'family' is a membership categorization device (MCD) which groups fact, as Watson (1978) and Jayyusi (1984) have noted, it is rlotjust activities
together 'brother', 'mother', 'baby' and so on while the MCD 'gender' which are conventionally bound to catcgories; a wholc ciJster of fea tu res
groups together the categories •mal e' and 'female'. may be expccted of categories: the traits and prefcrences ofÍhc incumbcnts,
It might seem that this is a rather clumsy way of referring to the fact, where they live, what they look like, what they wear abd so on.
noted earlier in our discussion of prototypes, that categories may be ordered The fact that membership categories can be conventiorÍally tied to, or
into hierarchies. However, the use ofthe terrn 'device' reflects the ethno- associated with, specific activities and other features, provitles people with
methodological emphasis on the constmctive nature of tal k; the collection a powerful resource for making sense oftheir social world~. In particular,
of a· set of categories together and the creation of an MCD are active
1

it allows them to make inferences, or discursive connections to the category


accomplishments. Categories are not nested in a clear-cut, natural way like membership of the actors. And, conversely, given that they know only
Russian dolls, for example, one inside another, but are grouped into a person's category membership, they can make a good ~uess as lo the
collections by the use of potentially complex and contradictory sets of kinds of things that person is likely to be doing. If someone is writing
interpretative procedures. We shall return to these differences between out a prescription we expect them to be a doctor; if thefe is crying we
ethnomethodological and social cognition approaches to categorization may loo k around for the baby. Put another way, if we hearla doctor crying
later. or see a baby writing a prescription we will expect an account of the special
Sacks suggests that there are certain basic practices which people draw on circumstances which led to this aberrant behaviour. 1

in their everyday discourse concerning categories. For example, there is Now of course, these features of categorization ha ve al so becn identu-ied
a maxim of 'consistency' which states that if ane category from a MCD by social psychologists. Tajfel (1981), for instance, talks fbout inductive
, ·;~· has been used it may well be appropriate to use another from that MCD. category errors where people mistakenly work from somr features of an
"'\1 For the hearer of the discourse this mcans that íf two categorizations which individual, ignoring other features, in arder to categorizc them. Neverthe-
, ··)' could be from the same MCD (but are nol necessarily so) are used, they less a very different perspective underlies the similar ~oints made by
2.') will be heard as coming from the same MCD. For example, the categoriza- ethnomelhodologists and social psychologists. This differe~ce will become
<'. _ tion 'sister' could appear in the MCD 'family' and the MCD 'feminists'. more and more obvious as we procecd to research exarhpics.
~' How it will be interpreted will depend in part on, for instance, whether Thc basic ethnomethodological insights ha ve provided lsome powerful

1
¡

130 Discou~se and social psychology Categories in discourse 131


1

tools for revJaling sorne of the details of social phenomena such as as from a single MCD. Watson suggests that this complex ofcategorizations,
blamings, acc~sations, and mitigations (Drew, 1978; Jayyusi, 1984; Lee, which both follows and fiouts the consistency rule, is carefully constructed
1984; Watsonl1978, 1983; Watson and Weinberg, 1982; Wowk, 1984). to produce a particular effect.
To illustrate tiow they can be used, we will take examp1es from studies The category 'sister' is derived from the MCD 'family' which, as we
by Rod Watsoh and Maria Wowk on the role of categorizations when b1ame noted above, is generally understood as duplicatively organized; it groups
is being al1olted in a murder investigation. together categories of people who are then seen to ha ve a variety of in ter-
personal bonds. The (black) suspect's categorization of the victim as a
black sister thus stresses his co-incumbency of the category 'black' and
Race, genderland blame trades on the affiliative, team-like properties of that co-incumbency.
Watson's point is that the categorizations pull together the suspect and
One of the at ractions of studying murder confessions and police inter- victim while contrasting the lictim with those she purported1y goes to bed
rogations of iuspects is that they bring issues of cu1pability and motive with ('white men' note, not 'white brothers'). This construction provides
to the forefroht of discourse. In general, suspects know that certain sorts the motive. The impassioned murder of someone who has dragged her
of motivation~ may be seen as more culpable than others, and that a range whole family - even if it is only a family in metaphorical terms - into
of circumstandes may mak.e them appear less blameworthy and more under- the núre of disrepute is so much more understandable than the cold-blooded
standab1e. Thbs when suspects construct descriptions of their crimes they murder of a young woman where the only motive is the theft of a small
do 'motive wbrk' (Milis, 1940). That is, they construct a version of the quantity of money.
crime and wh~t led them to commit it in such a way that the act is mitigated While in Watson's example the motive work is done more or less direct1y
or evenjustified (Scott and Lyman, 1968). In British and American legal by the use of explicit categorizations, the categorization in the murder
systems punishment is notan automa tic consequence of the sheer behaviour confession analysed by Maria Wowk must be inferred. The confession
of killing so~eone; circumstances, beliefs and motives are all taken into contains what Watson and Weinberg (1982) have described as a 'category
account and Punishment varied accordingly. puzzle'; a set of behaviours is described and the listener has the task of
In this context, categorization is a resource which suspects frequently making sense of them. The problem they are faced with is: what kind of
category member would act in this way? Here are sorne extracts from
~:~~c~nb~~o~f~Pf~; :xuac;;::. ~~:s~~p~~~;:a:~a~~~~r: aacr~ ~~tle~s~k~h;~:
0 0
the confes.sion.
plausible molivations for the murder of a black female by the black maJe
who is doingl the speaking: financia!, sexual and racial; and he suggests 3. Suspect. 1got to the interscction (.2) ofBrookland and Slade (1.0) when this
that this pass~ge of tal k works to preserve the racial motive and disclaim girl walked up to me (.6) and propositioned me
the financiallone. (l. O)
1
Policeman. what did she cxactly say to you Lewis?
Suspect. you loo k likc a tough guy (1.2) y'look like the member of a gang
2. Thcn hJ lUrncd off the ignition thcn thc one in the b.uck he snid "nlright (1.2) 1 told hcrc I'm nota mcmber of u gang (.) I'm an indcpendent (.7)
who dojng whoring nround out hcrc?" Then [thc victim] said well if you and she propositioncd me again
wnnt mbncy, 1 don't hnvc money so the tnccomplicc] in thc bnck said we [l
don't w~nt your moncy [ ] nnd thcn he snid "wc disgusted him evcry time Policeman. what did shc ask actually say lo you
onc of his black .risters gocs to bcd with white men nnd wns going lo pul [l
a stop tb it." He nsked [the victim] again why she went to bed with thc Sw.pect. she askcd me if I would likc to get laid
white m~n and she told him she had three kids and he said [she] was a God [latcr in confession]
damn lihr and punched her in thc face and about thc body a couplc of times. Suspecr. urh (1.4) the gir1 got (.8) might say kind of (.2) pricky (1.0) and er::
(Watso6, 1983: 50, emphasis added) [l
she propositioned me again (.6) and then she called me a prick hh (.) a
1
no good sonofabitch {.) hhh and she threw what was left at (.2) the remainder
In this pah of the interview the suspect deploys two membership
(.) of the bottlc of beer at me
categorizatidns: 'black sister' and 'white men'. In terms of Sacks' [l
Crom '"' Mr
consistency fute, we can clearly see the predicates 'black' and 'whit~' as
'moe'' Oowom, ;; ;, oo; =•
m o;ow •;;;~· ~· ·~·
al that (1.2) 1 threw al(.) a right handed punch (.2) from the waist towards
her shou1der. .. [goes on to describe killing] (Wowk, 1984: 80-1)
:

132 Discourse and social psychology Categories in di.~course 133


1

As we noted above, one pawerful resource used in everyday reasoning is For the social cognition researcher the prototype determihes perception
the idea ofthe category-boundedness of activity. People conventionally make through the operalion of a feature matching process; if the s timulus shares
1

inferences from categories to the activities of incumbents and, conversely, the prerequisite number of features with the prototype it will be included
from activities to the category membership of actors. In this extract we are i~ the category. For those who study the operation ofl categories in
presented with both a category- 'a girl'- and activities- propositioning dtscourse, however, the cluster of calegory-based attributes is not viewed
the suspect, calling him a 'prick' anda 'no good sonofabitch', throwing beer as a kind of mental picture influencing perception but a~ a potentially
at him etc. The puzzle for the listeners is to provide a categorization which inconsistent cluster of expectations and associations dr~wn on in an
can account for these activities in a way that 'girl' fails todo. occasioned manner. The point is that these expectations Áre selectively
Wowk suggests thal, given the described activities, the conventional managed, in the context ofpassages of tal k and writing, to bresent certain
puzzle solution will be to further speci[y thc category membership of the effects (Drew, 1978; Jayyusi, 1984). Thus so me featurcs/ may be made
girl as a 'tramp', 'slut' or even 'hooker', all categories which have strongly focal, others ignored, and yet others reconstructed. An understanding of
derogatory implications. These provide a negative moral assessment in the effectiveness of categorizations will thus depend on b~th elucidating
the way 'girl' does nol. Yet, as they are not offered explicitly by thc thc relevant cluster of attributc¡; and thc dctail ar thcir clcployment in '--
-:0 suspect, he can make a negative moral assessment of the victim available, discourse. We will illustrate how this can be done by lookirlg ata detailed
while maintaining a position which allows him to deny that he views lhe study of how the categorization 'community' was uscd in accounts of a
victim as a slut. Nevertheless, reapportioning sorne of the blame in this 'riot' (Potler and Reichcr, 1987; see also Halliday and Potter, 1986;
way is exactly the upshot of the suspect's account. Reicher and Potter, 1985).
In both of these analyses, then, we see a very different picture of the
way categories operate from the one offered by social cognition researchcrs
such as Wilder and Cantor. Rather lhan viewing the process of categoriza- Uses of 'conummity'
tion as a defect in perception - albeit an adaptive one - these researchers 1

have demonstrated how categorizations can be carefully inserted into In the spring of 1980 an evcnt occurred in Bristol which ca~e to be known
discourse as part of the joint activities of blaming (lhe victims) and as the 'St Paul's riot'. Fighting look place between policc!and youths on
mitigation, and thus how categories can be read. lhe streets of the St Paul's area of Bristol over a period df sorne hours.
Several police vehicles and properties were destroyed, a nulnber of police
and civilians were injured - although none seriously - !anct the event
Categories, content and community dominated the next day's natíonal news coverage and ~esulted in an
emergency debate in Parliament. This seemed an ideal oppdrtunity lo look
One of the features of the tradilional social psychological approach to at the way social categories are used in the representatidn of conflict.
categorization is its tendency to treat categories as if they were distorting The study worked with an archive of documents: copies lof reports and
enclosures anchored to unitary consistent prototypes. Their damaging social edilorials from the local and national press, transcripts of t6levision news
consequences are said to arise frorn effects such as the over-inclusion of and current affairs programmes, records of parliamentarY proceedings,
people into the category, and the possibly stereotyped nature of the official reports and lranscripts of interviews with people whÓ were actively
prototype (Petligrew et al., 1958; Tajfel, 1969). inv~lved in the events. . . 1

The studies examined in the last section reveal a rather different set of Smce much of the Ume the dtscourse of confltct was addressed to defining
categorization phenomena. They display categorizations used not so much the nature of the protagonists, it was an especially rich cohtext in which
for their 'boundary work', grouping and separating individuals, but forthe to study people's use of categorizations, and in particular their use of the
category-based inferences made available. Watson demonstrated in Extract categorization 'community '. 1

Two how the suspect's use of'white men' and 'black sister' was not simply The analysis was initiated by selecting out from the archi~e all instances
an exercise in inclusive grouping (although it inevitably achieved this) but af the term 'community' and its synonyms, and went on to identify the
the groundwork for an imputation of motive. Although, on the face of different predicales used with the term, along with the varyipg ways il was
it, the idea of categories carrying a cluster of category-based attributes is adopted to refer to the protagonists in the events. It was imrnediately apparenl
similar to the notion of categories anchored to a prototype with a limited set that certain sorts ofpredicates were repeatedly used, in partiJular predicates
of features, there is a crucial difference between these two models. describing a certaín cohesive style of social relationshi~: 'closcness',

1
!

134 Discourle and social psychology Categories in discourse 135

'intcgration' aJ 'friendliness', and lhosc associated wilh certain metaphors: problem and another group of accounts which treated them as a direct
spatial ('closejknit'), organism ('growth') and agency (a communily conflict between lhe 'community' and the police, Contras! the following
'acts')- see Figure 6.4. Without exception, where the term 'community' passages which illustrate these two versions. The first is a statement by
was used with ¡a strongly evaluative force it was positive: 'community' the then Home Secretary, William Whitelaw, made in the parliamentary
was seen as a .good thing. debate which followed the disturbance; the second is from lhe beginning
of a reporl in the newspaper Socialist C/zallenge.
;
1

1 FIGURE 6.4 4. Whitelaw. As to what the Right Honourable Gentleman says about the
Predicates and mctaphors used with 'community' community relations work ofthe police, it does so happen that a police officer
in this particular area ofBristol, as I understand, has been very active indeed
Samplc predicntcs Metuphors in the area of Community Council. He has been one of the most highly
(wherc rclcvunt) rcspccted mcmbers of thut council who hnve done a great den! of the sort
of work that the Right Honourable Gentleman has in mind.
Fricndly 5. ON WEDNESDAY 2nd April, the mainly black population of St. Paul's,
Wnrm a Bristol inner-city district, responded to police harrassment by mounting
Happy a counter attack. The police are the most visible instrument of state repression
Harmonious of the black minority in Britain toda y. Almost exactly ayear after the poi ice
Closc-knit Sputial riot in Southall, the black community of St. Paul's fought back against police
lntcgratcd
brutality and won.
Tight
Extract Four was typical of one kind of interpretation of the disturbance.
Grows Orgunic
Whitelaw was responding toa question about 'community policing'. He
Evolvcs
Malures formulatcd his reply in lhe ralher broader terms of lhe 'community relations
work of lhe poli ce' and claimed that this work airead y existed, citing an
Acts Agcncy
active member of lhe 'Community Council'. The effect is to include
Knows
prob!ems of policing under the tapie of 'community relations' - which

l
Fccls

\p-
. he had earlier characterized as gbod - and to support this characterization
wilh strongly positive, but at the same time very vague, allusions to the
work of a 'community po!ice officer'.
Throughout lhe parliamentary debate the 'riot' was described in lhis
Despile the broad agreemenl in the atlribute cluster connected with the
' \ \J
category 'comlnunity', considerable variability was evident in lhe applica- manner as a problem of 'community relations' and lhe police were depicted
''
tion of the tefm. In sorne accounts, for example, the 'community' had as parl of the 'community'; thus preparing lhe way for a response to the
been 'disrupte1d' or even 'finished' by the 'riot'; in others the 'riot' was events which follows from lhe cluster of features bound to the category
mere! y a sigo br the cohesiveness of the 'community'. In sorne accounts 'community'. If a 'community' had been disturbed lhe problem will be
the poi ice wer~ depicted as part of the 'community'; in others the disturb- one of fractured interpersonal relationships and trust. Thus the proposed
ance was a cohllict between the 'community' and the policc. Variability response is directcd at these points. It includes further representation of
on this scale h<Ís now become a very familiar phenomenon. Yet it represents police on community bodies as in Extract Four, the development of
an enonnous headache for any realist approach to participants' discourse, personal relationships and trust between police and locals, and a construal
and particular}y for those researchers bent on reconstructing exactly what of the contlict in 'human terms'. In general, the use of 'community' in
wenl on, bet~een whom, in the disturbance. It is for this reason that this way leads inexorably to the introduction of 'communitY, .•Policing': lrp¡-
1
research genetully uses procedures which render variability invisible.' But; the perfect solution to 'community relations problems'. ~/"J,'e- 1 • ,.
if wc look to the action orientation of the accounts, differences become IP-.; Extract Five is very different. Instead of fonnulating the event as a ri~t r:,..,..c,·

l"'"' "' '""" '""' "'.'"•' " " " ' "'"'" '•
10
much more uhderstandable. triggered by ailing 'community relations' it is depicted asan open conflict,¡-, -
One majar :difference in the data occurred between a broad group of With the two sirles explicitly characterized as the 'blnck community' and '&c.,.J1(>~

'~
the poi ice. Furthermore, a clear-cut and straightforward causal story of
"'"""" w"
1

136 Discourse and social psychology Categories in 'discourse 13 7


11' .q-TEGo n t ~ .+· 9 1,: ; t <-: (.'t :~:­
the conflict is offered: the 'blackcommunity' was responding to 'harassment' the expectations associated with a category is worked up in passages of
and 'brutality', so the conflict is a natural part of an ongoing illlergroup discourse. This is not to say that prior expectations and uh.derstanding are
struggle. not important; rather that these are drawn on as a resoJrce from which
In accounts of this kind, the cluster of category-bound attributcs and the detailed sense and implication of the category account ls manufactured.
the positive force of 'comrnunity' contribute three consequences. First, We will elaborate on this idea in Chapter Seven when J.-e introduce the
the police are implicitly blamed; for 'communities' are not reasonable notion of linguistic repcrtoircs. 1

r{,_·: targets for police attack (unlikc, say, 'thugs' or 'crazed rioters' - see Trew, Thc suggestion that categories have a fundamental role in biasing
1,, •''" 1979a). Second, actions against the poi ice are legitimated. As 'com- perception is replaced, in thc discourse analytic work, ríth thc idea of
b~ munities' ha ve the category-bound attribute ofharmonious personal rela- categorics being actively constructed and drawn on for many differelll
tionships, if they become involved in violence it will most likely be secn actions: bias and tolerance effects included. It may well bd that categoriza-
as extemally caused andjustifiable. Third, any poten tia! dismissal of people tions are often associated with phenomena such as ov~r-inclusion and
fighting with police as marginal or pathological is undermined; depicting simplification; howe~er, research on categoríes in discdurse has shown
them as 'the community' they are made central and indced representntive that categorization is vcry far from thc mcchanical cÓgnitive process
ofthe area. Versions ofthis kind lead toa distinct, alternative set of solu- implied by traditional work. Many different studies Have shown that
tions. Instead of 'community policing' being the answer the curbing of categories are selectcd and formulated in such a way tHat their specific
a repressive and racist police force is proposed. 11 ;\ \(1 2.... features help accomplish certain goals. This accompli~hment does not
"T
depend merely on the homogenizing properties of cateSarization, thcir
ability to group a set of stimuli, but on features specific Ita the category.
Categories and the construction of discourse For example, we saw how accounts could use descriptions ,of actions bound
to the category 'tramp' in the course of a blaming, or rJatures bound to
Earlier in the chapter we identified three kinds of assumptions which are the category 'community' to provide certain actions with '
legitimacy.
perenially made in traditional social psychological work on categories and One of the benefits of the discourse approach to categorization is that
which become questionable when a discourse-orientated perspective is it has directed attention away from the cognitive proceSses assumed to ,,,
adopted. These were that categories are prefonned and enduring, they be operating under people's skulls and on towards thb detail of how
have a fixed structure and inevitably lead to biased perception. Having categories are actually used. The study of categories hnfolds into the
explored sorne of the research on categories in discourse we are now in general study of the organization of discourse and its consbquences (Billig,
a position to see the problems with these assumptions more clearly. 1985, 1987). It is not surprising that categories are so im/JOrtant, because
The idea that categories are enduring, preformed entities is hard to they are the nouns from which we construct versions of !he collectivities
maintain in the light of widespread variation in the content of ciuegory in which we live. In a sense, they are the building b!odks of our many
accounts (as well as papers cited above, see Gilbert and Mulkay, 1984: versions ofthe social world; however, once we loo k closbly at the blocks
ch. 5; Yearley, 1984b). This variation and the active reconstruction of we see that they themselves are not salid and defined,l but have to be
categories cannot be easily reconciled with the popular cognitive models moulded in discourse for use in different accounts.
in this area, especially Eleanor Rosch's theory of prototypes, but is
perfectly compatible with discourse theory which sees variation as thrown
up in the wake of action.
Likewise. the concept of categories structured as fuzzy sets anchored
to prototypes and organized in hierarchies differs markedly from the notion {Ueo )
of structure revealed in discourse analyses. As we saw in the previous
section. the relatively static exemplar or prototype is replaced with the
idea of a cluster of potentially inconsistent features and expectations. It
would be wrong to think of these expectations as constituting a clear-cut.
prestructured, enduring list. Nor should our tal k of the 'function' of
category accounts be taken to mean that categorizations in rhemselves
notably lead to certain consequences. The relevance and detailed nature of
From representarions ro repertoires 139

of the chapter will then introduce a body of research using the notion of
interpretative repertoire which, although more modest in its goals, has,
7 we feel, a number of advantages in comparison with Moscovici's approach.
We hope to be able to demonstrate these advantages.
From representations to repertoires
!
The theory of social representations
One of the maij aims of social psychology is the identification of broad
patterns in hunian behaviour. Social psychologists have nol just been Social representations is not an easy theory to describe. Moscovici's
interested in individual actions, but in the way these actions are fitted iota writings are fragmented and sometimes contradictory (McKinlay and
episodes, systerhs or sequences (Clarke, 1983; von Cranach and Harré Potter, forthcoming) and researchers have tended to interpret lhe lheory in
1982; Forgas, l979). Recently, one particular theory with this kind of widely divergent ways (Parker, forthcoming). It has been used to examine
integrative goall has attracted an enormous amount of interest in social a range of social phenomena: for example, alliances in a student protest
psychology- Moscovici's (1976, 1981, 1982, 1984a) theory of social movement (Di Giacomo, 1980), explanations ofroad accidents (Barjonet,
representations.¡ Social representations are seen as mental schema~a or 1980), the design of experiments (Farr, 1984), and in two of lhe most
images which people use to make sense of lhe world and to commumcate developed projects, people's representations ofpsychoanalysis (Moscovici,
wilh each olher.llt has been suggested lhat lhis concept offers a new frame- 1976) and health and illness (Herzlich, 1973). In the following exposition
work for understanding the organization of attitudes, beliefs and attributions -( we will stay as clase as possible to Moscovici's various theoretical
and will be abl~ to provide a principled criterion for distinguishing the j)o\'.1 slatements (1981, 1982, 1984a, 1985).
members of dif;ferent social groups.
Because the theory of social representations has been so important
to the new styl~s of research emerging in European social psychology Representation and object
and because, irÍ certain respects, it covers the same kind of ground as
discourse analy~is, we will examine it in detail in this chapter. This will Social representatians are mental entittes. They are made up of both
give us an oppÓrtunity to demonstrate how discourse analysis can fulfil abstract and concrete elements (concepts and images, respective!y). In each
the need for mofe integrative perspectives, developing broader units lhan representation these elements ha ve a specific structure. For most represen-
those traditionálly used by linguists (phoneme, word, sentence) and tations the concrete elements or images are the most important; in these
conversational analysts (turn, adjacency pair, closing) to make sense
1
cases the representation will be built around what Moscovici calls a¡,
of social life.1 . . . .
'figurative nucleus'. Social representations provide, quite literally, the ·
One unit which has been explored m a number of d¡scourse stud1es IS
1
means for people to understand and evaluate lheir worlds. For Moscovici,
lhe interpreta ti V~ repertoire (Gilbert and Mulkay, 1984; Potter and Mulkay, all thought and understanding is based on their working.
1982, 1985; Wptherell, 1986). The interpretative repertoire is ba.sically For example, in traditional public opinion polling people are often asked
a Jexicon or register of tenns and metaphors drawn upon to charactenze and questions about, among other things, political parties. People give their
evaluate actions! and events. Por example, the categorization 'community', opinion about the party; so we ha ve, on lhe one hand, the party and, on
at which we lobked in Chapter Six, is achieved by the use of a cluster the other, the psychological entity: the opinion. In contrast to this, social
of terms and rhetaphors which are selectively put forward to provide representations theory propases a rather more complicated threefold model.
evaluative versions of the events taking place in a 'riot'. In this sense, First, there is lhe political party; second, there is the person's social
we can identiij a small-scale interpretative repertoire: the community representation of that party; third, there is the person's opinion, which
repertoire (Pot!er and Reicher, 1987). is derived from the representation.
This chapter \vm, lhus, contrast social representations and interpretative The person's representation ofthe British Conservative Party might be
repertoires. wel will first overview lhe lheory of social representations, and made up of specific abstract elernents, e.g. 1 a previously humanitarian
point toa numb~r of problems wilh bolh its coherence (McKinlay and Potter, grouping that has become authoritarian and racist', and concrete elements,
forthcoming) tind the way its principal concepts have been applied in an image, for instance, of a broad grouping of 'moderates' who ha ve been
research (Litdn and Potter, 1985; Potter and Litton, 1985). The rest 'penetrated' by a powerful group of 'extremists' from the 'far right'. This

11
!

140 Discourse and social psychology From representations 10 )epertoires 141


·'~ ',
~:;,)1\tversion ofthe party is clearly nota mere neutral description, but one which for conversation. On the other hand, disagreement in reJresentations will
~l \l'' provides a negati~e evaluation. If this person was asked by a pollster ]ead to conflict, dispute and misunderstanding; the convJrsationalists will
(~ which way they m1ght vote m a general elecuon they m1ght say they used by ~ty_l}lk.Q_ in their attempt to communicate. 1 ~-+ti\T¡ r1 :¡>-·t7-J< 1Z: 1 r
ú' to vote Conservative but no longer. Moscovici's important point is that The third sense in which representatians are social i~ their pravision

\
~ to understand why the persa o has offered this particular opinion we need of a theoretically coherent way of distinguishing betweén social groups.
\_10 understfi.ñi:Rhe person 's social representations (concepts and images) Because social representations supply a conventianal codé. for communica-
,_·\ \of political parties (as well as their associated representations of the tion, and because they are the central dynamic for under~tanding, al! who
1 share a representation will agree in their understandink and evaluation
,demacra tic process, the workings of the economy and so on).
· Social representations are also assumed to underpin attributians or a
of an aspect of the world. The representarían will thus be crucial unifying
the§msal explanat\~ns people give for events. We can see why if we and homogenizing force. In Moscovici's theory, then, wHat makes a group
carry on with the po 1tical party example. If the person with the represen- a group is exactly the sharing of representations among mJmbers; the edges
tation of the Conservative described above is asked to explain why the of representations will mark the edges of groups. 1

party supported a particular piece of racist immigration legislation his In terms of aur exarnple, the representation of the Conserva ti ve Party
or her explanation may centre on the malign influence of penetration as penetrated by right-wing extremists will not be an individual
'
construction
from the far right; racist legislation is not an expected and intrinsic part but the outcome of many conversatians about the Party, i4eas disseminated
of Conservative behaviour, for this person it is introduced from the outside. through the media and so on. As the representation is generated and
Yet another person, with a different representation of the party, may becomes more established it will become easier and eaiier to talk about
well attribute the legislation to the Party itself, acting naturally and the Party in these terms, and indeed, to presuppose th)s '
representation .
normally, in terms of its usual ideology. The representation thus provides as unavowedly literal and factual. The people who share this representationf __
a model of the causal processes speakers might use in explanations. are, in an important sense, a social group: they will und~rstand, evaluatej' C\.~.:
and ultimately act in the same way. 1

Wlzy social?
Construction, anchoring and ohjectification
From the foregoing it might seem that the social in social representations
is merely redundant baggage, and that representations are individual Social representations are said to ha ve a crucial constructive effect. They
phenomena -:- we each carry around our own unique versions of the a
do not mere! y nzediate between objects and belief5.Whe'n group makes
world, including representations of political parties and so on. In this sense of its world, that world will be constructed by, knd in terms of,
view, the theory could just become part of individual cognitive psychology social representations. Experience of poli ti cal parties, saY 1 is constrained
incidentally used to explain and understand social phenomena. However 1 by social representations. If the Ieader of the Conserva ti veS acts in a certain
Moscovici is keen to deflect this interpretation. For himl social represen- way the meaning ofher actions will be constructed by pedp!e's representa-
tations are irreducibly social in at least three senses. tions of the Party and politics in general. 1

FfrSl,--sOCiáf ¡:epreSérttations·-are- intrinsically linked to communication Mascovici also propases a mechanism for people's rnethods of coping
processes, and in particular to people's unstructured everyday talk: with new and un familiar experiences. Familiar experiencb can, of course,
their gossip, chat, pub arguments and family discussions. As Moscovici 1
be simply dealt with in terms of a person's existing stotehouse of social
puts it 1 'representations are the outcome of an unceasing _babble and a .6/./b~\t representations. The unfamiliar is more problema tic. Mdscovici suggests
permanent ·ctraiagÚe.betweeh individuzils' (1985: 95): so soéial.repfeseriia- ·~li,' that novel ar strange objects are dealt with in two sfuges, known as
tions" are social in the sense that'they originare in the course of social 'anchoring' and 1 0bjectification'. 1

interaction. In the first stage - anchoring - the novel object is assigned to one of
Second, they are social because they provide an agreed code for the categaries ofthought, or clcments, in un existing rcptcscntation. This
communication. That is, to the extent that people share representations, process is very similar to the way prototype theorists vie~ the assignation
they will be able to understand what other people are talking about and of stimuli to categories (see Chaptcr Six). The novel c~se is related to
will have fluid and intelligible conversations. The agreed representations typical or paradigm cases, and this allows the unfamiliar lo be understood
provide a stable, externa! version of the world which can form a tapie

e l
in terms of the more familiar. In the second stage- objcictification - the
1

142 Discoursh and social psychology From representations to repertoires 143

novel object is transformed into a concrete, pictorial element of the


1
groups and attempt to explicate their representations. The first problem
representatian to which it is anchored, and this new version of the represen- is that this presupposes the correctness of the notion that representations (y;._, , l..
tation is diffus~d, in the course of conversation, throughaut the social delimit groups. There is a vicious circle of identifying representations 1~~ j',:_,. '·
group. Thus wh~t was novel and disrupting now becomes, for that group, through groups, and assuming groups define representations.
part of their cohcrete reality. A second problem here is that researchers cannot easily identify
To go back to bur person with the social representation ofthe Conservative psychologically salient social groups independently of participants'
Party, if that pdrson encounters a new poli ti cal party of the right (he or representations of those groups. This leads to damaging inconsistencies.
she reads aboud it in The Times and chats to friends about it) he or she On the one hand, group categories will be treated as naturally occurring
may anchor it ~o his or her representation of the Conservatives. The phenomena which can be used as a clear base for research conclusions.
rationale mightjbe that it shares a stress on prívate ownership, law and On the other hand, group categories can themselves be understood as
arder, and the beneficia! effects of market forces, other elements in the social representations constructed in the course of participants' communica-
representation Óf Conservatives. Once ancho red in this way, the process tion. For instance, we might be interested in a particular protest movement
of repeated camhmnication with other group members generates a con~rete which we see as split between hard-line militants and moderates and
image of the nbw party, drawing on fea tu res of the representation of we may wish to interview both militants and moderates in arder to
Conservatives. Thus, it will be placed on a left-right continuum in relation discover the representations which distinguish them. But, in taking militants
to other parties!, an image of its leaders and supporters wil! be formed and moderates as natural groupings, we have already accepted one version
or representation of the protest movement which happens to draw on
and so on. 1

the militant/moderate categorization; there may well be other representa-


tions which do not make this distinction but create sorne other kind
of 'natural' groupings. These representations might not only give different
1

Problems witll social representations


labels to the factions in the movement (e.g. right-wingers and radicals
This theory has lprobably generated su eh enthusiasm among social psycho- for moderates and militants) but may well see the movement as divided
logists partly because of 1ts prom1se to throw hght on to the constructive into different kinds of groups altogether. There is no way of talking
(Berger and Ltickmann, 1966) aspects of social life, partly too because about the groups concerned independently of social representations,
of the claim tolpro~idc an i~!~~ra_~d frame _.for un~e-~st_~~~-~~--a~t_it~,S~es, yet Moscovici wants to argue that group membership determines those
attributions and behefs, and panly because'¡tprov¡des a ratmnale for a social representations.
specífical!yso-dar·psychological leve! of analysis. It seems to offer an anli- In ethnomethodological terms, we are back with the topic/resource
dote to the incrbasingly common cognitive reductionism in the discipline; problcm described in Chaptcr One - the object which is the tapie of
or the move to ~xplain our collectivc sociallivcs as a consequcnce of the analysis is also an analytic rcsourcc (Zimmerman and Pollncr, 1971).
operation of thd mental entities and processes of individuals. Al! the same,·~ As social psychologists we ha ve our own representations of the world
thcrc are a nurhber of difficullics which suggcst that it may disappoint \ and we use these as a resource in our analysis, but our research tapie
\( the enthusiast Jnd rcsist casy practica! application. Wc will conccntratc is thc naturc of rcprcscntations themselvcs. The methods we adopt to
on three: the rblation between groups and representations, the nature of make the research possible are the very thing we should be studying.
consensos assuÓled by the theory, and the roles of language and cognition. This would not matter if the categorizations of participants and researchers
were neutral, factual, descriptive acts. But, as we saw in the last chapter,
this is not the case.
Finally, difficu!ties arise because of the ambiguous empirica! status of
1

Groups and re{lresencations


social representations. If they are going to be related to social groups there
must be so me clear-cut, repeatable way of pointing to representations and
1

In this theory social groups are constituted by their shared social rep~esenta-
tions; the con~ensual adoptiori of representations 'establishes a' gioúp discriminating one from another. Yet Moscovici does not provide an
identity' (Mostovici and Hcwstone, !983: 116). Although theoretically analytic technique for performing this vital preliminary. The problem here
coherent, this ~remise raises immediate practica! difficulties when carrying is that it involves sorne wcll-defined notion of consensos and we will see
out research (Pbtter and Litton, 1985). Empirical studies of social represen- in the next section that this is no simple matter.
tations typical!y start with apparently wel!-defined and homogeneous social
'!
144 Discourse a11d social psycJzology From representations to rJpertoires 145
1
Consensus in representations With regard to the explanations which are used, rather than merely
mentioned, it is possible to make a further distinction betwhen use in theory
The premise that representations are consensually shared across a number -in a range ofbroad generalized fonnulations- and use id a range of more
of people is a central feature of Moscovici's theory. This claim secms specific or practica[ situations. For example, we can distlnguish explana-
relatively precise and unambiguous until we try to use it in practice. tions which refer abstractedly to the frequent part the polite force plays in
When we Iooked in detail at the way three different empirical studies of general in amplifying the severity of riots from the expiarlation given by a
representations operationalized consensos or agreement among a sample specific policeman defending his role in the St Paul's 'rlot'. There is no
(Potter and Litton, 1985), we found that they tended to draw upon ana!ytic hard and fast distinction here- it is mere! y meant to indica te the possibility
procedures which sirnply presupposed consensos and smoothed over of more or less theoretical and more or less practical accohnts. The crucial
interna! diversity. Sorne of these studies used numerical averaging tech- point for Moscovici is that consensus or agreement ambng a sample of
niques which, in their very narure, homogenize participants' responses. people could be assessed at either of these three levels.¡we could count
Di Giacomo (1980), for example, studying students' representations of al! the instances of mentior.. use in theary or use in ¡:jractice. Serious
a protcst movemcnt with a word association technique, simply prcsentcd confusions are gcnerated if thcsc differcnt levels are s· imply collapscd
1

average seo res for these associations across the sample. These seo res seem together as thcy usually are. A !'ter all, mentían can be tl~c polar oppositc
to demonstrate that the students share an undifferentiated representation of use. The i~portanc~ of distin:tio~s of _t~is kin~ ~ecqmes recognized
but an average may disguise considerable variation among u sample. A when wc bcgm a careful analys1s o! p<lrlicipants dJscqursc.
set of ten scores with an average of ten sounds a lot more similar that
five scores of zero and five scores of twenty (cf. Hewstone et al., !982). 1

We need to know the details of the spread and use of a representa ti en, Representarions: cognirive or linguisric? 1

whether sorne people apply it on many different occasions with a broad ¡

interpretation, or sorne on a more restricted basis. Other studies on social A furthcr problem with operationalizing social representations for research
representations (e.g. Herzlich, 1973) have presented only representative practice is at whar leve/ the operationalization should take place. As we
instances of accounts and similar! y seem to makc no attcmpt to look for havc sccn, Moscovici describes rcprcscntations as menÍa! entities madc
the kinds of variation which might cast doubt on their consensual status. up of both abstract and concrete elemcnts, that is, bo th concepts and
1

It is perhaps not surprising social representations researchers work in images. He seems to !acate representations as cognitíve lmits. However,
this way because, as it stands, the notion of consensus or the degree of for research purposes we are inevitably faccd with discdursc, whcther it
agreement among peop!e is a very difficu!t thing to define. For example, be participants' responses to structured procedures like sdaies or question-
in a study conducted with lan Litton (Litton and Potter, 1985) we tried naires, or thcir accounts elicited in open-ended intervie~s. This issue is
to identify consensual representations in a set of accounts of the St Paul's one we have come up against time and time again in this ' book. Given
1
street disturbances introduced in Chapter Six. What representations of the the essentially performativc and indexical na tu re of langu age use how can
'riot' will people commenting on the event Uournalists, residents, police, researchers construe it as a neutral record of seconctad phenomena, in
spectators) share in common? We found that it was essential to make a this case cognitive or mental states? 1

variety of distinctions befare our study of consensus could even get off More gene rally, the cognitive assumptions underlyink the theory are
the ground. In particular, we had to distinguish between what we shall highly problematic. Moscovici's account of the dual proceSses of anchoring
call 'use' and 'mention' and between 'theory' and 'practice'. and objectification for dealing with unfamiliar objects, il little more than
~~.l A representation which is used is one drawn on in an explanation of an excrcise in speculative cagnitive psychology. Givcn the metaphorical
r'~¡ events; for example 'the riot was caused by poor housing'. In contrast, language in which it is couc.:hed, it is not at a\l clear hbw claims about
¡ : } ..:.. a representation which is mentioned is not used to explain events but merely the proccss could be tested; for example, how would résearchers know
;, to refer toan available explanation: for example 'the left-wing press have if they had found an instance of anchoring? And are we really to assume
1

~:-· ,_. claimed the riot was caused by bad housing, but. .. '. Use and mentían ~hat, most of the time, people are understanding, say, thelpolitical proccss
';·'} are often combined in the same account, with a single 'correct' explanation tn terms of visual images? Do blind people ha ve a radically different notion
"\
being prometed in contrast toan alternative 'mistaken' explanation (this of political parties and the economy? 1

is just what would be expected in the light of Billig's, 1987, stress that There are important contradictions in the way thesc cognitive processcs
discourse is, in an important sense, argumentative). are described (McKinlay and Poner, forthcoming). Mosco!tici states that all

1
i

146 Discml~rse and social psyc/wlogy From repre.sentarions ro repertoires 147

cognition, ad mental cxpcriencc, Í.'l bascd on rcprcscntatians. But if our ATP is the basic chemical used to store energy in cells. Research into
pcrccption al· thc world is cntircly circumscribcd by our reprcsentations ATP has been dominated by two contrasting theorics. One group suggested
of the world, lhow can we even recognize new or un familiar social objects? that ATP is generated in a chain of reactions using a high energy che mi cal
How can the process of anchoring begin at all? Moscovici is vague on intermediary. The other group- championed by a biochemist Gilbert and
this point. sdme of these prob1ems might be resolvable, but it is question- Mulkay call Spencer- maintained that a chemiosmotic mechanism acting
able whetheria social psycho1ogical theory with such a heavy emphasis on at the cell wa11 was responsib1e for the transfer of protons. U1timate1y
the essentially social shou1d sadd1e itself with such a weight of unformu1ated the chemical intermediary preved elusive, the chemiosmotic theory became
cognitive ba!mage. widely accepted, and Spencer was awarded the Nobel prize (although this
is not to say that the alternative theory does not still have vociferous
i advocates). We are dea1ing, therefore, with an active, high-prestige area
.1 reper t mres
Interpretahve . of the natural sciences.
Gi1bert and Mu1kay jointly conducted extended, open-ended interviews
We will spJd the rest of this chapter introducing the notion of an inter- with thirty-four of the most productive researchers in this field. These
pretativo repclrtoire. This concept does sorne of the same sort of explanatory interviews were fully transcribed. They also amassed an archive of research
work as socih1 representations. That is, it attempts to look systematically papers, monographs, letters and other documents. Severa! different sources

l
at the organiZation of phenomena which social psychologists have tradi-
tionally undetstood in terms ofattitudes, beliefs and attributions. However,
this notion hrls thc advantage ofhaving been deve1oped in analytic practice.
of discourse thus formed the data base for their analyses.

1t is not handlcapped by presupposing a one-to-one concordance with group Contexts of discourse


boundaries, land it does not deploy the type of speculative cognitive
psychology 1hich underlies social rcpresentations theory. We will illustrate One of the first steps in this research was to compare the detailed content
the naturc anJ functioning of intcrprctativc repertoircs using one of thc of the accounts scientists gave in thcir publicutions with the accounts they
most develo~ed arcas of research in discourse analysis, namely the study gave in interviews (Gilbert and Mu1kay, 1980, 1984). Gilbert and Mulkay
of scientitic ldiscourse. designated these two situations in which scientists generated discourse the
One cxample of thi~ kind of research wa~ discussed in Chaptcr Three. 'formal' and 'informal' contexts, respectively.
In this chapler we will concentrate on an extended series of studies of The comparison concentrated on the introduction and method sections
biochemistd conducted by Michael Mulkay and Nigel Gilbert (Gilbert of articles, examining the way the same actions and beliefs were depicted
and Mulkay[ 1984). The central aim of this work was to reveal the in articles and then in interviews. The differences between the two were
interpretativb procedures used by scientists and the relationship between striking. For example, the introduction to an article might describe it as
the construdion of discourse and the particular ends to which it is put presenting a point of methodology; while in an interview the author of
(Potter and Mulkay, 1985). Mu1kay and Gilbert describe how scientists' the article might describe the same research as really militantly in support
accounts are ~ut together to portray their actions and beliefs in contextually of a particular, controversia! theory; or, the introduction to a paper might
appropnate. ¡vays.
'
st'ate that results suggest a new model; and yet in an interview the scientist
might describe the process as exactly the reverse ofthis: that is, claiming
that the model suggested the results.
Mapping rhJ /anguage of chemiosmosis The following twoextracts - the first from the introduction to a paper,
1
the second from the author's account of the same research- illustrate the
The starting jplace for their rescarch was a group of biochemists working sort of differences Mu1kay and Gilbert found.
in the UK and USA on the transmission of protons across the cell wall,
and on the ir\volvement of these protons in the formation of the complex
J. A long held assumption conceming oxidative phosphorylation has been that
molecule adclnosine triphosphate (ATP). In the next paragraph, we present thc cnergy availablc from oxidative-reduction reactions is used to drivc the
a thumbnail Jketch ofthis work. 1t is not necessary, however, for the reader formntion of [ ] ATP. Contrary to this view, reccnt rcsu\ts from severa!
to understarld the details of the science in arder to grasp Gi1bert and \aboratories suggest that encrgy is uscd primarily to promote the binding
Mulkay's cdnclusions regarding intcrpretativc repertoires. of ADP and phosphute in a catalytically compctcnt mude (1) and to facilitate

1
.
From representatwns to rep ert01res
1

149
148 Discourse and social psychology 1
.

the release of bound ATP (2, 3). In this model, bound ATP forms al the interpretative repertoircs to describe these ditTerences. Forl '
reasons which
catalytic site from bound ADP and phosphate wilh Jittle change m free will become clear, one was labelled the 'empiricist repertoire' and the
energy. (Gilbert and Mulkay, 1984: 41) other 'contingent repertoire'. 1

...::;.) 1-'J.-l~_,t._h,,: E: rr
1

2. He carne running into the seminar, pulled me out along wi~h one .o~ his lnterpretative repertoires are recurrently uscd systems of terms used 1
other post-docs and took us to the back of the room an? explame~ th1.s Idea for characterizing and evaluating actions, events and other phenomena, ;t C¡ (,¡,-;

that he had ... He was very excited. He was rea\ly h1gh. He sa1d, Wh?t 1

A repertoire, like the empiricist and contingent repertoires is constituted


,

if 1 to\d you that it didn 't take any energy to makc ATP at the catalytic through a limited range of terms used in particular stylistic ahct grammatical
si te, it too k energy to kick it off the catalytic ~it~?' lt too k hi~ a~out 30 constructions. Often a repertoire will be organized a~ound specific
seconds. But 1 was panicularly predispos~ to th1s '.dea .. Every~mg 1 d been
thinking, 12, 14, 16 different pieccs of mformauon m .the hterature t~at metaphors and figures of speech (trapes); we saw this, forlcxample, with
could not be explained, and then a\l of a sudden the simple explanauon the community repertoire dcscribed in the Jast chaptcr.
became clear .. , And so we sat down and designed sorne expcnments to The empiricist repertoire predominated in the context of formal research
prove, test this. [ ] lt took him about 30. seconds to .sell it to me. lt was papers. In this discourse the experimental data were giveh both chrono-
really like a bolt. 1 felt, 'Oh my God, th1s must be nght! Look at all the logical and logical authority. That is, the data were const~ued as coming
things it explains.' (Gilbert and Mulkay, 1984: 47) first and forming the foundation for theory. The author's involvements
and comnnitments were almost never explicitly mentionJd. Laboratory
In the formal paper experimental results suggest a model - in the interview work was characterized in a highly conventional mannet, as following
the central idea of the model beco mes a dramatic revelation. Moreo:er, from impersonal rules which were universally effective.¡The discourse
the results described as doing the suggesting in the paper are descnbed hada neutral style which seldom referred to the judgements of the author.
in the interview being produced from the intuitively formulated m~del. And when the scientist did apppear he or she was presentkd as forced to
Other differences are also apparent, particularly the Impersonal sct~nce undertake actions by thc demands of natural phenomena orlthe constraints
reporting style of the journal article and the depiction of scientific ~ctwns of invariant rules. The basic principie of the empiricist ~epertoire was,
and judgements, ifthey appear at al!, as constramed by data, expenments therefore, that actions and beliefs are a neutral medium through which
and, ultimately, the demands of the natural world. . empirical phenomena make themselves felt. 1

Gilbert and Mulkay observed similar variations when lookmg at the In informal interviews, when scientists were tallcing toa social researcher,
method sections of articles. Despite the exact and exhaustive aura of method the empiricist repertoire also featured. But, in these situadons the contin-
sections in science journals, and despite the idea t~at ~hey are s.upposed gent repertoire was also prevalent. This portrayed actionJ and beliefs as
to give sufficient detail to allow any competent sctenust to re.pl~cate the heavily dependent on spcculative insights, prior intellectuail co111mitments,
research procedures, a vety different view was frequently fou~d m t~fon;ml personal characteristics, unspecifiable craft skills, sociallties and group
talk. This presented method sections as ~ague outhnes ":htch sc.tenttsts memberships. A much wider range of lexical, grammati~al and stylistic
only interpret with the aid of their constdcrable craft sktlls. Thts vtcw resources was uscd in this repertoire. The general conn6ction between
is illustrated in the following extract. scientists' actions and beliefs and the real m of biochemiJ:ai phenomena
was much less clear-Cut. 1

3. You get a fc.cl for what you need. 1 can tell you a story about this. 1 went
The basic principie of the contingent repertoire was that professional
to thc workshop once to get something made. There was no ~ay they co~ld
do anything for me for a week or a month. They, ~ere mak1~g so~ethmg actions and beliefs are crucially influenced by factors outside the realm
for Dr x. 1 said, 'Whal are you making for Dr X? Dr X requ¡res h1s water of cmpirical phenomena. That is, actions and beliefs hre not forced
bath to operate nt36.5 C and ltothing els~.' 1 said, 'T~at's ridi.culo.us.' An~ by empirical realities but are a contingent product of rbanifold extra-
1 consulted with Dr X and he produced th1s paper showmg that m tlus expen- neous influences. The nature of the two repertoires is Jummarized in
mental protocol, they'd worked at 36.5 C. It didn't matter a damn really, Figure 7 .l. 1

whether it was 35 or 40 C, as long as you stayed roughly where you were. It is not, however, sufficient for analysis to simply identifY these different
Dr X was not an experimenter and no longer d.oes a~ y. If you a~e an 1

experirnenter you know what is important and what IS not tmportant. (Gilbert
forms of !anguage in the abstrae t. We need to know, firs t, the uses and
~~ --~-----·· 1

and Mulkay, 1984: 53) functions of different repertoires, and second, the probl<¡ms thrown up
by their existence. One important function Mulkay and Gilbert identified
These examples are just sorne of a range of diffcrences which Gilbert and was accounting for error. 1

Mulkay identified. Overall, they argued for the existencc of two broad

1
From represemations to repertoires 151
1

150 Discoufse and social psychology

1 FIGURE 7.1 4. 1 had no axe to grind. It's an advantage not being ab\e to contribute in the
! Scicntists' two inlerprctulivc rcpcrtoircs thcorctical scnse. I mean, you don 't fee\ that you ha ve time and publications
and reputation based on previous contributions and it's very casy to go the
( 1) Thc Empi~idst Repcrloire
Experimental data is given logical und chronological priority
way the cvidcnce seems to point. It lends to more flexibility. People like
Gowan and Fennc\1 cspccially and Milner, ccrtainly had many publications
The authbr·s personal and social commitments urc not mcntioncd and they discusscd one theory as thcy wcnt along and they had a lot invcsted
Laboratoh.- work is conventional and follows from impersonal rules in that ficld and 1 think they were psychologically a litt\e bit reluctant to
Basic pn1nciple: Aclions and bclicfs are a neutral mcdium through which fol\ow the lead of- utterly ncw, strnnge and different coming from somebody
cmpiribal phcnomcnu muke themsclvcs felt clsc completely. Ccrtuinly thnt remains thc case with Pugh. (Gilbcrt and
1
Mulkay, 1984: 64-5)
(2) Thc Contipgcnt Rcpertoir~ . . . . 5. !71is is the effcct of removing membrane potential. Now we ask what happens
Action apd belicf are deptcted as dependen! on speculattvc mstght, pnor 1f wc r~ow prevent [ ] a hydrogcn ion accumulation inside when we don't
commi'tmcnt, personal chumcteristics, social tics think we can have any membrane potential. Now you wiÜ havc, you will
Thc con~cction bctween action und belicf ami the phenomena undcr study ha ve people, particular! y people at [a particular university], who will give
is lessjcteur-cut . · . you absolute hc\1 about those experirnents. But the people at [that university]
Basic prjnciple: Scicntists' professionnl uctions nnd be\lefs nre 1mportantly are wrong. Thc pcople nt [that university] are wrong becausc they are too
inOueTcd by fnctors outsidc thc rcnlm of physical phcnomcna damned dogma tic. They think this an insuperable barrier to the chemiosmotic
theory or at lenst it is beyond the range that's acceptable to the chemiosmotic

tlcco/lllling fr·
1

error
theory. And that's no way todo science. The facts are pretty clear experi-
menta\ly and thcse people are sort of misquoting the fact. (Gilbert and
Mulkay, 1984: 65)
Mulkay and oilbert note that there was a basic asymmetry in the way
scicntists account for truc belief and their accounts for error (Gllbert
1
In each of these accounts, the speaker's own view - the correct view!
and Mulkay,ll984; Mulkay and Gilbcrt, 1982a). Each scientist partici- - is connected directly to the experimental evidence. That is, their
pating in the research took his or her own scientific beliefs as basic and theoretical position is pictured, through the empiricist repertoire, as arising
unproblematic. Indeed, this assumption could be posited as the fundamental directly from the natural world as revealed by experiment. In Extract Four,
principie of ~cicntists' discoursc. Wc are not, of coursc, claiming that the thc scientist claims that as he does not ha ve an axe to grind it is easy for
scientist's bdlief is a static inner object of reference - each researcher him to follow the direction pointed to by the evidence. Likewise, in Extract
modified and reformulated his or her view in a flexible and contextualized Five, the scientist describes the facts as being clear experimental! y, despite
manner. Thé point is that wlzatever the version the scientist currently the fact that other scientists cannot recognize them,
espoused it Jas taken as primary. Scientists warranted their beliefs through In contrast to this, the speakers charncterize erroneous belief- the
reference to fue constraining role of experimental evidence and the correct views of scientists they disagree with! - using the contingent repertoire.
application br procedural rules, that is, warranting via the empiricist False claims are represented as a consequence of flawed personalities or
rcpertoire (qilbert and Mulkay, 1982; Mulkay and Gilbert, 1983, 1985; distorting group allegiances. In Extract Four, certain scientists are said
Potter, 19 8'\). to ha ve too m ueh interest in a theory, so they are unable to give it up
Scientists have a problem, however, when they account for mistaken when c~mtradictory findings appear. In Extract Five, reference is made
beliefs: howl can they make scnse of the many errors committed by other to the personality of scientists at one institution, who are too dogmatic
scientists? If scientific views are prompted JUSt by the expenmental to see the val ue of a new theory.
evidence- ~s the empiricist repertoire has it- why do sorne scientists' The contents of the contingent repertoire are much more varied than
get it wroni? This problem was handled with the contingent repertoire. the empiricist repertoire. For instance, factors adduced to explain scientific
The clear-cllt connection between empirical phcnomena and the behefs error include: failure to understand, prejudice, commitment to one's own
br
and actions scientists was undermined and reference made to a Variety theories, reluctance to make the effort, complexity of the theory, dislike
of personalland social factors influencing belief. At its simplest, other of the new theory, extreme naivety, narrow disciplinary perspective, threat
scientists make mistakes bccausc sorne socwl factor comes between them to status, insufficient experimental skill, false intuition, subjective bias,
1 accepting the views of an authoritative figure, being out of touch with
and the facts.
Take the ~~following pair of error accounts for example. reality, personal rivalry, emotional involvement, general cussedness, being
!

152 Discourse and social psychalogy ~rom representatwns to repertotres


. 1 . 153
1

too busy, living in a country where theory is not popular, stupidity, introduced a package of non-scientific resources for expluirling these erran;
pig-headedness, being American and therefore thinking in a woolly fashion! and, as a consequence, allowed each scientist to maintain a Óoherent version
Each ofthese factors was used at sorne point to explain why other people of their social world which featured their own beliefs as ti\ e unthreatened
got it wrong (Gilbert and Mulkay, 1984). truth. 1

Tite 'chat 's how it is' response 17le truth will om device
1

One possibl~ response to this set of accounts is to simply accept them as So far, v:e have indica~ed in general terms the nature af the empiricist
a document of haw it is in the field of chemiosmosis, to say 'that's how and contmgent repertOJres, and documented onc important use of the
it is' in the field and take them at face value. But there are a number of c.onti~gent repe~oire. The notion of.rep~rtoires has en~bled us to dis-
fea tu res which rule out this path of action. First, the accounts of different ungu¡sh contrastmg sets of terrns used m d1fferent ways. These differences
scientists are afien highly inconsistent. For example, one scientist may should, however, also be salient to the participants. Do tlie scientists act
describe a theory as so complicated that it is very difficult to grasp, while as though the repertoires were contradictory? If scientists simply blend
another scientist describes the main virtue ofthe theory as its simplicity. the two repertmres together without perceiving any probledJ then the claim
Second, Mulkay and Gilbert repeatedly found that the error accounts of to have discovered two contradictory interpretutive systefus would seem
the same scientist were highly inconsistent. There was an enormous rather flimsy. · 1

flexibility in the way these scientists canstructed their discourse in arder On~ way ~f checki~g whether analytic discoveries havJ reality for the
to claim the high ground of 'truth' so vital for their self-presentation. part!c¡pants IS to se~ ¡f they themselves orientate to theml (Atkinson and
An even more startling conclusion emerged. If the accounts of these Hentage, 1984; Levmson, 1983). We use the term 'orientdte' rather than
researchers were combined, virtually everyone in the network had their say, 'notice' or 'understand' very deliberately, because J...e are not con~
work explained away by at least one other scientist. Between them the ~erned whether participants are consciously aware of thesd organizations,
scientists provided a contingent account of the entire scientific field in JUSt _whether they are a genuine feature of their interpret~tive practices.
which every contributor was either scientifically incompetent or crucially Gtlbert and Mulkay's contingent and cmpiricist rcpeftoires are two
affected by non-scientific factors. Again, this seems like a very good reason highly contrasting ways of making sense of scientific actidns and beliefs.
for not taking their discourse as a mode! of what is the case in biochemistry. On t?e whole thes: repertoires were kept separate in time ~nd space. The
Looking back to Chapter Three where we described how scientists contmgent repertmre was rarely used in formal papers and although both
presented the operation of rules for theory choice, we can now see that were drawn on in interviews, most of the time they appeatect in different
the pattem of accounts found there conformed to the broader pattern
for error accounting. Scientists in that case presented the operation of
t1
pas~ages oftalk or, as in error accounting, were applied very different
top1cs. However, there were occasions in interviews when thc twa reper-
rules with respect to their own theories in a clear-cut determinist fashion: toires appeared together and were applicd to the same dass of events.
the rule constrained the action of choice. An empiricist account of rule If a genuine dimension of the scientists' interpretative app~ratus has been
operation dominated. But, when describing the rule use of other scientists ca~tu~ed these cases ought to cause problems requiring discu1rsive solutions.
who they felt were mistaken, rules became a flexible strategic resource Th1s Is exact!y ~hat Gilbert and Mulkay found. 1

open to many interpretations. A contingent account of rule operation In. the followmg passage. the speaker uses both the ~ontingent and
became pre-eminent. Our knowledge of interpretativc repertoires helps empmctst repertmre tn g1vmg a general account of the field.
us understand the fine grain as well as the global pattern (McKinlay and i
Potter, !987; Yearley, 1985). 6. lnten•iewer. 1 think thnt's al!. Perhnps we could nsk one furthcr question
Mulkay and Gilbert's general finding, then, was that the researchers Are there any things which you think are important nboÚt this field bu~
they interviewed used an asymmetrical pattern of accounting for 'correct' w~ich.we hnvcn't toucbed on'? 1 '

and 'false' belief. This pattern of accounting solved an interpretative Sclt!l~fl.rr. Y7s, you havcn't touchcd on personnlitics l'cr) much. Spcnccr
and so on. I m not su re I want to tal k aboUl them. But I lthink thcy ha1•e
dilemma, reconciling the basic idea of the empiricist repertoire - facts contributed.
arise naturally from experimental findings - with the observation that other lnr.en•l:ewer. W~uld ~ou say something general without paming namcs?
scientists regularly seem to get them wrong. The contingent repertoire Sctentlst. Thc thmg ol which I'm wcll awarc is that thc anitudc tha! Mulhcrn

1
1

154 Discours~ and social psychology From represenrations to repertoires 155

took to anJ.hing Burridge published, which was of severe, critical, bitter This extract demonstrates how a speaker can use the TWOD, a very
opposition.jHc didn't like hirn. His bittemess has dis~dvantnged.h~, Mulhem comrnon rhetorical device, to deal with a particular interpretative problern.
enormously. Because it meant that other people dtstrusted hts JUdgmcnt. Up until the start ofExtract Six the scientist had made extensive use ofthe
And thcrc havc been occasions when people have said 'Oh, him' instead empiricist repertoire. In this extract he tentatively starts to introduce notions
of 'oh, tha~·, Somctimes peoplc ha ve been out to prove that somebody else
from the contingent repertoire. The problern is that his own scientific actions
is wrong, rhther than [uncleur]. But I think thnt incvitably things were seen
in that wayl !'ve seen othcr fields whcre things have been much more bitter. begin to seem arbitrary and pointless. Why continue researching ifbeliefs are
But scienc~ gencrally does progress very well and objectively, despite the only accepted because ofpersonalities and power? However, the Jooming
subjcctive 1eicment. 1 think there is a subjective element. contradiction can be dealt with by producing a temporal separation of
Interviewe~. Do you have any idea how this personal element gets eliminated? eiements. There may be contingent factors operating al the present but in the
Scientist. Only because a sufficient number of experimenters try to make future empiricist factors will cometo dominate. The structure ofthe TWOD,
the positioh clear. If other people are interested enough, if it's important which was recurrent in the interviews, emphasizes that in due course, in the
cnough, thbn the work will be done again or, more likely, its :amifications
fullness oftime, the facts will become fully established: the truth will out.
·wilJ be putsued. Predictions will be followed up, more expenments done,
and in the ¡runness of time a much clearer position will become apparent. For the scientist, this device is important not because it reconciles potential
Just as happened with thc chemiosmotic theory. And then, any personal contradictions between versions, but because it re-establishes the importance
rivalry will be seco for what it was, in relation to the facts, as they become ofthe empiricist repertoire. As we indicated when describing accounting for
more full~ establishcd. error, this is the dominant repertoire. Scientists make the basic presupposi-
lnten,iewer. So thc experimental evidence ... tion that their own actions and beliefs are determined in an empiricist manner
Scientist. !At thc cnd of the day salves cverything [general laughter]. by experimental facts. The TWOD allows this version to be maintained while
!tucn,iew~r. Ovcrwhclms thcse private antagonisms.
at the sarne time giving the speaker leeway to apply the contingent repertoire
Scienrisr.¡That's right. (Gilbert and Mulkny, 1984: 96)
toa wide range of persons, events and even whole scientific fields.
It is worth exploring this exchange step by step. At the start ofthe extract For the analyst, this device is irnportant because it corroborates the
the interview i~ drawing to a ciase. The two social researcher!i and the argument that thc discourse of these scientists is organized iota two bread
scientist (it is jvery useful to interview in pairs when con~erned with interpretative repertoires. Interpretative repertoires are used to salve
technical material ofthis kind, Mulkay, 1974) have been talking for over problems, but they also generate difficulties oftheir own (Levinson, 1983).
two-and-a-halfihours and thc interviewer asks the stock closing question: In this case it was expected that difficu\ties would arise when the repertoires
what has rwt bben covered? The scientist rathcr hesitantly brings up the were used together, and that is what was found. Ifthese difficulties had not
issue of persorialities and the interviewer, sensing something of interest, emerged Mulkay and Gilbert would have had rather Jess confidence in their
basic analytic claims.
probes. 1
At this point despite bei~g asked to speak in gener~l te;ms, the scientist Once interprctative repcrtoires ha ve bcen identified in this way analysts
describes the antagomsm d1splayed by a particular scJentJst, Mulhern. He can make .sen se of di verse areas of sociallife. Repertoires ha ve been used
then goes on (o characterize events in the wholc field in these sorts of in a nurnber ofstudies to systematically understand scientists' accounts of
terms; that is, Iic gives a contingcnt account of the developrnent of the field. theory choice (Gilbert and Mulkay, 1982; Mulkay and Gilbert, 1983, 1985;
The intervidwer then asks the scientist to explain how this personal Potter, 1984); their versions of the application of scientific knowledge
element mi_ghtibe e\iminat~d. _How is it tha~ scienc: can be both riven ':'ith (Potter, 1982; Potter and Mulkay, 1982), their readings ofscientific texts
personal disputes and prcJUdlccs and yet progress. very well ~~d. objec- (Potter, 1986), their accounts ofthe role ofmodels (McKinlay and Potter,
tivcly"? In an aiytíc tcrms, how can thc contrad1ctory cmpmclst and
1 1987), the construction ofargumentative letters (Mulkay, 1985); and the
contingent ver~ions of the field be sensibly articulated in the same passage construction of scientific jokes (Mulkay and Gilbert, 1982a).
oftalk. It is at fuis point that the scientist draws on a particular interpretative
technique: whht Mulkay and Gilbert have called the 'truth will out device'
or TWOD. Hb suggests that work which, presumably, has beeninitially Language, representation and repertoire
rejected will be re-evaluated in the future. More experiments will be
completed and •in the Jullness of time the position will become clear; the Having overviewed sorne of the research conducted with linguistic reper-
Jacts will becbme Jully established; the experimental evidence will so/ve toires we are now in a position to spell out, in a bit more detail, the
advantages of this analytic notion compared with social representations.
1. ' .
everytung ¡'

1
'
156 Discourse and social psycho/ogy "'
1· rom represematwns to reRertotres
• 1 •

157
!

'
In the first place, repertoires are not construed as entities imrinsically place made for cognitive phenomena in people's explanatibns. Discourse
linked to social groups, so research has not been hampered by the need analysis has eschewed any form of cognitive reductionisn\, any explana-
to engage in the often problematic exercise of identit'ying natural group tion which treats linguistic behaviour as a product of m~ntal entities or
boundaries. In the work on biochemistry, for example, it was found that processes, whether it is based around social represendtions or sorne
al/ the biochemists drew regular!y on both interpretative repertoires other cognitive fumiture such as attitudes, beliefs, goalslor wants. The
when constructing their accounts. These repertoires have been found concern is firmly with language use: the way accounts are constructed
in the discourse of other scientists, and it has even been suggested that and different functions. We are not denying the importarice and interest
analogous interpretative procedures are used in the legal professior. of cognitive science and the insights it has to offer; the poinf is that analysis
(Year!ey, 1985). Rather than make the somewhat unlikely assumption that and explanation can be carried out at a social psychologibal leve! which
all these people- biochemists, social scientists, and lawyers- are members is coherently separable from the cognitivc. Thc irony, ar¡ course, is that
of the same social group, it is much more fruitful to accept that repertoires a coherently social, social psychology is exactly one ofthe espoused goals
are available to people with many. different group memberships, and of social representations theory. However, it is discourse ~analysis which
1

patterns of accounting may not be the neatest way of dividing up society, offers a systematically non-cognitive social psychology aJ an alternative
or confirming conventional group categorizations. to the increasingly pervasive cognitive variety. 1

A more serious problem with any proposed relation between groups Finally, we should stress that there are no grandiose claims accompanying
and repertoires is the sort of realistic, decontextualized notion of group the notion of interpretative repertoires. Moscovici's infention was to
membership necessarily assumed. As Chapter Six demonstrated, group produce a theory of social representations. We are not anguing that this
or category mernbership is an occasioned phenomena, an achievernent. theory should be rep!aced with a theory of interpretative rJpertoires. The
This is not to say people are always free to change membership and concept of repertoire is but one component in a systernatic approach to
categorization in any way they desire, much of the time membership is the study of discourse. It cannot be isolated frorn other kinds of discursive
an occasioned product of other people, or of broader institutional arrange- phenomena. In addition, it is a preliminary step, further analyses
' are bound
ments (one only has to think ofthe comp!ex apparatus used in South Africa to refine this analytic unit. It may appear awfully gross iÓ severa! years
to police membership status). Yet studies such as Garfinkel's (1967) show time, and it rnay well be replaced by a hierarchical understa~ding of narrow
the constructed nature of even apparently obvious natural categories like to bread regularities in the content of discourse. We won't say that one
gender (see.also Potter et al., 1984: ch. 1; Wetherell, 1986). Mulkay and day 'the truth will out', only that, in this case, 'time mb tell'.
Gilbert's research, apart from the aspect we have considered, also displays
how biochemists produce highly varying group divisions in their field when
engaged in different interpretativo tasks (Gilbert and Mulkay, 1984: ch. 6).
The general point, of course, is that in discourse analysis, groups and the 1

way they are constructed in the course of accounts, have become an


important tapie of research in themselves.
A secbnd majar difference with social representations theory is that there
is no attempt in discourse analysis to find consensos in the use of repertoires
in the sense that sorne people are found to always use a certain repertoire,
and certain people another. Interpretati ve repertoires are used to perform
different sorts of accounting tasks. Because people go through life faced
with an ever-changing kaleidoscope of situations, they will neecl_!o_clr_a_\V
upon· very different repertoires to suit the needs at hand:f'rom this
theoretical perspectivo what is predicted is exactly variability rather than
consensos. Consistency is important in discourse analysis, it is useful to
identify the occasions where sorne people draw on one repertoire and sorne
another, but analysts do not assume that on other occasions these people
would necessarily produce the same repertoires.
The third majar difference between the two perspectives concerns the
How to analyse discourse 159

The premisc that certain crucial experiments can be identified which


overturn one theory and show the correctness of another has equally
8 been shown to be largely mythical (Lakatos, 1970). Historical work
has suggested that crucial experiments are not a factor in the actual
How to analyse discourse choice between competing theories, rather they are performed or at
least recognized as 'crucial' after the choice, to provide illustration and
legitimatian (Kuhn, 1977).
· In most textbo ks and monographs the methodology chapter comes near Similarly the image of experimental replication as a hard criterion for
the front for lhe perfcctly sensible reason that ii helps to know how the adequacy of findings has taken a severe knock recently as a result of
experiments are conducted, surveys carried out and so on, befare looking
' 1

sociological investigations by Collins and others into the actual process


at findings in ctdtail. There are very good reasons for following the reverse of replication (Collins, 1974, 1985; Mulkay, 1986). Collins has shown
procedure whdn introducing discourse analysis. Unlike experimental that decisions about the successful replication of an experiment are
methodologies lvhich ha ve become conventionalized and formalized over inextricably bound up with the question of what counts as a competent
1
the years, allof.ing an alrnost recip:-style format, .d'tscourse ana1ysts
. .IS experiment in the first place and, further, the exact nature of the
just emerging, tleveloping and changmg. We have tned to present severa! phenomemon of interest. Replications do not simply fail or succeed, they
specimen ana!Jses and theoretical origins in previous chapters befare are highly negotiable events.
attempting to a~stract general methodological principies in this penultimate There are many ways of interpreting the practica! upshot of this new
chapter. Tradidional social psychological textbooks also take a relatively perspectivo on science. Sorne philosophers ha ve argued, for example, that
cavalier attitudcl to the issues raised by the philosophy of science. For them, falsification is the only clear-cut criterion for developing knowledge
methodology rhrely raises troubling questions. For us, however, things (Popper, 1963), others that the only proper way to assess scientific
are not so cu~ and dried. We will begin our systematic overview of development is to compare entire research programmes (Lakatos, 1970),
discoursc anal~sis mcthodology with a considcration of sorne points raised and sorne havc claimed that now 'anything goes': the more plurality in
by the philoso~hy, history and sociology of science. theory und method the better (Feyerabend, 1975). Sociologists of science
have tended to argue that the problem is not one of generating better
philosophical or practica! criteria for the progression of knowledge but
Philosophy, sLiology and methodology our hopelesly idealized view of the scientific enterprise, the quality of the
1
knowledge it produces, and its role in transfonning our lives (Barnes, 1982;
As the 1980s progress, it has become more and more difficult to turn a Collins, 1985; Latour and Woolgar, 1979; Law and Lodge, 1984; Mulkay,
blind eye to the radical changes occurring in our understanding ofthe way 1979b; Yearley, 1984a).
science operat~s. We have to thank philosophers, historians and sociologists Psychologists ha ve been particular!y prone to the trap of comparing their
ofscience for this new view, or, more precise! y, this new collage ofviews discipline to mythical versions of the natural sciences. They ha ve tended
which has cryJtallizcd in the last few decudcs. Mention of just thrce areas to see the natural sciences as data driven, guided by experiment and almost
of change givbs a reminder of the veritable revolution evident in our exclusively concerned with the production of generallaws (Harré, 1986b;
conception ofiscience. Sloman, 1976). It is probably as a consequence ofsuch misperception that
The traditional and fundamental, distinction between observations, or the theoretician and the non-experimental are still regarded with suspicion
data statements and theoretical statements has been thrown into doubt
1 '
in psychology. There is a dearth of full-time theorists. The fact is, of
by people such'as Hesse (1974), Kuhn (1962) and Popper (1959). It is course, that this occupation is seen as absolutely essential in natural sciences
now taken for Jgranted that any observation of the physical or social world such as physics. However, the message we wish to get across in this work
is imbued with theoretical interpretation. It has been demonstrated that is that developing an adequate theoretical understanding or interpretation
cven the simplest scientific description is dependent on a whole .variety is at least as important as perfecting a cast iron methodology, and theories
of theoretical dssumptions. Without the distinction between fact and theory can be assessed using a set of empírica! techniques, experiments being
or observationland conceptual frarnework the si.mple-m~nded p.ositivist.view only one example.
of science deyeloping as a steady accumulat10n of s1mple unvarmshed
facts' censes to be tenable (Feyerabcnd, 1975).

1
160 Discourse and social psychology How ro ana/yse discourse 16/
1

Ten stages in the analysis of discourse facts and mak.ing others' errors understandable. Conversatidn analysts have
examined how coherent sequential discourse is construct~d with the aid
For convenience and clarity we will split the process of discourse analysis of a limited set of tum-taking rules, and also how partidipants perform
into ten stages. In practice, of course, these stages are not clear sequential ..--particular acts within the constraints of this rule system.¡ To summariza \
steps but phases which merge together in an arder which may vary - our research questions give priority to discourse, in any form, and ask J
(
considerably. However, we hope the didactic advantage will outweigh any about its construction in relation to its function. 1

potential confusions. Throughout our exposition we will use illustrations


from conversation analysis and Mulkayand Gilbert's study ofbiochemists' -- 1

discourse, as these indicate a range of different issues and practica! Stage two: sample selecrion 1

considerations.
Although many considerations in sampling remain unchdnged whatever
the area of social research, there are always sorne differendes in ernphasis.
Stage oue: research questions The first point, where discourse analysis diverges most radically from the
traditional view involves the basic question of - sarnple
-··· ....size.
T
It should be perfecúy clear by now that the questions asked by discourse Discourse analysis, at least at present, is arl Cxtremely labour-intensive
analysts can be many and varied. We can investigate mjcro:-_conversations approach. There is no discourse equivalent to feedind results into a
between two people, say, husband and wife, patient and doctor, teacher computer and then making sense of a limited pattern oflsignificant and
and pupil and the discourse produced by public figures aimed atan audience insignificant differences. Even after the demanding process of transcription
, .. of millions. We can focos on the detail of an exchange and then hypothesize a,nd preliminary coding is complete, the researcher is ineyitably required
about its functional effects as it becomes part of collective social interaction. ~o rcad and reread large bodies of transcript and documrnts. Thcre is a
In this new and fast-growing field the surface of possibilities has hardly danger here of gclting bogged down in too much data an9 not being ablc
been scratched. Yet there is one coherent theme or restriction linking these to Jet the linguistic detail emerge from the mountains artext. If one is
\ possibilities together. Participants' discourse or social texts are approached interested in discursive forms, ten interviews might provide as much val id
in their own right and notas a secondary route to things 'beyond' the text information as severa! hundreclreS~to a structurdd opinion poli.
p~oL .¡,
like attitudes, events or cognitive processes. Discourse is treated as a Because one is interested in language use rather than the pJople generating e-
potent, action-orientated medium, nota transparent information channel. the langua_ge and beca use a large number of linguistic patterns are likely ív2tt1 .- cl.e./..-!.._: 1
Crucial questions for traditional social psychological research thus cease ( to emerge from a few people, smJ!llsamples or a fewl intervie~are
to be relevan t. For example, we are not asking whether a sample of people gen uite ade te for investigating an interes.t0gl ancLpractically
are revealing their 'genuine' attitudes to ethnic minorities, or whether fans' imp_oriant-range..Q[p.!Jen9_mena. ForoiscoílrSeañálysts \he success of a
descriptions of what happens on the soccer terraces are 'accurate'. The study is 1101 in the least dependen! on sample size. It is 1¡ot the case that
concern is exclusively with talk and writing itself and how it ca_n_b~ a larger sample necessarily indicates a more painstaking or worthwhile
not with descriptive acuity. By restricting the questions asked in this way, piece of research. Indeed, more interviews can often siffiply add to the
discourse analysts are liberated from many pseudo-issues which have been labour involved without adding anything lo the analysiJ.
a recurring headache for traditional researchers. This does not mean that The crucial determinant of sample size, hawever, rriust be, here as
we are indulging in easy escapism, postponing the 'big' questions or ruling elsewhere, the specific research question. A number of ciaksic studies have
them out of court. As must be apparent from previous chapters, discourse concentrated on a single text, with the goal of showing hoJ., a certain effect
analysis adopts a radically new approach to the formulation ofthese issues can be achieved (e.g. Elgin, 1979; Gusfield, 1976; Potter et al., 1984;
and we will return to this point in the final chapter. Smith, 1978; Woolgar, 1980). In these cases the value origeneralizability
The research questions discourse analysts do focus on are broadly of results depends on the reader assessing the importance and interest of
related, as we have seen, to construction and function: how is discourse the effcct described and deciding whcthcr it has vital cÓnsequences for
put together, and what is gained by this construction. For example, the
research on scientists' discourse looked at how accounts were constructed
using the empiricist and contingent repertoires, and the function these
1 thc areu of social life in which it emerges and possibly l~or othcr di verse
areas. One can analyse, too, in a fine-grain way, onc-off representative
instances of what are unmistakably commonplace phenombna. At the other
accounts served in prescnting scientists' own views as constrained by the extreme, rescarchers havc conductcd interviews ucrofs an cxtensiv~

1
162
!

Discour1e and social psychology


1
l How to analyse discourse 163

sample, becaJse the commonplace or important patterns are not we have also looked at scientists' conference disputes (Chapter Three),
rccognizable i~ advance and recurrently used systerns of terms need to letters to newspapers (see Mulkay, 1986, on letters gene rally) and
be elucidated] (Gilbert and Mulkay, 1984; Wetherell and Potter, Par!iamentary debates recorded in Hansard. This latter source of data not
forthcoming a). only has the advantage of recording MPs and ministers constructing their
In many cas~s, practice will be governed by what is available: if you own versions of the social world and undermining opponents, but is an
have access to ~n extensive archive oftranscribed telephone conversations, ideologically powerful form of spoken material which comes ready
orto a collectidn of media reports ofriots, then that is the analytic starting transcribed, saving considerable work (see also Reeves, 1983; for a useful
point. GenerallY'
there is no 'natural' boundary line to be drawn in these exploration of the possibilities and problerns of documentary research see
cases, or no point at which sampling can be said to be complete. lt is simply Plummer, 1983).
a case of giving a clear and detailed description of the nature of the material It is at this stage in many research projects that two types of problem
one is analysing and its origins. 'Important' texts, like Nobel prize speeches cometo the fore: one practica! and one ethical. The practica! one concerns
or Minister's official reports caÍl be interesting; but rnundane conversation taperecording. If you are handling natural conversational material you will
is equally so, ahd is perhaps always the analytic first base (Heritage, 1984; need a recording of sufficient qua!ity to enable transcription and bear
Sacks, 1972).1 repeated listening. Fortunately, modem taperecorder technology has
provided machines which are pocket-sized, run offbatteries, have internal
1

microphones, and are sensitive enough for a high quality recording of,
Stage three: cbllection of records and docwnents say, a group of people having a discussion in a medium-sized room..J:he.
pragmatics of taperecording are usefully discussed by_Stubbs (1983)-
lt should be Jident that discoursc analysts work very frequently with and Labov (1972b). The ethical problems are much less tractab1e. The
records and ddcuments of interaction, as opposed to material garnered conundrum concems the dubious morality of taking surreptitious recordings
from the rcsea~cher's own dealings with participants. Traditionally, one weighed against the potential disadvantages of participants' knowledge
ar the most iniportant advantages of collecting naturalistic records and that rccording is occurring. There are various approaches to this problem,
documents is the almost complete absence of researcher influence on the al! of which have merits and demerits (see Stubbs, 1983). 'Obviously,
data (Webb etjal., 1966). Transcripts of everyday conversations, news befare using any letters or private documents, conference and workshop
reports, scient1fic papers, lcttcrs, official documents, are features of the transcripts etc., full permission should be obtained from all the participants.
social fabric ttÍat the researcher has had no part in producing. From the
discourse analhis point of view in particular, this material is helpful
bccause it alloWs the researcher to capture the widest possible variation Stage four: inten,iews
in accounts. Fbr example, a person may offer a specific, limited version
of their worldl in an interview compared to their writings or their un- Although naturalistic records, documents and transcripts of conversations
structured conrersation with their peers (Gilbert, 1980). By collecting have a number of advantages, interviews have the virtue of al!owing the
documents from many sources, recording interactions, and then combining researcher room for active intervention. In particular, they enable the
' with more' directive intervicwing, it is possible to build up a much
this researchcr to deliberately question an en tire sample of people on the same
'ruuer idea oflthe way participants' linguistic practices are organized issues, giving greater camparability in responses, and increased simplicity
i compared to one source alone. in initial éoding.
: --y ct another Jctvantagc accruing from records of interaction is that people Interviews, however, are a very different too! for discourse analysts
undermine eadh others' versions in these documents in a way which is than for orthodox social researchers (Brenner, 1985; Cannell and Kahn,
illuminating aÓd probably beyond the bounds of the researcher's com- 1968). The goal oftraditional interviews is to obtain or rneasure consistency
petence, not tolmention the morality and ethics guiding research. In effect, in participants' responses; consistency is valued so highly because it is
you can use people's own ability to artfully (and very helpfully) poke.holes taken as evidence of a corresponding set of actions or beliefs. If the
in each others positions to reveal their constructed nature. The most
1

' interview talk is consistent, the argument goes, it rnust reflecta consistent
obvious material of this kind is ordinary conversation, and, as we ha ve reality beyond; consistent discourse demonstrates the interviewer has
seen, therc is a! large and growing litera tu re about the analysis of this form found sorne genuine phenomena and not biased or distorted responses.
al' discourse (~tkinson and Heritage, 1984; Levinson, 1983). However, Consistency is important for the discourse analyst as well, but not in
i

164 Discourse and social psychology How to ana/y.\·~ discourse 165

this sense, only to the extent that the researcher wishes to identify regular
. i

detatled schedule which sets outthe questionsto be as~cd, and specilles


'

patterns in language use. the probes and follow-up questions which should be produced ifparticular
Given the theoretical primacy of the tal k itself in the discourse mode responses are offered. Thisguidesthe interviewer thro~gh the questions,
of research and lhe,fgg¡s_on.howJ¡tltj!;~QDstructed.andwhalit achieves mnkes sure the same questton ts asked of each respondent, and recards
rather than whether it is an accurate description of the participant's interna! any information which rnay be of use to the interviewe~. Bringing off an
state, consistency is often less useful and desirable for analysis than interview which systematically covers a range of topics,lyct is open-ended
\
\variation in interviews. Consistency suggests that participants are drawing enough to allow the respondent to elaborate on their vie:ws in a relatively
on a limited number of compatible discourses or interpreta ti ve repertoires naruralistic conversational exchange, is a craft skilll that takes sorne
when answering questions. Analyses which identify only the consistent developing. It will always be something of a compromise. It pays to
responses are thus sometimes uninformative because they tell us little about conduct and transcribe sorne pilot interviews befare th~ rescarch proper
the fu!l range of accounting resources people use when constructing the so thatthe adequacy of the schedule can be assessed, tind any necessary
meaning of their social world and do not so e!early reveal the function moditications made to question wording and overall Órganization.
of participants' constructions. Fi~ally, it is important to stress that since the interfiew is no longer
How, then, does the researcher go about generating an interview which consulered a rescarch instrument for accurately revealing an unbiased set
allows rather than restricts the diversity of participants' accounting of opinions, but seen as a conversational encounter,¡ the researcher's
practices? The first answer to this question is that you cannot, in fact, questions become justas mucha tapie of analysis as ¡the interviewec's
usua!ly stop this diversity appearing, except perhaps by confining answcrs. These questions set sorne of thc functional contdxt for the answers
respondents to simple yes and nos - and even here it will seep out. As and they must be included. In practice this means that thJ Jinguistic nuance
we argued in Chapter Two, a clase look at a verbatim transcript of of the question is as important as the linguistic nuancb of the answers.
virtually any interview reveals a degree ofvariation in responses normally The whole interview must be transcribed, rather thanjust thc interviewee's
only repressed by sorne sort of precoding or by burying under gross part. Unlike traditional interviews, the researcher's qucistions are seen as
categorization. active and co.nstru~tive .and ?ot passive and n~ut~al. 1

The second answer to this point is to make the interview a much more To summanze - mtervtews m dtscourse analysts dtffer from conventional
interventionist and confrontative arena tlmn is normal, dropping the formal intcrvicws in three ways. First, variation in response /s as importunl as
procedures which act as a device to restrict variation in traditional inter- consistency. Second, techniques which allow diversityl rather than those
views (Poner and Mulkay, 1985). This does not mean the interview should which eliminate it are emphasized, resulting in more informal conversa- ,
1
be turned into sorne sort of dispute. We are suggesting that thc researchcr tional cxchangcs and, third, intcrvicwcrs are sccn a!'i '1 ctivc participants/
should try to generate interpretative contexts in the interview in such a rather than like speaking questionnaircs. {
way that the connections between the interviewee 's accounting practices 1 .
and variations in functional context become cl"ear. One of the ways this
can be done is to tack.le the same issue more than once in an interview, Stage .ftve: transcription 1

in~erviews
in the course of a number of different general tapies.
For example, we asked those interviewed in our New Zealand studies When working with either records of interaction or it will be
questions relating to equality of rights in relation to three different tapies: necessit.ry to make transcriptions from audiotape. Both the importance and
the Springbok tour, Maori Jand disputes and explanations of unemploy- dif'ficulty of this task are often underestimated. A good trrlnscripl is essential
ment. The different kinds of answers given allowed us to illuminate the for a form of ana!ysis which involves repeated readings Óf sections of data,
more general features of interpretations of equal rights. Another method nnd the process of transcription itself can be helpful in forcing the
is to adopt follow-up questions to responses which pose alternative or transcriber to clase! y read a body of discourse. HoweJer, even the most
problematic views or facts for the interviewee. If the interview is seen d
basie transcription is extremely time consuming, and is skill that requires
as a forum in which the respondent regurgitates preformed and largely practice to perfect. The idea that transcription is 'simply[ putting the words
static opinions this approach wi!l seem strange; but if it is viewed as an down on paper' is very far from reality. Transcription is a constructive
active site where the respondent's interpretative resources are explored and conventional activity. The transcriber is strugg!ihg lo make c!ear
and engaged to the fu!l this will seem perfeelly natural. decisions about what exactly is snid, and then to reprbsent those words
It is important when conducting interviews of this kind to construct a in a conventional orthographic system (Stubbs, 1983).
!
i

166 Discoursel and social psychology How to analyse discourse 167


.
Th e t1me ta klr . .vanes
en 10r transcnpt!On . wtt "hh ..
t e transcnptmn system Stage six: coding
adopted. If you ate working with high quality recordings of small numbers
of people engage1d in relatively formal conversations like interviews, and The first thing to note regarding coding is that it is quite distinct from ,
your system pay~ little attention to pause-length, hesitations, overlaps and doing analysis itself. The goal is not to find results but to squeeze an /1
intonation, the rdtio of tape time to transcription time is about one to ten. unwieldy body of discourse iota manageable chunks. It is an analytic Y
That is, one hou~ of tape will take ten hours to transcribe. This includes preliminary preparing the way for a much more intensive study of the
the time taken toiaudio-type a first hearing, using a foot-pedal to stop and material culled through the selective coding process.
start the tape, into a word processing system, check a draft printout against The categories used in coding are obviously crucially related to the
the tape, and thdn put in corrections. If you wish to include phenomena r~search questions of interest. Sometimes they can be fairly stra1gffi:'
such as timed p~uses and overlaps in your transcription scheme it will fÜfWUfd. For example, Whellresearching scientists' criteria for theory
take much longér. For the full 'Jefferson style' transcription used by choice (Potter, 1984- see Chapter Three), the first step was simply to
conversation anrllysts (Atkinson and Heritage, 1984; and see Appendix select out from a body of transcribed conference interaction all references
One) the tape tinje to transcription time ratio is probably increased to more to criteria, however oblique, and the selection of theories. When looking
than one to twenty. at the tapie of community (see Chapter Six), al! occasions where this
The question br exactly how detailed the transcription should be is a category was used were selected during the initial coding. At other times
thomy one. Thefe is certainly a variety of work attesting to the significance the phenomenon of interest may not become clear until sorne analysis has
of intonational féatures of discourse (e.g. Brazil, 1981; Jefferson, 1985; taken place and a number of attempts at theoretical interpretation of the
Kreckel, 1981, !982). However, for many sorts of research questions, data. In these cases the process will be a cyclica1 one of moving between
the fine details dr timing and intonation are not crucial, and indeed they analysis and coding. For examp1e, in our study of New Zealanders'
can interfere witJ\ the readability of the transcript, particular!y when dealing accounts of racial inequality our understanding of what should be coded
with extended sdquences and for peop1e unused to the system. Moreover, out of the transcripts changed repeated1y as our analysis became more
for a project wdrking with, say, ten interviews, an hour in length, the sophisticated (Wetherell and Potter, forthcoming a). Once the tapie of
Jefferson systern is like1y to tu m three weeks of hard slog into six or more analysis was clarified, however, the return to the data and preliminary
weeks. So it is Ünportant to think very carefully about what information coding became less problernatic.
is required fromJthe transcript, and at what leve! the analysis will proceed. It is important ta stress that as cading has the pragmatic rather than
One advantge of transcribing on to a word processor is that it allows editing analytic goal of collecting together instances for examination it should be
in more detaile . information from the tape if it is found to be relevant. done as inc/u~f!:g_?y_ª~-P-~-~~i_~_le. In this sense it is quite different frorn
A final pragrn~tic point concerns the way transcript is presented on the standard techniques of content analysis where, for all intents and purposes,
paper. Transcri~t is a very expensive and vital commodity -a researcher coding data into categories and looking at the frequency of occurrence
may well read Jame sheets fifty or sixty times. Thus it is importan! to is simply equivalen! to the analysis (Berelson, 1971; Holsti, 1968; Mostyn,
print it out as c19arly as possible: using a readable type face, a new ribbon, 1985). At this stage in the research we are in the business of producing
double spacing ¡t on the page, and using short Unes. Cutting corners in a body of instances, nottrying to set limits to that body, Thus al! borderline\
any of these arE:as is false economy. cases, and instances whi~hseemmitllllly oíily vague1y related, should be
( included. All pages of transcript coded as containing relevan! instances
are then photocopied and p1aced in a file of their own. This file serves
lntermission 1

!
eventually as the basis for detailed ana1ysis. If a number of codings are
i done simultaneously and, given the time it takes to read through a large
At this point thci researcher usually sits back in contentment and surveys body of text, this is worth doing; it may well be that the same page of
a wholc set of chrdboard boxes with bits of paper in them. Sorne contain transcript will appear in a number af different categories, and hence a
documents of Jarious kinds, others are full of transcript and cassette number of data files, At the end of this process we are ready to move
tapes. There ark literally hundreds of thousands of words waiting to be into the most interesting stage: analysis proper.
1

transforrned int<) exciting research findings, and with this thought content-
ment can casi!~ be transformed iota total immobility and panic. Where
should one star¡'? We find it reassuring lo begin with sorne coding.

1
i
168 Discourse and social psyclwlogy How ro analys1 discourse 169

Stage seven: analysis We can see how lhis operates by looking once mole at Mu!kay and
Gilbert's research on lhe biochemisls (Chapter Seven).' In lheir search
Analysis of discourse is like riding a bicyclc compared lo conducting for pattern Mulkay and Gilbert slarted with accourtls of the same
experiments or analysing survey data which resemble baking cakes from actions, events or beliefs produced in different circulnstances. In this
a recipe. There is no mechanical procedure for producing findings from situation consistency would be expected, if one wasl working with a
an archive of transcript. There is no obvious parallel to the well-controlled realist view oftalk, and any variation between accountslis thus extremely
experimental design and test af statistical significance. In fact the results revealing. As we saw, this approach did demonstrate radical and pervasive
of studies of discourse are warranted, and critically examined, in a way variability on many le veis as well as a great deal oficonsistency. Bul
that is novel to psycho!ogy. We will explore techniques of validation in consistency was not due to accounts describing a consistent world; it
the next section; for the moment we will concentrate on how to get analysis arase because different types of accounts were restrictedl to different kinds
off the ground. of contexts. 1

Just as with bike riding, it is not easy to convey the analytic process Mulkay and Gilbert suggesled that lhe consistent pattern resulted from
in abstract. Words fail us at this point, it is not a case of stating, first scientists constructing their discourse in tenns of two con~rasting interpreta-
you do this and then you do that. The ski!ls required are developed as ti ve repertoires. They went on lo hypothesize about thci function of these
one tries to make sense of transcript and identify the organizational features two different repertoires: for example, the empiricist repertoire seemed
of docurnents. Nevertheless, there is a basic lesson that is inescapable: to be an extremely effective resource for justifying onJ•s own scientific
analysis involves a lot of careful reading and rereading. Often it is only be!ief, beca use it tied that belief directly to the facls m~king personality,
after long hours struggling with the data and many fa!se starts that a social and economic interests irrelevant. The contingent repertoire em-
systematic patterning emerges. False starts occur as pattems appear, bodied an alternative set of resources and was a per~uasive means of
excnement grows,only to fina that the pattern postulaled leaves too much accounting for others' errors and the failure of the scie ntific community
1

unaccounted, or results in an equally large file of exceptions. to endorse one's own beliefs. 1

Academic training teaches people to read for gist- which is precise! y The strength of Mulkay and Gi!bert's analysis !ies in the fact lhat it
lhe wrong spirit for discourse analysis. lf you reactan article or book the explained both the bread organization and much of lhc mpment-lo-moment
usual goal is to produce a simple, unitary summary, and to ignore the detail. The hypothesized functions provided an cxplanat¡on for the pattern
nuance, contradictions and areas of vagueness. However, the discourse evident in the discourse. This explanation was analytically more coherent
analyst is concerned with the_ detail qf_passage~_~f discourse, however than the alternative possibility that accounls were patt~rned in this way
fragmented and contradictory, and_with_w.hatis.actuilll.y_saidor written, because the scientific world simply is like this. 1

~só_in_e_-gener<J.!.:_Ídeá:t&áúel'.!!lsto be intended. In ethnomethodological lt should be clear, then, lhat there is no analytic method, at !east as
terms, we are so used to 'repairing the indexica!ity' (Garfinke!, 1967) this term is understood elsewhere in social psycho!ogy! Rather, there is
of talk, and reconstructing it in ways that make sense for us, that it is a bread theoretical framework, which focuses attention ah
the constructive
very difficult to throw offthis habit. Thus, part ofthe process is inevitably and functional dimensions of discourse, coupled with the reader's skill
a critica} interrogation of our own presuppositions and our unexamined in identifying significant pattcrns of consistency and va~iation. This does
techniques of sense making (Ashmore, 1985; Potter, forthcoming). The not mean lhat lhe reader has to take the conclusions of this kind of ana!ysis
analyst constantly asks: why am l reading this passage in this way? What on trust, however, because there are severa! stages ofivalidation; sorne
features produce this reading? are an extension of the analysis, others intrinsic to the presentation of
Ana!ysis is made up principally of two clase! y related phases. First,' findings. 1

there is th~_search for.gattern in~ta. This pattern will be in the form


of both variability: differences in either the content or form of accounts,
and consistency: the identification of features shared by accounts>.')~_:_gpa,
1

Stage eight: validation 1

there is the concern with function and consequence. The basic theoretical
lhrust of discourse analysisls the argumenllhat people 's tal k fulfils many There are severa! analylic tcchniques which can be us¿d lo validate the
functions and has varying effects. The second phase of analysis consists findings of this kind of research. The four main ones aJe: (a) coherence,
of forming hypotheses aboul these functions and effects and searching for (b) participants' orienlalion, (e) new problems, and (d) fruitfulness. We
the linguistic evidence. will take them in turn.
i

170 Discourse¡ and socio/ psychology How to analyse discourse 171

(a) Coherence 1 recognized by the scientists themselves. On the one hand, their discourse
A set of analytic claims should give caherence to a body of discourse. was organized in such a way that the two repertoires were kept separate.
Analysis should letus see how the discourse fits together and how discur- On the other, when the repertoires were produced on the same occasion
sive structure pfoctuces effects and functions. If there are loase ends, special difficulties were created for the scientists which had to be resolved
features of the chscourse evident in the data base which do not fit the by the use of a particular interpreta ti ve device: the TWOD. If the par-
explanation, wel are less likely to regard the analysis as complete and ticipants had not experienced these predicted difficulties, that is, if they
trustworthy. Iftne explanation covers both the broad pattern, and accounts had not orientated to the suggested inconsistencies, then we would be very
for rnany of thel micro-sequences, then we will take it more seriously. suspicious about the validity of the findings.
Apparent excCiptions to the analytic scheme are particular! y relevant to
the assessment df coherence. This tapie was discussed in Chapter Three (e) New prob/ems
1

where we introd uced the notion of confirmation through exception. If a One of the primary goals of discourse analysis is to clarify the linguistic
regular pattern br accounting is discovered, which we hypothesize is resources used to make certain things happen. However, these resources
designed for a sp¡ccial goal, then we must look for and examine exceptions
1

will not only so/ve problems, but will also create new problems of their
to this pattern. chses that lie outside the explanatory framework of a theory own. If we think of a car engine, it converts chemical energy into
are almost alwa~s more informative than those that lie within, and oftcn mechanica1 propulsion reasonably effectively, but in doing so it generates
dredge up imporftant problems. If there is clearly sorne special fcature of heat. Thus the car needs a cooling system to mop up the excess heat and
the exceptions '~hich marks them off from the standard examples and keep the engine working smoothly. The existence of new problems, and
thereby dctermihes thcir status as exceptions, the explanatory scopc of solutians, provides further confirmation that Iinguistic resources are being
our scheme is cdntirmed. If lhere are no special features which plausibly used as hypolhesized.
explain differende, the exclusive nature of our schcme must be questioned. If we look back to the TWOD again, we can see it as an example of
1
this process. The two interpretative repertoires used by the scientists are
(b) Participant~' orientation very serviceable for certain tasks; however, they create problems oftheir
The kinds of phenomena which interest discourse analysts have genuine own, particularly when used together. The TWOD deals with the contra-
consequences fÓr people's social lives. We are not interested in the dictions which arise and its exlstence provides further confirmation of the
dictionary definitions of words, or abstract notions of meaning, but in basic analytic suggestions. The extremely powerful turn-taking system
distinctions partibipants actually make in their interactions and which have posited by conversation analysts (Saeks et al., 1974) provides another
important implidations for their practiee. When looking at variability and example. The system is extremely effective for producing coherent
consistency, it i~ not sufficient to say that as analysts we can see that these sequential discourse, but it creates a problem of how to end conversations;
statements are donsistent and these dissonant; the important thing is the after all, if the system was too effective people could starve to death. What
orientation of tHe participants, what they see as consistent and different. is needed is an additional system which enables people to clase conversa-
There are a nJmber of ways of checking this orientation. In conversation tions down (Schegloffand Sacks, 1973). The existence ofthis secondary
analysis the rese~rcher is helped by the fact that conversations are organized system acts as a validity check on the existence of the primary system
on a turn by tufo basis, where each turn orientates to the previous one. (Levinson, 1983).
So although a t4rn of tal k may, for example, take the standard syntactic
form of a question, if the recipient treats it as an accusation the analyst (d) Fruitfulness
is alsojustified i!n interpreting it this way. Ifthe participants view the turn The fourth criterion of validity, and in many ways the most powerful,
as an accusatioh this overrides the theoretical consideration of what a is fruitfulness. This refers to the scope of an anal y tic scheme to make sense
question ought to \ook like. Atkinson and Drew's study of courtroom
1

of new kinds of discourse and to generate novel explanations. This is,


exchanges described in Chapter Four illuslrales the utility of this approach. of course, a general criterion of validity for scientific explanations and
In Mulkay arld Gilbert's study of biochemists' discourse one,of.the theories; if they can be used lo generate fresh solutions to the problems
central claims fas that scientists constructed their discourse largely in a field of research then we accord them more respect.
from two incompatible interpretative repertoires. To warrant this claim Both the sequenlial model at the heart of conversation analysis and the
Mulkay and Gilberl not only demor.straled that the repertoires appeared notion of repertoires deployed in work on scientists' discourse ha ve been
1

incompatible lo lhe analysl, they also showed that this incompatibilily was highly fruilful in this sense. For example, the sequential model has taught

1
i

172 Discourse and social psycltology How to analysJ discourse 173


1

us a great deal about the fundamental problem of accounting for indirect patterns and organization in the materials. The way cti1caurse is used in
speech acts, such as the use of questions like 'ha ve you gota Mars bar?' this kind of analysis contrasts with superficially simil~r work in social
to make requests (Levinson, 1983; Merritt, 1976). As we ha ve indicated, psychology and interpretative sociology in which extr~cts are presented
the notion of repertoires has helped delineate many varied aspects of to illustrate a causal story or model derived from partidipant observation
scientists' discourse, including unexpected discoveries like the structure (Collins, 1983) or reconstructed from accounts in the m~nner of ethogenic
of scientific jokes (Gilbert and Mulkay, 1984; Mulkay and Gilbert, 1982a). analysis (Harré, 1979). In discourse analysis the extracts are not character-
1

izations or illustrations of the data, they are examples [of the data itself.
These four techtúques for validating the findings of discourse analysis allow Or, in ethnomethodological terms, they are the tapie itself, not a resource
for a stringent examination of any claims. It is often suggested that non- from which the tapie is rebuilt. 1

experimental work or qualitative research is less rigorous than the stand~rd Discourse analysis reports will thus also contrast with ¡traditional content
altemative. Textbooks, for example, often promote an image of this kind analysis reports (Holsti, 1968). Although content analysis was developed
of work as the arena for hypothesis generation, with the rigorous work for use with documents and open-ended materials, its tlieoretical assump-
of testing and evaluation done through experiment. Yet there is no need tions, and assumptions about coding, are very differentl It uses discourse
for this to be so; research ofthe kind discussed in this book has undergone as a convenient dependent variable which records causal ~rocesses revealed
a searching and critica! examination, on a variety of levels, to assess its by the numerical assessment Or occurrenccs within categories. We have
adequacy. This is not to say that the criteria described he re are absolutely alrcady noted that one diffcrence between this appro¿ch and discourse
watertight or infallible. However, the philosophy and sociology of science analysis líes in the lack of a theory o[ language asan ac~ive, performative
described at the beginning of this chapter tell us that infallible criteria exist realm (see Chapter Two). A second contrast in terms 6f coding is worth
only in the land of positivist mythology: there are no crucial experiments, reiterating here. 1

knock clown refutations or definitive replications in the real world of Essentially, content analysis trains coders to reliably place passages or
science. parts of discourse into particular categories. For example, Tetlock (1984)
has coded politicians' speeches into different categorles of 'integrative
complexity' and then related these codings lo other yariables, such as
Stage nine: rhe report political allegiance (do right-wingers, for example, mal<,e simpler political
speeches than left-wingers?). The reliability of the coding is checked by
In work of this kind the final report is a lot more than a presentation of mensures of inter-rater reliability, which are basically catrelations between
the research findings, it constitutes part of the confirmation and validation the categorizations of different coders. 1

procedures itself. The goal is to present analysis and conclusions in such The problem here, as we have noted, is that although a high inter-rater
a way that the reader is able to assess the researcher's interpretations. reliability tells us that coders are agreeing we do not knbw the exact basis
Thus a representative set of examples from the area of interest must be ofthis agreement (Abraham, 1984). All we know is thrlt coders are using
included along with a detailed interpretation which links analytic claims much the same interpretative procedure. The readers br the final report
to specific parts or aspects of the extracts. In this way, the en tire reasoning are presented w~th _a definitio~ of the category, _and even _one or two
1

process from discursive data to conclusions is documented in sorne detail example categonzat10ns, but VJrtually none of the mter¡pretat¡ve work on
and each reader is given the possibility of evaluating the different stages which the research is based appears in the text. Al! we ,see is a numerical
of the process, and hence agreeing with the conclusions or finding grounds summary. This makes sense given the theoretical assu~1ptions of content
for disagreement. In this sense discourse analysis could be said to be more analyscrs, but is less than adequate if wc theorizc languagc itsetf as the
rigorous than experimental reports as it is often impossible to independently tapie of interest. Hence discoursc unalytic papers bring to thc forc those
check the analysis in these cases. aspects which traditional content analytic studies leave¡the reader to take
In practice this means that the analytic section of a discourse article will on trust. Namely, the detail of interpretation of a text. If a discourse
be considerably longer than the corresponding section of more traditional analysis report sometimcs looks Jess rigorous than al rcport of content
empírica! reports (an issue which journal editors will need to consider if analysis research this is probably more todo with the rt1etorical effective-
they are not to discriminate against the publication of this research). A ncss of tables of numbers than any lack of stringencY.
sizeable portian of the article will be taken up with extracts from transcript In tcrms of thc practice of discoursc analysis, the diScipline of making
or documents and the rest will be detailed interpretations which pick out intcrprctalions fully explicit for final rcports oflen rcvlls diflicultics and

!
¡

176 Disco1me and social psychology 1

analysis the same discourse can seem sol fragmented and contradictory
that it is difficult to see how it could e~~er be taken as sensible in the
first place. 9
Discourse analysis has radical implications for our understanding and
interpretation of findings derived from tr~ditional methods. As we have Controversial tapies and future directions
emphasized in a number of chapters, if w~ take a functional view of tal k
and writing we cannot simply put that vie-f to one si de when we deal with
the talk participants produce in experiments or the written answers people This book has described a dislinctive and novel social psychological
give to survey qtiestions. The functional petspective demands that we think approach to language. In the final chapter we would like to consider
about what people are doing in their expJrimental 'responses' and what three controversia! issues touched upon at various points in the book
perf~n_native effects follow from survey janswers. but which now need to be drawn togethcr and re-emphasized. These
It rs rmportant to stress finally that we are not making a general case that are the problematic relationship between language and mental states, the
experiments or surveys are intrinsically flkwed and should be abandoned connection between language and the world 'out there', and the issue of
by social researchers, or even that they might not have an important role reflexivity. In the rest of the chapter we willlook at sorne fu tu re directions
to play in the study of discourse. At this pqint in time the very phenomena for discourse analysis and its developing relationship with traditional social
of interest to discourse analysts are still being defined and understood, psychology.
and much of the most exciting research ha~ come from naturalistic studies
of texts or conversations. It may be thatj in the future experiments will
come to the fore, but at the present the sheer lack of basic knowledge Discourse and the world 'under the slwll'
of thc factors governing language use, coJpied with the extreme delicacy
ofthe processes of interest, makes it very difficult
' to design well-controlled Open souls telling more than they know
experiments. We would not, however, ~ish to advocate hegemony for
discourse analysis in social psychology, or rule out any futurc ovcrlaps If wc wantcd to pie k out two of thc mostlliamctrically opposcd positions
with other established methodologies. on the status ofpeople's talk, Harré's early work with Secord (1972) and
Nisbett and Wilson's (1977) research would be good candidates. Harré
v;otf. . '11-1'-¡G /. . .t
.f..l
and Secord propounded the 'apeo souls' doctrine which, at its most basic,
''"-
' (! responds lo the question 'why did these people do that?' with the answer
'why not ask lhem?' (Harré and Secord, 1972: 101-23). Of course, as
our discussion of ethogenics in Chapter Three demonstrated, Harré's
posilion is much more subtle than this formulation of the open souls
doctrine suggests. Harré and Secord are nol claiming people have perfect
insight into their own motives, or that telling researchers about those
motives is an uncomplicated matter, but they are suggesting that the
appropriate analysis of accounts will reveal to the researcher the nature
of the social competence responsible for generating both accounts and ..;/"' /<2<,. ~··
actions (Harré, 1977b). The basic aim of the research is to move from ··
an analysis of language to an understanding of an internal, causal,
mechanism - the person 's social competence.
In contrast, Nisbetl and Wilson (1977) propase that people's 'verbal
reports' about the causes of their actions are highly inaccurate. When
people give verbal reports, they argue, much of the time they are 'telling
more than they can know'. Nisbett and Wilson discuss a series of studies
which seemed to show that people's explanations of why. they liked a
stimulus, for example, or why they changed their mind on an issue, were
1 74 Discourse and social psyclzology r 1

11
How to analyse discourse 1 75

raiscs questions nol apparent at earlier stages. On sorne occasions a of the positive fruits of discoude analysis is to promate an informed critica!
discursive organization which seemed clear-cut breaks down and leads attitude to discourse of this lcind; to be more aware af its constructiVe
the researcher back to the coding, or even to the 'raw' documents and nature and the clase connectidn between the way textual versions of the
transcripts. A useful lesson can be leamed here: the process of writing world are put together and sp~cific policies and evaluations are pushed.
helps clarify analytic issues and thus it is often better to do a rough draft There are various models fa~ the application of discourse analysis. One
of the analysis and discussion quite early on than leave the report as a possibility is popularization,fgiving the knowledge away as freely as
separate postscript to the research. More generally, this reminds us that possible (Miller, 1980). An example of this kind would be Atkinson's
these stages are a conceptual scheme rather than a rigid temporal narrative. (1984a) work on how convedation analysis has helped demonstrate the
Discourse analysis involves fluid movement between the different stages, techniques used by political orlators to elicit responses from an audience.
with coding, analysis, validation and writing each leading back to earlier He suggests that an audience bétter educated about the workings of various
phases and ultimately to the talk and writing which were the original point rhetorical devices may resporid to speeches in a more sophisticated and
of departure. critica! manner. To promete tlüs end he produced a popularizing book .
as well as being involved in a fvidely discussed television programrne (in
which a 'naive orator' was shol"'n generating rapturous applause ata party
Stage ten: application poli ti cal conference after being trained to use the devices).
A second possibility is to o~en up a dialogue with the peop1e who have
The final stage in the process is one often ignored in social research - been researched. An example Ófthis k.ind would be Mulkay's (1986) study
application. Indeed. the whole issue of application is vexing. There is little in which a scientist is activelY and critically involved in the process of
understanding of how it occurs and the knowledge we do possess is analysing scientists' discourseJ It is too early to say whether these models,
fragmented and contradictory, with many studies finding it difficult to or sorne other, perhaps involving the educational system, will be the most
identify any c\ear-cut applications of social science rcsearch (Kallen ct al., fruitful. The main point is th~t application is very much on the agenda
1982; Potter, 1982; Weiss and Bucuvalas, 1980).lnterestingly, the position and should not be relcgated tb an optional extra.
is similar in the natural sciences. Recent findings in the sociology of science 1

ha ve 'undermined our twentieth-century notion of technology - non-stick


frying pans, bridgcs, atom bombs, micro-computcrs- arising naturally Conclusions 1

and straightforwardly from advances in pure science (Layton, 1977;


Sherwin and lsenson, 1967), so perhaps social scientists should not feel It is important to re-emphasize lthat there is no method to discourse analysis
their position is unique. Nevertheless, we feel that researchers should pay in the way we traditionally ü~ink of an experimental method or cantent
considerably more attention to the practica! use af their wark over and analysis method. What we have¡1is a broad theoretical framework concerning
abave the amassing of research findings and the furtherance of careers the nature of discourse and its role in social life, along with a set of
(Stringer, 1982). The image ofa benign body ofpractitioners waiting to suggestions about how discauhe can best be studied and how others can
read the jaurnals of pure scientists and put research findings into practicc be canvinced findings are geriuine. The ten stages we ha ve outlined are
is heartwarming but unrealistic. intended as a springboard rather
'
than a template.
This is a particularly intercsting issue for discourse analysis because Just like the biochemistry ctlscussed in the previous chapter, discourse
of the criticism sometimes voiced that it is 'just looking at words - not analysis is heavily dependent Ón craft sk.ills and tacit knowledge (Collins,
real things'. 1t is implied that a discipline concerned with language and 1974; Polanyi, 1958; Ravetz,l1971). People who run social psychology
its function will perhaps be of abstraer interest but no practica! use. It is experiments will of course know that these too are camplex social
important to remember that virtual! y the entirety of anyone's understanding accomplishments which take dansiderable sk.ill and organization to work
of the social world is mediated by discourse in the form af conversations, effectively. In describing disbourse analysis, it is particular! y difficult
newspapers, novels, TV stories and so on. to canvey the stirts of frust~ations and breakthroughs the researcher
For example, none of us has actually 'seen' the National Health Service experiences when engaging J.rith discursive data. Profound changes in
- it is not the kind of entity that could be seen: it is geographically highly understanding take place in t~e caurse of repeated readings. The initial
disparate and largely abstract. Yet we have conversations about 'it', read reaction is afien that it all maRes perfect, consistent sense, and that there
newspaper articles about 'it', and express opinions about 'its' future. One is no phenomenon to be resbarched. However, in the later stages of

1
178 Discours~ and social psychology Controversia[ tapies and future directions 179
1

simply mistakch. Thcy go on to make the alternative suggestion that so on. But because discourse analysts do not accord a different status to
people's accouÓts are not in any way accurate reports of mental cvents the 'inner' and the 'outer', orto the 'mental' and the 'non-mental', the
and procesSes. lnstead explanatory accounts offer conventionally derived question of how, precise! y, a person 's description of their mental state
beliefs about bhaviour ('I carne to a stop because the light started to represents or matches that mental state becomes irrelevant. The descriptive
change'), or arb ad hoc and possibly mistaken causal hypotheses which accuracy of discourse and its adequacy as a map or chart of prívate,
make sense of fue behaviour ('l liked the party because of the people'). subjective, mental experience is the non-issue from our perspective.
There are va~ious positions on the nature of this disagreement between
Nisbett and \\¡ilson and those who favour the open-souls approach
(Bainbridge, 1985; !ckcs et al., 1986). Sorne argue that this is, in fact, Beyond intra-psyclzic explanations
a pseudo-conflibt, because Nisbett and Wilson confuse what is essentially
a conceptual theoretical
'
issue about how a social science should be It may seem odd to be promoting a non-cognitivo social psychology when
conducted with an empírica! point (Gergen, 1982; Sabini and Sil ver, 1981;
1

the vast majority of current research is attempting to give intra-psychic


Shotter, 1981).¡ln general, mainstream social psychology splits down the explanatians ofsocial behaviour. However, as we have seen, there are both
middle on this issue. Many studies implicitly adopt the open-souls approach theoretical and methodological advantages to this orientation. In the course
and see no dirficulty in taking a verbal or written utterance as a clear of this book we ha ve looked critica!! y at a number of fundamental! y intra-
reflection of a cbgnitive or mental state. Large amounts of attitude research, psychic explanations which draw on notions of attitudes, rules, the self,
for examplc, fahs into this category, as does the study of social representa- categories and representations and which try to move from language to the
. and so on. In many
tions, aspects of' research on the self, person perceptlon mental state. In each case we ha ve described an alternativo approach which
other cases, hdwever, the social psychologist beco mes suspicious of the caneen trates on description and explanation in tenns of discursive organiza-
veracity of peÓple's descriptions of their internal states. As we saw in tions instead of moving, in Garfinkel's (1967) phrase, under the skull.
Chapter Two, lome attitude theorists such as Fishbein and Azjen (1975) There is a powerful tradition of non-cognitivo thought in linguistic
argue that people can misrepresent their basic underlying attitude as a rcsult philosophy and ethnomethodology which has been concemed particular! y
of nonnative p~essures. There is considerable tension in social pyschology with our vocabulary of mentalistic terms: words like belief and memory
betwcen the principie that people can be trusted to describe their interna! which, at first sight, appear to be references to mental states and entities.
// states and the 1 principie that the researcher must remain vigilant and Gilbert Ryle (1949) and Ludwig Wittgenstein (1953, for a more elaborate
,~--' sensitive to the possibilities of conscious or unconscious fraud. This issue, account see Remarks 011 the Philosophy of Psychology, 1980) have been
however, is rarely discussed systematical!y and is generally left to the central in propounding this approach in philosophy, while Jeff Coulter
commonsense jof the researcher. (1979a, 1983, 1985), Rom Harré in his more recent writings (1983), and
Discourse analysis takes a rather different position when faced with this John Sabini and Maury Silver (1982) have, in different ways, done much
problem ofthe !relationship between utterances and mental states. We argue to develop the implications for social research.
that the researbhcr should bracket off the whole issue of the quality of The central plank of this position is the argument that much of our
accounts as ai:curate or inaccurate descriptions of mental states. The vocabulary of mentalistic terms has no 'inner' referent at all; instead of
problem is beÍng construed at entirely the wrong leve!. Our focus is being merely descriptions af mental states these words are themselves an
exclusive! y on! discourse itself: how it is constructed, its functions, and autonomous part of particular social practices. Take the term 'under-
the consequenbes which arise from different discursive organization. In standing' as an example. When first thinking about this word, we are
this sense, disbourse analysis is a radically non-cognitive form of social probably tempted to view it as a description of sorne private or inner
psychology. j experience. However, when we Iook closely at the way this word is actual! y
lt is importdnt to be clear about this point. Discourse analysts are not used this view seems less convincing. Ryle has pointed out that this term
suggcsting thal the phenomena traditionally thought of as mental should be if afien used to mark a claim to success, the sort of situation where one
excluded from! analysis beca use they are in sorne way prívate, mysterious, might say '1 have been working at the problem and 1 think I understand
unknowablc. As we saw in Chapter Five, the la11guage of the self and it now'. However, merely having a moment of insight ora feeling ofhaving
mentallifc is ~ public one which is available for analysis (Coulter, 1979a). cracked thc problem is not sufficient. If the person with the feeling of
In practice, n\uch of the phenomenon of the mind is intersubjectiv_ely comprehension tries to apply their ideas and finds they do not work, they
constituted as lthe person speaks, writes, reminisces, talks to others and will know that they wcre mistaken in the belief that they understood.

1
1

180 Discourse and sacia/ psyclwlogy Cnwroversia! topics and future directions 18/
1

Furthermore, if other people assess the understanding gained, they may Barrles, s/zips and lzidden construction 1

decide that the person only thinks they understand - they do not real/y
understand. The general point is that although cognitive processes are To begin with we need to look al how this issue is formulatetl. A dichotomy
clearly going on, and people without a brain certainly do not understand, is created between ·real' events and linguistic represen~ations of those
this is not a sufficient condition for understanding. Understanding is events. This formulation, like the dichotomies between mehta!inon-mental
assessed by public criteria and practica[ tests. The term understanding or subjective/objective, is not, of course, a neutral or n~tural
1
division.
is properly used when these criteria can be, or have been, satisfied, not It is a philosophical construction with a long history in W estem societies
merely when people ha ve a certain experience. Similar arguments ha ve and by agreeing to discuss in these terms one is fated to ~nd up hung on
been used with a variety of other terms, such as 'knowledge' (Harré, 1983), one ofthe contradictory poles. Why do we talk about a -Jorld 'out there'
'memory' (Coulter, 1979a, 1983; Harré, 1983), 'belief (Coulter, 1979b), in the first place? Js this world in sorne way independent oChuman affairs,
'intention' (Anscombe, 1957), 'envy', 'anger' (Sabini and Silver, 1982) a realm removed from human constructive activities? 1

and 'motives' (Mills, 1940; Peters, 1959; Sharrock and Watson, 1984- Collins (1985) has provided a rather nice analogy inl relation to his
and see Chapter Four). research on scientists which captures sorne of the constraints this kind
For the most part, this argument in linguistic philosophy and ethno- of thinking generales. Social researchers and people in general populate
methodology can be smoothly integrated with discourse analysis as it has their worlds with many different k.inds of basic objects: sbcieties, selves,
been outlined in this book. In suggesting the relationship between language biologies, physical phenomena, material realities and sd on. Following
and mental states is a non-issue, we are agreeing with this tradition which Collins' analogy these .objects are like ships in bottles. ¡Their presence
treats descriptions of mental states as discursive social practices. There IS unquestwned and miraculous. The sh1p IS m the bott e, it looks like
are, however, two drawbacks that might arise. The first stems from the it must always have been there, it is impossible to thirik how it could
danger of getting involved in fruitless debates about the reality or non- be otherwise, the finished product is all that is visible. Those who
reality of mental entities, which can easily end in the kind of linguistic marvel at ships in bottles, of course, have never observdct the elaborate
imperialism which denies all significance to cognitive processes. The and detailed process of construction, as the ship is re-crdated with small
second problem is that it is easy to move from this position to making pulleys, matchsticks and so on. The process of manhfacture is left
the assumption that other sorts of discourse are straightforwardly referen- obscure. In the same way discourse analysts argue that the constructive
tial. One ofthe main arguments ofthe discourse analysis described in this process used to furnish the world 'out there' and the dichotomies which
book is that al/ language can be analysed in terms of construction and result remain obscure. The world and its objects appear nlady completed,
function, not rnerely the language of self-reference. As we have seen, There seems no alternative but to accept il as it app¿ars. Discourse
descriptions of scenes and events are as rnuch an arena for presenting analysis aims to explícate the constructive activity involvetl in the creation
motives, giving disclaimers, and justifying actions, as accounts using the of a 'world out there' and for this reason is reluctant to taki: any dichotomy
mentalistic terminology of desires and beliefs. for granted, without researching why problems should ~e formulated in
. 1
th IS way. 1

Discourse and the world 'out there' Factual accounrs as ropic 1

People sometimes assume discourse analysis denies the existence of a world There are two problems which immediately arise whenlthe question of
'out there'. 'Why this concentration on language', they ask, 'when people discourse and 'the world out there' is posed. First ofall, as we have noted,
out there are giving birth, making money and being murdered by oppres- the form of the question implies access to sorne clear-cutf non-discursive
sive regimes? Why don 't you study these real processes and not just realm which discourse 'relates to'. However, any formulation ofthe nature
language which is a second-hand superficial medium?' This contrast ofthat realm will inevitably be framed in discourse and aJ we have shown
between the real material world and language opens a veritable snake-pit repeatedly throughout this book, you do not need to study many of these
' 1

of philosophical and political issues but it is important to briefly review formulations t.o find. si?nificant variation between them.¡so, in practice,
why discourse analysts argue one should focus on language construction when addressmg thts tssue, the nature of the world out there is either
and function in this context. stipulated by analytic fiat, sorne versions are rcified ancl
sorne ironizcd
1 '

1
1

182 Discoursk and social psychology Controversia[ tapies and future directions 183
1

which is clearl~ unsatisfactory, or the focus should shift to the question Rejle:xivity as a tapie and a strategy
of how the versions themselves are managed (Cuff, 1980).
The second problem with discussions of 'discourse and the world' is One upproach has been to take social research, and its discourse, as a
the implication that the resolution will necessarily be a philosophical matter. tapie of study in its own right. We have already described sorne research
Discourse ana!ysts ha ve preferred a different route. They have focused of this kind. For example, in Chapter Three psychologists' justificatory
attention on sitJations where issues of 'fact' or 'perception' cometo the accounts of their theory choices were examined, In another example,
fore, such as tHe discourse of science or the courts (cf. Pollner, 1974,
1
Stringer looked at how factual discourse was used as an unexplicated
1975; Smith, 1978). Discourse analysts ask questions such as: what resource in a sociopsychological analysis (Janis, 1972) ofthe 'Bay ofPigs'
procedures arel
used to authorize factual accounts? How are factual incident from the early months of the Kennedy presidency (Potter et al.,
accounts produeed to perform a specific act? How is the effect of 'mere 1984: 100-15). Stringer's account shows that Janis's group dynamics
description' geÓerated in discourse? For example, in Chapter Four, we explanation tended to use his source texts (particularly Schlesinger's book
discussed how tlefendants in court constructed factual accounts in such about the presidency, A 17wusand Days) as neutral historical docurnents,
a way that theyl provided mitigation for potentially culpable behaviour.
1

and failed to acknowledge the strategic purposes these texts might have
In general, this path seems a more fruitful one than being caught within been fulfilling, in particular their role in justifying Schlesinger's own
a certain formulation of languagc and the world. actions. For other work of this kind see Bazerman (1984) and Gusfield
!
(1976).
! The problem with approaching the issue in this way is that the constant
Discourse andl re!lexivity emphasis on other people 's texts can blunt the radical implications of
retlexivity. Thc point is that alltexts, cven thc one we are writing and
Reflexivity is another issue that has increasingly come to thc fore in you are reading at this very moment, are subject to the same processes
discourse anaiY¡sis. We met ane sense of this term in Chapter One in and have the same constructed, action-oriented aspects. Taking reflexivity
nur discussinn of Wicclcr's wnrk. Rcncxivity in this scnse rcfcrs to thc mcrcly as a tapie defers attcntion from the most focal, present text. To
fact that talk hhs the property ar being both abow actions, events and overcomc this problcm, sorne researchers ha ve wanted to draw attention
situations, and ~t the same time part of those things. However, there is more directly and forcefully to the construction of tlzeir own texts by
a second, morelgeneral, sense ofthis notion which draws attention to the departing from the standard conventions of academic writing and exploring
point that the findings of discourse analysis apply equally to the social 'new literary forms' (see Woolgar, forthcoming).
texts produced by discourse analysts as to anyone el se. Put another way, The point of these 'new literary forms' is not to document or describe
if the upshot cif research of this kind is to question the simple reu!ist problems of re!lexivity but to give a practica! demonstration of their effect
model of the dperation of discourse, and suggest that 'realism' is, at (Woolgar, 1982). The goal is to remind the reader that the text is nota
least partly, a ~hetorical effect constructed through the careful choice of mere record of events, a collation of facts-, but is a complex, multifaceted
particular lingJistic forms, then what are we to make of the discourse in social achievement (Stringer, 1985). One way the conventional and con-
which this clairh is itse!f couched? How should we deal with thc fact that structed nature of the analyst's text can be pushed to the foreground is
our accounts of' how people's language use is constructed are themselves to present the analysis using a very different set of conventions. For
constructions? 1 example, Mulkay (1984b, 1985) has used a dramatic, theatrical play formal
Most of the time this problem does not actively trouble discourse analysts to illustrate the multiple versions of the idea of replication drawn on by
as ihey practis.! their craft. It is possible lo acknowledge that one's own scientists and social researchers when legitimating their activities. 'otliers
language is coÓstructing a version of the world, while proceeding with have used the literary form of the !ecture and encyclopedia (Ashmore,
analysing texts hnct their implications for people's social and politicallives. 1985), parody and dialogue (Mulkay, 1985).
In this respectl discoursc analysts are simply more honest than other Another method of i!lustrating problems of re!lexivity in practice has
researchers, re bognizing their own Work is not immune frorn th9. SSJ.~i.~_J beco to construct anal y ses with a self-referential quality. These studies
1
psychological proccsses being studied. Most of the time, therefore, the examine at the same time the tapie and their own investigation of that tapie.
most practica} Way of dealing with this issue is simply to get on with it, Typically, this operates by utilizing two, or more, parallel discourses, one
1
and not to get ~ither paralyzed by or caught up in the infinite regresses of which is concerned with the subject of the research while the second
possible. How~ver, there are other intriguing responses to reflexivity. tackles the construction of that subject in the first discourse. For exarnple,

1
¡

184 Discourse a11d social psyc!w/ogy Cuntroversialtopics cmd .fíllure d~rections J8j

Wynne (forthcoming) has examined accounts of the onset of multiple shows the potential for integrated work. In our view thc areL of agreemcnt
sclerosis and her own discursive formulation of those accounts in a parallel are .sufficient to warrant this integrated approach, and wb hope each of
text (sce also Ashmore, 1985; Mulkay, 1985). Rcficxivity in these cases the :·ndividual perspectivcs and rcsearch traditions will beryefit from more
is a meansrofexploring and illustrating rhetorical construction through crms fcrtilization than has occurrcú un ti! now. As wc pointcd -out at thc
analy~ one's own analysis, and is an cxciting development which ver~· bcginning of the book, a lot of research has bceÓ conducted in
challenges the literary skills of the social scientist. rela jvely isolated groupings witl1in different disciplines, yel there are many
pandlels in thc questions asked and the sorts of tcchniqueslused to answer
then. i
The development of discourse analysis

Tize theoretical coherence of discourse analysis Discourse analysis and traditiona/ social psychology 1

.1

It will be apparent that in the course of this book we ha ve discussed research There are potentially thrce kinds of interaction which could' occur between
derived from a number of rather different theoretical perspectives. We discourse analysis and what we have referred to as tr~ditional social
have treated this work as a whole because it shares several basic assump- psychology or the vast majority of research which is qoncerned with
tions about participants' discourse. In pürticular, al! the research we ha ve cognitive process and the cxplanation of social behaviourl The f1rst kind
covered is concerned in one way or another with the action orientation ofinteraction is, basically, no interaction at all. Much oftheltime, discoursc
of discourse. It takes discourse as a research tapie in its own right rather analysis simply asks a different set of questions using a different kind of
than treating it as a transparent medium through which the 'real facts' method. Insofar as tlús is thc case, it occupies the role of a nbw, but distinct,
of attitudes, events or behaviours can be recovered. Moreover, it takes facet of the many which make up social psychology. 1

a social perspective which focuses on the role of discourse in interaction The second kind of interaction which could develop would allow
and sense making rather than being concerned with, for example, abstract discourse analysis to actas a resource for more traditiondl work. One of
1
questions of semantks, text coherence or aesthetics. its main inputs here would lie in improving our understan ding of the sort
Ncverthcless, therc are still important issues oftheorctical disagrecmcnt of lliscoursc which is thc bcdrock or traditional social psychological
between the different perspectives discussed he re. One issue which particu- methods. As so much of social psychology is bascd more or less dircctly
larly distinguishes those adopting a conversation analysis approach centres on participants' discourse in one form or other, a clcarcr picturc of its
on the questian of sequcnce. Conversatian analysts, such as Sacks, have org:.mization and working should aid thc construction of fucttcr mcthodo-
argued that a sequential analysis is fundamental for explicating the sen se of logical techniqucs and the intcrprctation of ·verbal rcspbnses'.
As we notcd in Chapter Two, there is airead y a large lit~ raturc on issues
1
utterances for participants (see Chapter Four). A full analysis o fa particular
speaker's turn of tal k will, in this view, depend on an explication of the of response bias, experimenter cffcct:;, demand charactcristics and so on.
sequences of turns in which the particular turn is embedded; only through It is well-known that participants' discourse varíes accorctiAg to subtle cucs
this procedure will its nature and consequences be adequately understood. in the wording or sequencing of questions and in experimbntal protocols.
Those taking a more semiological perspective emphasize the importance Research on language function could fit these phenomenÁ into a broader
of a combinarion of sequential considerations and considerations of replace- theoretical framework, and ultimately may be able tal make detailed
ment. That is, they argue that the researcher needs to understand both predictions asto the sorts of wordings which will produce¡ certain effects;
the discourse which is actual/y produced, and the discourse which could what kinds of question wording, for cxample, willlcad to answers stressing
have been produced, but was not on this occasion (Culler, 1976; Fiske social tlesirability and what constructions might lead to ariswcrs strcssing
and Hartley, 1978; Hartley, 1982; Hodge and Tripp, 1986). From this soc ·.al differentiation'? 1

perspective, to get the necessary analytic grasp on an account talking of The third kind of interaction would build on the tension between the
a 'disturbance', say, we need to know the sequence in which that particular theories and findíngs ofthe two arcas. Tension and connic~ are, of course,
account was embedded, and the fact that alternative accounts talked of an t~ssential and creative part of scientif1c development. '(Ve stressed this
a 'riot' rather than a 'disturbance' (Potter and Reicher, 1987). kim l of interaction in Chapters Two to Seven where we trie# to demonstrate
Despite these differences and possible disagreements, we have chosen how sets of findings from areas of traditional social psyc~ology could be
to present this research in a way that draws attention to similarities and better understood with perspectives from discourse analys~s. Initially, one
186 DiscoJse and social psyclzology Controversia[ tapies and future direcrions 187
1

of the import<lnt advantages of discourse analysis may well be to raise Trew, 1978; McHoul, 1982; Smith, 1974, 1978, 1982, 1984; Woolgar,
difficulties wlth currently acceptcd models and theories. Eventually, 1980; Yearley, 1981). The classic work of this kind comes, however, not
however, the tlebate is likely to work in both directions, with traditional from social research but from literary theory, namely Barthes' (1974)
researchers aftempting to explain discursive organizations as the upshot analysis of the sense-making processes at wark in a short story by Balzac.
of the workin1gs of cognitive processes. In general, there is a great deal of scape in this field for canstructive
Discourse analysis should also be highly pertinent when questions of interchange between social researchers and literary theorists (Culler, 1983;
generaliza ti en! and application becorne dominan t. In most cases, when the Harari, 1979; Rorty, 1978; Suleiman and Crosman, 1980).
results of ex~rimental work are checked in naturalistic situations, people's The second likely area of expansion is the study of what has classically
utterances or lwritings are all that is available to the analyst. A proper been known as rhetoric (Billig, 1987), the use of discourse to persuasive
evaluation ofjexperimental results thus depends, perhaps ironically, on effect. This tapie has always been important, and is one where the concems
analysis of discourse. At sorne point event the most rigorously controlled of discourse analysis coincide with those of political scientists (Edelman,
behavioural eJtperiments will need to be backed up with techniques capable 1977; Shapiro, 1981, 1983; Shapiro and Banham, 1983). This field has
of dealing wi\h the kinds of naturalistic records examined in this book. recen ti y received a powerful impetus as a result of research applying the
Similarly with application, as Tajfel (1972) pointed out, to properly apply techniques and theories of conversation analysis to recordings of the
a social psychblogical theory it is necessary to develop a detailed analysis interaction between political orators and audiences (Atkinson, 1983, 1984a,
of the social cbntext in which the application is to take place. The variables 1984b, 1985; Grady and Potter, 1985; Heritage and Greatbatch, 1986).
modelled in ajtheory are only defined in an abstract, operational fashion, There is room here for an approach to the perennial social psychological
and may be manifested in a variety of different ways when applied, issue of persuasion in terms of the process of interaction between pro-
depending onl specific features of the social context. Put simply, what ducer and recipient, and the detailed linguistic format of the persuasive
counts as, saY, helpfulness in one context may become something very text.
different in ahother (Van der Vlist, 1982). Yet, as with evaluation, the The third area of expansion encourages discourse analysts to address
detailed analybs of the social context, very largely, reduces to an analysis concems in the arena ofideology. Thompson (1984) has pointed out three
of participantb• discourse. However precise and clear-cut the theory, if central ways in which discourse contributes significant ideological con-
it is to be pro~erly applied in a practica! context, the researcher or practi- sequences. lt can be used in the presentation of situations of domination
tioner must d~ with naturally occurring discourse with all the complexities and exploitation as legitimate and proper, to mask the existence of these
that it involvb. situations, and to reify current social relationships as natural or, indeed,
! essential. This is an enormously complicated area which raises many
considerations outside the detailed analysis oftexts. However, a number
1

Future directfons of studies have tried to tackle questions of discourse with ideological
import, for example, in the political representations of the 'the Orient'
i •
Someone is supposed to have asked Humphrey Lyttleton, the great JUZZ and middle-eastem politics (Said, 1978, 1981), in the organization of
musician, wh~ere jazz was going, and he replied 'If I knew where jazz accounts of employment opportunities (Wetherell et al., 1987), and the
was going I \vould be there already!' 1dentifying future directions for justification of racial disadvantage (Trew, 1979; Wetherell and Potter,
discourse anilysis is somewhat similar. As the work develops it throws forthcoming a).
up its own bfdnd of problems and suggests exciting and interesting new One of the advantages of discourse analysis is that the data are every-
avenues of rJsearch. Nevertheless, it is possible to discern three areas where- in conversations, on television, in the newspapers, on advertising
in which thefe is increasing interest. hoardings - and the resources needed to start work are minimal. If we
The first isl in the painstaking examination of camplex, worked-over, have indicated the interest and value of the systematic analysis of accounts
written texts.l There are an increasing number of studies which are con- of all kinds this book will have succeeded in its aim. The most exciting
cerned with tbe role of these kinds of texts (children's stories, joumalism, developments, as always, lie in the future.
democratic te~ts, other polit~cal texts etc.) as potent con~tructions of social
life which, t~rough the ach¡evement of particular verswns of the world,
encourage daluations and actions from readers (Adams, 1986; Chua,
1979; Elgin, \ 979; Kress, 1983, 1985; Kress and Hodge, 1979; Kress and
! 1

AppendLt 189

Square brackets indicate that sorne transcript has been delibe!ately omitted.
Material in square brackets is clarificatory information, g.:e
Appendix: transcription notation
A: Brian [the speaker's brother] said [ ] it's okay

The fonn of notation used throughout this book was developed by Gail
Jefferson. A more complete description is found in Atkinson and Heritage
(1984).
Extended square brackets IJlark overlap__ p¡;:_tV(_e_e_n_ utterances, e.g.:

A: Right [so you


B: I'm not sure

An equals sigo at the end of a speaker's utterance and at the start of the
next utterance ind1cates the absence of a discernable gap, e.g.:

A: Anyway Brian=
B: =O ka y, okay
-vc.:.t.>¡¡t,- 1 '~ •

Numbers in brackets indicate pauses timed lo the nearesftenth of a second. '!.


A full stop in brackets indica tes a pause which is notidlblé'but too short
to measure, e.g.:

A: 1 wcnt (3.6) a lot furthcr (.) than 1 intended

One or more calons indica te an extension of the preceding vowel sound,


e.g.:

A: Yca::h, 1 see::

Underlining indicates that words are uttered with addedemphasis; words


in capiials are uttered louder than the surrounding talk, e.g.:. . .

A: It's not right, not right AT ALL


-¿ ¡j (f" !f
A full stop befare a word or sound indicates an audible intake of breath,
e.g.:

A: I think .hh I need more

Round brackets indicate that material in the brackets is either inaudible


or there is doubt about its accuracy, e.g.:

A: I (couldn't tell you) that

! ,, !
1
References 191

Bnrthes, R. (1964) Elements of Semiology. New York: Hill und Wnng,


1 Bnrthes, R. (1972) Mytlwlogies. London: Paladio.
References Bnrthes, R. (1974) SfZ. London: Cape.
Bnrthes, R. (1985) 111e Fashion System. London: Cupe.
Bnumcisler, R. F. (1982) 'A Se\f-Presentntionnl View ofSocia\ Phenomcna' Psychnloglcal
Abraham, C. (198fl 'Problems in Cntegorizing Content: A Discussion ofthe Limit.s ofContent Bulletin, 91: 3-26. '
Annlysis', Pnper prcsented al the British Psycho\ogicnl Society, London Conference, Bazerman, C. (1984) 'Modern Evolution ofthe ExperimentaJ Physics Report' Social Smdies
Dccember. 1 of Science, 14: 163-96. '
Adnms, J. (1986)1 TI1e Con:rpiracy ofrhe Text: 1J1e Place of Narra1ive in tire Development Benttic, G. W. (1983) Talk: Analysis of Speech and Non-Verbal Behaviour in Conversa-
of Thought. ~ondon: Routledge and Kegnn Pnul. tian. Milton Keynes: Open University Press.
Allen, V. L. andi D. A. Wilder (1979) 'Group Categorizntion and Attribution of Belief Becker, H. S. (1963) Outsiders: Studies in the Sociology of Deviance. New York: The Free
Similarity', sJnall Group Behaviour, 10: 73-80. Press.
Allport, G. W. ( 1935)
' 'Attitudes', in C. Murchison (ed.), Handbook of Social Psychology Berelson, B. (1971) Content Analysis in Communication Research. New York: The Free
Vol. 2. Worchster, Mnss.: Clark University Press. Press.
1

Althusser, L. (19 71) For Marx. London: Allen Lnne. Bergcr, P. L. nnd T. Luckmunn ( 1966) 111e Social Constmcrion of Reality. Harmondsworth:
Anscombe, G. Ej M. (1957) lnrention. Oxford: Blnckwell. Penguin.
Argyris, C. (1975) 'Dangers in Applying Results from Experimental Social Psychology', Best, E. _{~922) Spin'tual and Mental Concepts ofthe Maon', Wellington 1 New Zea\and:
American Ps)ichologist, 30: 469-85, Domm10n Museum Bulletin No. 2.
1 Billig, M. (~985) 'Prcjudicel Categorizntion nnd Particulurization: From a Perceptualto
Ashmorc, M. 0¡985) 'A Question of Reflexivity: Wrighting Sociology of Scicntific
1
Knowledgc • ;t.Jnpublishcd D. Phil Thcsis: University of York. u Rhetoncal Appronch', Eurapean Joumal of Social Psychology, 15: 79-103.
Atkinson, J. M. (jt983) 'Two Deviccs for Gcnernting Audience Approval: A Comparative Billig, M .. (1987) Arguing and T11inking: A Rheton'cal Appraaclr ro Social Psyc/rology.
Annlysis of ~ublic Discoursc nnd Texts', in K. Ehlich and H. van Ricmsdijk (eds), Cambndgc: Cambridge University Press.
Connectedness in Selllence, Discourse and Te..'Ct, Tilburg: Tilburg Papers in Linguistics. Bogdnn, R. nnd S. J. Tnylor, (1975) lntroductian to Qualitative Researc/r Methods New
Atkin!ion, J. M. d984a) Our Mastcr's Voices: 71zc lnnguage and Body lnnguagc ofPalitics. York: Wiley. ·
'
London: Methuen. Bourhis, .R. Y. and H. Giles (1977) 'The Lungunge of Intergroup Distinctiveness', ¡0
Atkinson, J. M. ~l984b)
' 'Public Speaking nnd Audience Responses: Sorne Techniques for H. Giles (ed.), Lnnguage, Etlmicity and lnrergroup Relations. London: Acndemic
lnviting Applause', in J. M. Atkinson and J. C. Hcritnge (eds), Stmcmres aJSocial Acrion: Press.
Srudics In c+ver.mtion Analy.rls. Cambridge: Cambridge University Press. Bower, G. H. ~.nd R. K. Cirilo (1985) 'Cognitive Psychology nnd Text Processing', in
Atkinson 1 J. M. (!1985) 'Refusing lnvited Applnusc: Preliminnry Observations from a Case T .. A. van DIJk (ed.), Handbook of Discourse Analysis Vol. l. London: Acndemic Press.
Study of Chrulsmntic Orotory', in T. A. van Dijk (ed.), r1 Handhook of Discourse Analysis Brnnm?nn, .A. (1981) Tire Social Ba.ris of Scientific Discoveries, Cambridge: Cambridge
Vol. 3. New IYork: Acndemic Press. Umversrty Press.
Atkinson, J. M. ahd P. Drew (1979) Ordcr it1 Court: 11w Organiz.ation ofVerballnteraction Brazil, D. (198!) 'The Place of Intonation in n Discourse Modcl', in M. Coulthard and
in Judicial sJuings. London: Mncmillnn. M. Montgomery (eds), Srudies in Discourse Analysis. London: Routledgc nnd Kegan Paul.
Atkinson J. M. ~nd J. C. Hcrit.nge (eds) (1984) Stmcturcs of Social Action: Studies in Brenner, M. (1985) 'Survey lnterviewing', in M. Brenner, J. Brown nnd D. Cuntcr (eds),
Conver.rationiAnalysls. Cambridge: Cambridge Univcrsity Press. 11Je Researc!J lntef>liew: Uses and Approac!Jes. London: Acndemic Press.
Atkinson, P. (19~5) 'Talk nnd Idcntity: Sorne Convergences in Micro-Socio\ogy', in H. Brewer, M. B., V. Dull and L. Lui (1981) 'Perception of the Elder\y: Stcreotypes as
J. Hellc and S. N. Eisenstndt (cds), Micro-Sociological T11eory. London and Bcver\y Prototypcs', Joumal of PersoTUJ/ity and Social Psyc!Jology, 41: 656-70.
Hills, Culiforinia: Snge. Broadbent, G., R. Bunt nnd C. Jcncks (eds) (1980} Signs, Symbols and Arcl!itecrure. New
York: Wiley,
Austin, J. (1961) lA Plea for ExcuSes', in J. D. Urmson nnd G. Wamock (eds), Philosophical
1' Paper.s. Oxf9rd: Clurcndon Press. Brown, D. und S. Levinson (1978) 'Universuls in Lungunge Use: Politeness Phenomenn'
Austin, J. (1962) How ToDo 111ings IYilh IYords. London: Oxford University Press. in E. Goody (ed.), Questions and Politeness: Scrategies in Sociallmeraction. Cambridge;
Bn.inbridge, L. (1$85) 'lnfening from Verbal Report.s to Cognitive Processes', in M. Brenner, Cambridge University Press.
J. Brown nnd¡D. Cunter (eds), 11zc Research llltef>liew: Uses and Approaches. London: Brown, .o. nnd G. Yule (l983) Discourse ATUJ!ysis. Cambridge: Cambridge University Press.
Acndemic Press. Brunsw1k, E. (1955) 'Representntive Design nnd Probnbilistic Thcory in a Functional
Bnr-Hillel, Y. 0:954) 'Indexical Exprcssfons' 1 Mind, 63: 359-79. Psychology', Psychological Review, 62: 193-217.
Barjonct, P. E. (jl980) 'L'intluence socinle el des representntions des causes de l'accident Butt.on, G. nnd .N. Cnscy (1984) 'Generating Tapie: The Use of Tapie Initial Elicitors',
ILc
de In route' 1 Travail Humain, 43: 243-53. In J. M. Atkinson nnd J. Herit.nge (eds), Stmctures ofSocíal Action: St1J.dles in Conversa-
tion Analysis, Cambridge: Cambridge University Press.
Bnrnes, B. (1982) T. S. Kuhn and Social Science. London: Macmillan.
Sumes, B. nnd J .ILaw (1976) 'Whatevcr Should be Done with lndex.ical Expressions?, 71wory Cannell, C. F. and R. L. Kuhn (1968) 'Interviewing', in G, Lindzey andE. Aronson (eds),
rmd Saciery, j3: 223-37. 11Je Handhook ofSocial Psychology (2nd Edition) Vol. 2. Reading, Mass: Addison-Wesley.
1

192 Discourse and social psychology References 193


i
Cantor, N. and W. Mischel {1977) 'TrnilS as Prototypes: EffeclS on Recognition Memory', van Dijk, T. A. (1984) Prejudice in DiscourJ:e: An Analysis of Erlmic Prejrjdices In Cognition
In L. Berkowitz (ed.), Advances in E:cperimcmal Social Psyclwlogy Vol. 9. London: . ond Co~zversarion. Amsterdam: John Benjamins. !

Acadcmic Press. :d;¡\~Van Dijk, T. A. (cd.) (1985) Handhook of Discour.1·e Ana/ysis Vals 1-4. lLondon: Acndemic
Cantor, N. and W. Mischcl {1979) 'Prototypes in Pcrson Perception', in L. Berkowitz ;'lf ~~ Press. !
t..~ 1

(ed.), Advances in Er:pcrimcmal Social Psycholvgy Vol. 12. London: Academic van Dijk, T. A. and W. Kintch (1983) Srrategies of Discourse Compr~hen.sion. London:
Press. Acntlemic Press. _ 1

Cattell, R. B. (1966) 77w Scientijic Analysis of Persvnality. Chicago: Aldine. Drew, P. (1978) 'Accusations: The Occasioned Use of 'Religious Geogrnphy' in Describ-
Chalmers, A. (1980) What is 17Jis 17zing Called Science? Milton Keynes: Open Univcrsity ing Events', Sociology, 12: 1-22. 1

Press. Drew, P. {1984) 'Speakers' Reportings in lnvitation Sequences', in J.¡ M. Atkinson and
Chomsky, N. (1965) A.rpects of a 1J¡eory of Syntar:. The Hague: Mouton. J. Heritnge (eds), Stmctures of Social Action: Swdies in Convl!r.wrion Analysis.
ChomsJ..-y, N. ( 1966) Canesian Linguistics: A Clwpter in the Hiswry of Rationalist TJwught. Cambridge: Cambridge University Press. 1

Ncw York: Harper and Row. Drew, P. ( 1985a) 'Analyzing thc Use of Langunge in Courtroom Intcrnclion', in T. A. van
Chua, B. H. (1979) 'Democrncy as a Textual Accomplishment', TJ¡e Sociologica/ Quanerly, Dijk (ed.), Handbook of Discmme Analysis Vol. 3. London: Acn~emic '
Press.
20: 541-9. Drew, P. (1985b) 'Sorne Properties ofCompeting Descriptions of EvcntS'. Pnper presented
Cicourel, A. V. (1974) 1Jteol)•and Metlwd in a Swdyof Argemine Fenility. New York: Wilcy . al the British Psychological Society Social Psychology Section Confcfencc, Cumbridgc,
.-•"'ii Clarke, D. {1983) IAnguage and Action: A Srmctural Model oJBehaviour. Oxford: Pcrgamon. Scptember. ¡

Collett, P. (1977) 'Rule as a Scientific Conccpt', in P. Collett (ed.), Social Rules and Social Drew, P. (1986) 'A Comment on Taylor and Johnson', British Joumal offocial Psycho/ogy,
Behal•iour. Oxford: Blackwell. 25: 197-8. 1

Collins, H, M. (1974) 'The TEA Set: Tncit Knowledgc and Scientific Networks', Science Duncan, B. L. (1976) 'Differential Social Perccption and Attribution of Intcrgroup Violencc:
Smdies, 4: 165-86. Testing the Lower Limits of Stereotyping of Blacks', Joumal of Pedanalitv ami Social
Collins, H. M. (1983} 'The Meaning of Lies: AccounlS of Action and Purticipatory Research', Psychology, 34: 590-8. 1 .

in G. N. Gilbert and P. Abell (cds). tlccmmts nnd Action. Aldcrshot: Gower. Edclman, M. (1977) Politica/ Lnnguage: Words that Succeed mrd Polic.Jies that Fail. Ncw
Collins, H. M. (1985) Clwnging Order: Replication and lnducrion in Sdentiftc Praclice. York; Academic Press. 1
London and Beverly Hills, California: Sage. Elgin, P. (1979) 'Resolving Reality Junctures on Telegrnph Avenuc: A Study of Pructical
Cooley, C. H. (1902) Human Narure and rile Social Order. New York: Scribner's. Reasoning', Canadian Joumal of Sociology, 4: 359-75. 1

1
Coulter, J. (1979a) 77w Social Constmcrion af Mind. London: Mncmillnn. Elms, A. C. (1975) 'The Crisis of Confidcncc in Social Psychology', American PsyclwloKüt,
Coulter, J. (1979b) 'Bclicfs and Prnctical Underslanding', in G. Psnthas {cd.), Evt:l)•day 30: 967-76. 1

Languoge; Studies in Etlmomethodology. New York: Irvington. Eysenck, H. J. (1953) 77w Stmcture of Human Personaliry. Ncw Yorf: Wiiey.
Coultcr, J. (1983) Retliinking Cognitive 17wal)'. London: Macmillan. Eyscnck, H. J. nnd S. B. Eyscnck ( 1964) Manual of rlw Eysenck Per.ronaUo• lnvelltofJ',
Coultcr, J. (1985) 'Two Concepts nflhc Mental', in K. J. Gcrgcn und K. E. Duvis (eds), London: Hodder und Stoughton. 1

71ze Social Cmwruction of tite Person. Ncw York: Springer-Verlag. Farr, R. M. (1984) 'Social Reprcsentalions: Thcir Role in thc Design· ami Exccution of
Coulthnrd, M. (1977) An /ntroduction to Discourse Analysis. London: Longman. Laboratory ExperimenlS', in R. M. Farr and S. Moscovid (eds), Socibi Rcpresemation.r.
Coward, R. (1984) Femnle Desire. London: Paladio. Cnmbridge: Cambridge University Prcss. 1
'
Coward, R. and 1. Ellis ( 1977) Úlnguage and Materiolism: Developmems in Semiolagy and Fnrr, R. und S. Moscovici, (eds) (1984) Social Representations. CamOridge: Cambridge
the 111eory of the Subject, London: Routlcdge and Kegan Paul. University Press. 1

Van Cranach, M. and R. Harré (cds) (1982) 71w Analysis af Action. London: Academic Prcss. Festingcr, L. (1954) 'A Theory of Social Comparison Processes', ~wtwn Relariom,
Crnne, D. (1980) 'An Explorntory Study ofKuhninn Pnradigms in High Energy Physics', 7: 117-40. 1

Social S1udies af Science, JO: 23-54. Festinger, L. (1957) A 1Jzeory ofCogniti\'e Dissonance. Evnnston, I11inois: Row, Peterson.
Cuff, E. C. (1980) 'Some lssucs in Studying the Problem ofVersions in Everyday Situations', Feyernbend, P. (1975) Againsr Metlwd. London: NLB. 1

Occasianal Paper No. 3. Oepartment of SOciology, University of Mnnchester. Fishbcin, M. and l. Azjen (1975) Be/ief, Attitude, lntemion and Belwviour: An lmroducrion
Cuff, E. C. and D. E. Hust\cr (1982) 'Storics and Story-Time in an Infant Clnssroom: Sorne to 71wory and Research. Reading, Mass: Addison-Wesley. 1

Features of Languagc in Social Interaction', Semiotica, 42: 119-45.


1

Fiske, J. antl J. Hartley (1978) Reading Te/evi.rion. London: Methucn •

Culler, J. (1975) Srrncturalist Poetics. London: Routledge and Kegun Paul. Forgas, J. ( 1979) Social Episodes: 71re Swdy of lnteraction Rowine.s. Londdn: AC'udcmic Press.
Culler, J. {1976) Saussure. London: Fontana. Foucault, M. (1970) 77w Order of17rings. London: Tavistock. 1

Culler, J. (1983) On Deconstruction. London: Routledge and Kcgnn Paul. Foucault, M. (1971) 'Orders of Discourse', Social Science /nformatiofr, 10: 7-30.
Dahrendorf, R. (1973) Hamo Sociologicu.r. London: Routlcdge and Kegan Paul. Foucault, M. (1972) 77ii! Arclweology of Knowledge. London: Tavist9ck.
Davidson, J. (1984) 'Subsequent Versions of lnvitations, Offers, Requests and Proposa\s Foucnult, M. (1981) 77ze Histof)'ofSe:waliry, Vol./; Anlntroducrion.ILondon: Penguin.
Dealing with Potential or Actual Rejection', in J. M. Alkinson nnd 1. Heritnge (eds), Frederickson, C. H. (1986) 'Cognitive Models and Discourse Analysis~, in C. R. Cooper
Stmcmres of Sociul tlction: Smdies in Conversmíon Analysis. Cambridge: Cambridge and S. Greenbaum (eds). Swdying IVrirtng: lin!{uisric Appmaclws. Lpndon nnd Bcvcrly
Univcrsity Prcss. Hills, California: Snge. 1
Denzin, N. K. ( 1978) 1711! Researclt Acr (2nd edition). Chicugo; Aldinc.

1
194 Discourse land social psychology References 195 \
1

Frecdle, R. O. (ed.)l(1977) Advances in Discourse Processes. Norwood, N. J.:.Ablex.. Hammersley, M. and P. Atkinson (1983) Ethnography: Principies and Practice. London:
Garfinkel, H. (1967)1 Smdics in Ethnomelhodology. Englcwood Cliffs: Prentice Hall. Tavistock. ·
Gnrfinkel, H. (1974)1 'On the Origins of the Tenn "Ethnomethodology" ', in R. Turner Harari, J. V. (ed.) (1979) Textup/ Strategies: Esiays in Post-Strncturalist Criticism. London:
(ed.), EiJmometh?dology. i-IIH111ondsworth:. Penguin. . Methuen.
Garvey, C. (1984) C(hildren 's Talk. London: Fontana. Harré, R. (1974) 'Sorne Remnrks on "Rule" as a Scientific Concept', in T. Mischel (ed.),
Gergen, K. J, (1973} 'Social Psychology as History', Joumal of Personaliry and Social Understanding Other PersohS, Oxford: Blnckwell.
Psyclwiogy, 26: ~09-20. Hnrré, R. (1977a) 'Rules in the Explnnntion of Social Behnviour,' in P. Collett(ed.), Social
Gergen, K. J. (1978} 'E:tperimentntíon in Social Psycho\ogy: A Renpprnisal', European Rules and Social Behaviour. Oxford: Blackwell.
Joumal of Social Psyclwlogy, 8: 507-27. Hnrré, R. (l977b) 'The Ethogenic Approach: Theory nnd Practice', in L. Berkowitz
Gcrgen, K. J. (1982) Toward Transfomlation in Social Knówledgc. New York: Springer-Verlag. (ed.), Advances in Erperimental Social Psychology Vol. 10. London: Academic
Gergen, K. J. (1985) ' 'Social Constructionist Inquiry: Context nnd Implicntions', in Press.
K. J. Gergen nndiK. E. Davis (cds), 111e Social Construction ofthe Person. New York; Harré, R. (1977c) 'The Self in Monodrama', in T. Mischel (ed.), 71te Self: Psycltological
Springcr-Vcrlag. · and Philosophícallssues. Oxford: 8\ackwell.
Gergcn, K. J. (forthcoming) 'Warrnnting Voice nnd thc Elnboration of lhe Selr, in J. Shotter ~jhrré, R. (1979) Social Being: A Theory Jor Social Psychology. Oxford: Blackwell.
and K. J. GergeÓ (eds), Texts of ldentity. Greenwich, Conn.: JAI Press. Harré; R. (1983) Personal Being: A 11~eory for Individual Psychology. Oxford: Blackwell.
Gergen, K. J. and I<J. E. Davis (eds) (1985) 11w Social Construction of the Person. New Harré, R. ( I985a) 'Situationnl Rhetoric nnd Self-presentntion', in J. Forgas· (ed.), Language
York: Springer-Yerlag. and Social Situations. New York: Springer-Verlug.
Di Giacomo, J.-P. q980) 'lntergroup Alliance and Rejections Within a Protest Movement Hurré, R. (1985b) 'The Language Game of Self-Ascription: A Note', in K. J. Gergen and
(Annlysis ofSoci~ RCpresentntions)', European Joumal o] Social Psyclw/ogy, 10: 329-44. K. E. Davis (eds), 71te Social Constrnction ojthe PÚson. New York: Springer-Verlng.
Gilbcrt, G. N. (1980) 'Being Interviewed: A Role Annlysis', Social Science lnformation, Harré, R. (1986n) 'Selves in Tnlk', British Journal of Social Psycho/ogy, 25: 271-3.
19: 227-36. . 1
Harré, R. (1986b) Varieties of Realism. Oxford: Blackwell.
Gilbert, G. N. nnd t1· Mulkuy ( 1980) 'Contexls of Scientiflc Discourse: Social Accounting Hnrré, R. (forthcoming) 'Problcms of Self-reference nnd the Expression of Selfhood', in
in Experimental Pupers', in K. Knorr-Cetinn, R. Krohn und R. Whitley (eds), 111e Social J. Shotter and K. J. Gergen (eds), Texts of ldentity. Greenwich, Conn.: JAI Press.
Proce.fs of lnvesfi¡:ation. Dordrecht: R.cidcl. -y!Jarré, R. and P. F. Secord (1972) 111e Explanatiorl ofSocial Behaviour:Oxford: Blackwell.
Gilbert, G. N. und M!. Mu\kuy (1982) 'Wurrnnting Scientific Belier, Social Smdies ofScience, -{Harré, R., D. Clurkc und N. de Cario (1985) Motives and Mechanisms: Anlntroduction
12: 382-408. 1
to the Psychology of Action. London: Melhuen.
; Gilbert, G. N. and M. Mulkny (1984} Opening Pandora's Box: A Sacialogical Analysis Harris, R. (1980} 111e Language Makers. London: Melhuen.
· of Scientists' Di.fcourse. Cambridge: Cambridg~ University Press. Hartley, J. (1982) Understanding News. London: Methuen.
Giles, H. (ed.) (197p) IAnguage, Ethnicity and lntergroup Relations. London: Academic Henlh, S. (1982) 111e Sexual Fix. London: Macmillnn.
Press. 1 Hebdidge, D. (1979} Subculture: 11w Meaning of Sryle. London: M~thuen.
Giles, H. and R. ~- St. C1air (cds) (1979) IAnguage and Social Psychalqgy. Oxford: Heelas, P. andA. Lock (eds)"{l981} lndigenous Psychdlogies. London: Acndemic Press.
Blackwell. ! . Heider, F. (1958) 111e Ps}'chology of lmerpersonal R_elations. New York: Wiley.
Goffman, E. {1959)171~e Presentation of Self in Everyday U fe. Harmondsworth: Penguin. Henriques, J., W. Hollwuy, C. Urwin, C. Venn nnd v:Wa1kerdine (1984) Changing the
Goffman, E. (1961) Asylums. Harmondsworth: Penguin. Subject: Psychology, Social RegulaÚon and Subjectiviry. London: Methuen. .
Goffman, E. (1971) Relations in Public: Micro-Smdies ofthe Public Order. Harmondsworth: Heritage, J. (1978) 'Aspects of lhe Flexibilities of Natural Language Use', Sociology, 12:
Penguin. 1
79-105.
Goffmnn, E. (1981) Forms afTa/k. Oxford: Blackwell. Heritage, J. (1984) Garjinkel and Etlmomethodology. Cambridge:- Polity Press.
Gombrowicz, W. {i979) Ferdydurke. London: Boyars. Heritnge, J. and D. Grentbntch (1986) 'Generating Applause: A Study of Rhetoric and
Grady, K. and J. Poder (1985) 'Speaking and Clapping: A Comparison of Foot and Thatcher's Response al Party Politica1 Conferences, American Socio/ogícal Review, 92: Il0-57.
Orntory', I.nngdage and Communication, 5: 173-83. Herzlich, C. (1973) Health and lltness: A Social Psychological Perspedive. London: Academic
Grice, M. P. (1975)1'Logic und Co~versation', in P. Cote andJ. L. L. Margan (eds), Synrax Press. .·
and Semantics 3: Speech Acts. Ncw York: Academic Press. Hesse, M. (1974) 11w Stmcrure of Sctenti.fic lnferenÚ. London: Macmillan.
Gusfield, J. (1976) •The Literary Rhetoric of Science: Comedy and Palhos in Drinking Driver Hewitt, J. P. and R. Stokes (1975) 'Disclaimers', Amen·can Sociological Review, 40: \-11.
1 Hewstone, M., J.Jaspars nnd M. Lalljee ( 19.82) 'Social Representations, Social Attribution
Reseurch', American Sociological Review, 41: 16-34.
Halliday, M. A. K (1978) Lnnguage as Social Semiotic. London; Edwnrd Amold. and Social ldentity: The Intergroup Imnges of "Pub\ic" nnd "Comprehensive"
Halliday, Q. and J .1 Potter (1987) 'Community Leaders in Discourse: Lay Envli-oniriCntal Schoolboys', European Joumal of Social Psychology, 12:241-71.
Psychology in Áccounts of a "Riot" ', mimeo, University of St Andrews. Hinde, R. A. ( 1975) 'The Conccpt of Function', in G. Buerends, C. Beer nnd A.. Munning
8 .
Hnmilton, D. L. ( 19~9) 'A Cognitive-Attributionnl Anulysis of Stereotyping', in L. erkowltz
1
(eds). Funclion and Evolwion in Behaviour. Oxford: Clarendon.
(ed.), Advances fn Expen'mental Social Psyclwlogy Vol. 12. New .York: Academic Press. Hodge, R. and D. Tripp (1986) Children arui Television: A Semiotic Approach. Cambridge:
Humlyn, D. W. ( 19?4) 'Person Perception nnd our Understanding of Others', in T. Mische\ Polity.
(ed.), Understahding Other Persons. Oxford: Blnckwell.

1
196 Discourse and social psychology IReferences 197

Holsti, O. R. (1968) 'Content Analysis', in O. Lindzey nnd E. Aronson (eds), Handbook Kuhn, T. S. ( 1977) 1Jw E<sential Tenswn: Se/ected StudieJ in Smnt(jiciTmdttton and Clwngc.
of Social Psyclwlogy Vol. 2. Reading, Mass: Addison-Wes\ey, London: University of Chicago Prcss. 1

Husband, c. (1982) 'lntroduction: "Race", the Continuity of a Conccpt', in C. Husband Kulka, R. A. nnd J. B. Kcssler (1978) 'ls Justicc Rcally Blind? Th~ lnfluence of Litignnt
(ed.), 'Race' in Brltain: Comirwity arul Change. London: Hutchinson. Physical Attractivcness on Juridicinl Judgment', Jmmwl of Appl~ed Social Psyclwlogy,
Ickes, W., E. Robertson, W. Tooke andO. Teng (1986) 'Naturalistic Social Cognition: S: 366-81. 1

Methodo\ogy, Assessmcnt and Vnlidation', Joumal of Personality and Social Psychology, Lnbov, W. (1972n) Language oftlw Jnner Ciry. Philndelphia: Philade¡phia University Press.
51: 66-82. Lab~;~~~Q.( 1972b) 'Sorne Principies of LingUislic Methodology', ~mgua¡:e in Society, 1:
Israel, J. and H. Tnjfel (eds) (1972) T11e Comext ofSocial Psycl!ology. London: Academic
Press. Lnbov, \V. and D. Fanshel\ (1977) 71Jerapeutic Discourse: PsyclwrhJrapy as ConversatiorL
Jacobs, S. and S. Jackson (1983) 'Speech Act Structure in Conversution: Rational Aspccts Academic Prcss: London. 1

of Prngmatic Cohcrcnce', in R. T. Crnig and K. Trncy (eds) Convcrsational Colwrence: Lakutos, I. (1970) 'Falsificntion nnd the Methodology of Scientific Rpsearch Programmes',
Form, Structu.re and Strategy. London and Beverly Hills, California: Sage. in J. Lakntos andA. Musgrnve (eds), Cn'ricism and rl!e Growth of Knawledge. Cambridge:
James, W. {1968) 'The Selr, in C. Gordon and K. J. Gergen (eds), Tiw Selfin Social Cambridge University Press. 1

lnteraction, Vol. 1: Classic and Contemporary Perspecti\•es. New York: Wiley. Lakoff, O. and M. Johnson (1980) Meraphors We U1•e By. Chicugo: Uni~crsity ofChicago Press.
Janis, l. (1972) Victims of Groupthink. Bastan: Houghton Mifflin. Lalljee, M., L. B. Brown and G.P. Ginsberg (1984) 'Atlitudes: Disposition, Behaviour
Jayyusi, L. {1984} Categorization and the Moral Order. London: Routledge and Kegan or Evaluntion', British Journal of Social Psyclw/ogy, 23: 233~.
Paul. Lntour, B. and S. \Voolgar (1979) Lnboratory Lije: 77w Social Consrntction of Sciemific
Jeffcrson, G. (1985) An Ex.ercise in lhe Transcription and Analysis of Laughtcr', in Facts. London nnd Beverly Hills, Cnlifomin: Sage. 1

T. A. van Dijk (ed.), Handbook ofDiscourse Analysis Vol. 3. London: Academic Press. Lnw, J. nnd P. Lodge (1984) Science For Social Scientists. London: Mncmillan.
Johnsan-Laird, P. (1974) 'Experimental Psycholinguistics', in M. R. Rosenzweig and L. Layton, E. (1977) 'Conditions for Technological Development', iniL Spiegel-Rosing and
W. Portar (eds). Annual Review of Psychology Vol. 25. Palo Alto, California: Ann. D. J. de Solla Price (cds), Science, TecJmology and Sociery, Lon¡ion and Beverly Hills:
Revs. Inc. Sage. 1

Jones, E. E. and K. E. Davis (1965) 'From Acts to Dispositions: The Attributional Proccss
1

Lee, J. (1984) 'Innoccnl Victims and Evil·Doers', Women's Studie¡ /mernmional Fomm,
in Person Pcrceplion', in L. Berkowitz (ed.), Ad1•ances in Experimental Social Psyclwlogy 7: 69-73.
Vol. 2. New York. Academic Press. Lcvinson, S. (1983) Pragmarics. Cambridge: Cambridge Universib' Prcss.
1
Kal\en, D. B. P., O. B. Kosse, H. C. Wagenaur, J. J. J. Kloproggc anU M. Vorbeck (eds) Litton, I. untl J. Potter (1985) 'Social Rcpresentntions in the Onli nary Exp\anntion of 11
(1982) Social Science ReJearc/1 and Public Pulicy-Muking. Windsor, Bcrks: NFER· "Riut"', Europea11 Jmmml ofSodlll P.\'yclwlogy, 15: 371-HHf
Nelson. Lock, A. (1981) 'Universnls in Humnn Conceplion', in P. Heehls nnd A. Lock (cds),
Kelley, H. H. (1967) 'Attribution Theory and Social Psychology', Nebraskn Symposium Jndigenous Psyclwlogies. London: Academic Press.
Ofl Moth•mion, 15: 192-238. Lodge, D. (1978) Clmnging Places. Harnmndsworth: Pcnguin.
Kinder, D. R. and D. O. Seurs ( 1985) 'Publie Opinion and Political Action', in G. Lindzcy Lyons, J. (1967) Jmroductiun to 711t!arerical Unguistics. Cambridge:¡ Cambridge UnivcrHity
andE. Aronson (eds), T11e Handbook of Social Psyclwlogy (3rd Edition) Vol. 2. New Press.
1

York: Rundom House. Lyons, J. (1978} T7lt! Jnvention ofthe Self. Carbondale, Illinois: Soulhcm Illinois University
Krnmpen, M. (1983) 'lean:; of the Road', Semioticu, 43: 1-204. Press. 1

Kreckel, M. (i981) Camnmnicarive Act.r and Slwred Knowledge in Natural Di.I'CO/Irse. McCabe, C. (1974) 'Renlism nnd the Cinema: Notes on Some Brec'htian Theses', Screen,
London: Academic Press. 15: 7-27. 1

Kreckel, M. {1982) 'Communicative Acrs and Ex.tralinguistic Knowledge', in M. Van McConnlmy, J. B. (1985) 'Modem Racism, Ambivalence and the Modero Rucism Scnle',
Cranach and R. Harré (eds), 711t! Analysis of Action. Cambridge: Cambridge University in S. L. Gucnner (cd.), Prejudice, Discrimbwtion and Racism:IT11t!ory ami Researc/1.
Press. New York: Academic Press.
Kress, G. (1983} 'Linguistic and ldeologica\ Transformations in Newspaper Language', in MncDonell, D. (1986) 17wories of Discourse: An /mroduction. Oxford: Bluckwell.
H. Dnvis and P. Wahon (eds}, l.nnguage, Jmage, Media. Ox.ford: Blnckwe\1. McGuire, W. J. (1973) 'The Y in and Yang of Progress in Socinl Psybhology: Seven Koan',
Kress, G. (1985) 'ldeo\ogicnl Structures in Discourse', in T. A. van Dijk (ed.), Hm1dbook ., Journal of Personality and Social Psychology, 26: 446-57. 1

of Discaurse Analysis Vol. 4. London: Acndemic Press. U :-rJicGuire, \V. J. (1985) 'Altitudes and Attitude Changc', in G. Lindzey nnd E. Aronson
Kress, G. and R. Hodge (1979) Lnnguage as Jdeology.- London: Routledge and Kegan 0 (eds}, Hnndbook ofSocinl Psychology (3rd Edition) Vol. 2. NewiYork: Random House.
Paul. McHoul, A. \V. (1982) Telling How Text.\' Talk: Essays 011 Reading qnd Etlmomethodology.
Kress, O. and T. Trew ( 1978) 'Ideological Trnnsformations of Discourse: Or How the Sunday London: Routledge and Kegan Paul. 1

1imes Gol its Message Across', 'fl1e Socio/ogical Review, 26: 755-76. McKinlay, A. and J. Potter (1987} 'Model Discourse: Interpretative Repertoires in Scientists'
_....--l$-roger, R. O. (1982) 'Ex.ploralions in Ethogeny: With Special Referencc to the Rules of .;;_'Q Conference Tal k', Social Studies of Science. 1

- Address', American Psyclwlogist, 37: 8l0-20. McKinlny, A. nnd J. Potter (forthcoming} 'Social Representations: A Conceptunl Critique',
Kuhn, T. S. (1962) 71JeStmcmreofScientijic Revalmions. Chicago: University ofChicago Prcss. Jau mal for the 17Jeory of Social Beha1•iour.
198 Discou(se and social psyclzology References 199
!
McLellnnd, D. a. (1961) 17u! Achievíng Sociery. Princeton: Van Nostrnnd. Mulkny, M. ( 1981) 'Action, Belief or Scientific Discourse?', Philosophy ofthe Social Sciences,
Mursh, A. (197~) 'Who Hntes the Blncks?', New Socíery, 23 September, 649-52. JI: 163-71.
Mnrsh, P. (1982) j'Ruh:s in the Organization of Action: Empiricnl Studics', in M. van Cranuch Mulkay, M. (l984a) 'The Ultimute Compliment: A Sociological Analysis of Ceremonial
und R. Hnrré (eds), 17u! Analysis of Action. Cambridge: Cambridge University Press. Discourse', Sociology, 18: 531-49.
1 Mulkny, M. (1984b) 'The Scientist Tnlks Back: A One Act Play, with a Moral, nbout
Mursh, P., E. Rosser und R. Hurré (1978) 17ze Rules ofDi.mrder. London: Routledgc und
Kegan Pnul.l Replicntion in Science nnd Reflexivity in Sociology', Social Smdics ofScience, 14: 265-82.
Maslow, A. H. (1 968) Taward a PsychologyofBeing (2nd Edition). Princeton: Van Nostrand. Mulkay, M. (1985) 77Je Word and rhe World: Explorations in the Form of Sociological
1
Mut.son, F. W. (1973) Wirhour/Within: Behaviourism and Hwnanism. Monterey, Califomiu: Analysis. London: Allen Unwin.
Brooks/Colc.l Mulkay, M. nnd G. N. Gilbert (1981) 'Putting Philosophy to Work: Karl Popper's Influcncc
Mcud, G. H. (1~34) Mind, Selfand Society. Chic:ugo: University of Chicugo Press. on Scientific Prnctice', Philosophy of rhe Social Sciences, 11: 389-407,
Mchun, M. (1979) Leaming Lessons: Social Organizatian in the Classroom. Cambridge, Mulkny, M. nnd G. N. Gilbert (1982a) 'Joking Apnrt: Sorne Recommendations Conceming
Mass.: Harvdrd University Press. the Annlysis of Scientific Culture', Social Studies of Science, 12: 585-615.
Merritt, M. (1976) 'On Questions Fol\owing Answers (in Service Encounters)', Lnnguage Mulkay, M. and G. N. Gilbert (1982b) 'Accounting for Error: How Scientists Construet
in Saciety, 5~ 315-57. Their Social World When They Account for Correct and Incorrect Belief, Sociology,
Merton, R. (1973) TI1e Saciology af Science. Chicago: University of Chicago Press. 16: 165-83.
Milgram, S. (19p3)
' 'Behnvioural Study of Obedience', Joumal of AbnomJal and Social Mulkny, M. nnd G. N. Gilbert (1983) 'Scientists' Theory Tnlk', Canadian Jouma/ of
Psychology, 67: 371-78. Sociology, 8: 179-97.
Miller, G. A. (1980) 'Giving Awny Psychology in the 1980's', Psychology Today, 13: 38-50, Mulkay, M. nnd G. N. Gilbcrt (1985) 'Opening Pnndorn's Box: A New Approach to the
97-98. i
Sociological Analysis of Theory Choice', Knowledge and Sociery, 5: 113-39.
Milis, C. W. (1940) 'Situnted Actions nnd Vocabularies of Motive', American Sociological Mulkny M., J. Potter and S. Yearley (1982) 'Why an Analysis of Scientific Discourse is
Review, 5: 9Ü4-13. Needed', in K. D. Knorr-Cetina and M. Mulkay (eds), Science Observed: Perspectives
Milroy, L. (198Ü) Lnnguage and Social Networks. Oxford: Blnckwell. on the Social Study of Science. London and Beverly Hills, California: Snge.
Mischel, W. (19~8) Personality and Assessment. New York: Wilcy. Myers, G. (1985) 'Texts ns Knowledge Claims', Social Studies of Science, 15: 593-630.
Mixon, D. (1972) 'Instcad of Deceptlon', Joumal for rhe 11Jeory of Social Behaviour, Nisbett, R. E. and T. D. Wilson (1977) 'Telling More Thnn We Cnn Know: Verbal Reports
2: 145-77. 1
on Mental Processes', Psyclwlogical Review, 84:231-59.
Moscovici, S. ( 1976) Ln psyclwanalyse: son image et son public. (Revised Edition}. Pnris: Ome, M. T. (1969) 'Demand Chnraeteristics and the Concept of Quasi-Controls', in
Presses Univbrsitnircs de Frunce. R. Rosenthnl and R. L. Rosnow {eds), Artifact in Behavioural Research. New York:
_, --j~oscovici, S. c[981) 'On Social Represcntntion', in J. Porgas, (ed.), Social Cognition: Aeadcmic Press.
:/ Perspectivcs rn Everyday Undersranding. London: Acndemic Press. Pnrker, l. (forthcoming a) 'Discourse and Power', in J. Shotter and K. Gergen (eds), Texts
_L Moscovici, S. (1982) 'The Coming Era of Representntions', in J. P. Codal and J. P. Lyons of ldentiry. Greenwich, Conn.: JAI Press.
. (cds), Cognifive Analysis of Social Behaviour. The Haguc: Nijhoff. Parker, l. (forthcoming b) 'Socinl Reprcsentations: Social Psyehology's (Mis)Use of
...,.--Moscovici, S. Ó984a) 'The Phenomenon of Social Representations', in R. Farr and Sociology', Journal for rhe 17Jeory of Social Behaviour.
S. Moscovicij (cds), Social Represcntations. Cambridge: Cambridge Univcrsity Press. Pecheux, M. (1982) Language, Semantics and Ideology: Stating the Obvious. London:
Moscovici, S. {19B4b) 'The Myth of the Lonely Purudigm: A Rejoinder', Social Rcscarch, Macmillnn.
1 . 51: 939-67.¡ Perls, F. (1971) Gestalt TI1erapy Verbatim. New York: Bnntam.
_ . ,~ Moscovici, S. ( 1985) 'Comment on Poner nnd Litton', Brirish Joumal oJSocial Psyclzology, Peters, R. S. (1959) TI1e Concept of Marivarion. London: Routledgc and Kegan Paul.
24: 91-3. 1
Pettigrcw, T. F., G. W. Allport andE. V. Bnmett (1958) 'Binocular Resolution and Perception
\Moscovici, S. a1d M. Hewstone (1983) 'Social Rcpresentntions nnd Social Explnnutions: of Rnce in South Africn', British Jaumal of Psychology, 49: 265-78.
From the "N.pvc" to the "Amateur" Scientist', in M. Hewstone (ed.), Attriburion T11eory: Plummcr, K. (1983) Documems of Ufe. London: Allcn Unwin.
Social and Funcrional Extensions. Oxford: Blnckwell. Polynni, M. (1958) Personal Knowledge. London: Routledge nnd Kegnn Pnul.
Mostyn, B. {1985)' 'The Content Analysis of Qualitative Resenrch Datn: A Dynumic Pollner, M. (1974) 'Mundanc R.casoning', Phílosaphy ofthe Social Sciences, 4: 35-54.
Approach', iA M. Brenner, J. Brown and D. Canter (eds), 77w Rescorch /nten1iew: Uses Pollner, M. (1975) 'The Vcry Coinnge ofYour Brnin: The Anntomy ofRenlity Disjunctures',
'
ond Approac(1es. London: Acndemic Press. Philosophy of the Social Sciences, 5: 411-30.
Mulkay, M. (19~4) 'Mcthodology in the Sociology of Science', Social Science lnfonnation, Pomerantz, A. (1978) 'Compliment Responses: Notes on the Operation of Multiple Con-
13: 107-19.¡ strnints', in J. Schenkein (ed.), Studies in rhe Organizarían ofConversational/nteraction.
Mulkay, M. (1919a) 'lntcrpretation and the Use of Rules: The Cnse of Norms in Sciencc', London: Academic Press.
in T. F. Gichm (cd.), Science ond Social Strucrure. Transactions of the New York Pomerantz, A. (1984) 'Agreeing nnd Disugreeing with Assessments: Sorne Features of
Acadcmy of !scienccs, Series IIJ, 39: 111-25. Preferred/Dispreferred Tum Shnpes', in J. M. Atkinson and J. Heritage (eds), Stntcfllres
Mulkay, M. ( 19~9b) 'Knowlcdge nnd Utility: Implications for the Sociology of Knowlcdge', of Social Action: Srudies in Conversar ion Analysis. Cambridge: Cambridge University
Social Swdics of Scicnce, 9: 63-80. Prcss.
200 Discourse aJtd social psychology Rljerences 201
'

Pomernntz, A. ( 1986) 'Extreme Case Formulations: A Ncw Way of Lcgitimuting Claims', Sncks, H. (1974) 'On the Annlyzability of Swries by Children', iA R. Turner (cd.),
in G. Bullan, P. Drcw and J. Heritnge (eds) Human Swdies {lnteraction and Language Etlmomerlwdology. Harmondsworth: Pengum.

1
1

Use Special Issuc), 9: 219-30. Sacks, H. (1979) 'Hotroddcr: A Revolutionary Catcgory', in G. Psathas (ed.), Everyday
Popper, K. {1959) 11ze Logic of Scicmijic Disco\'ef)'. London: Hutchinson. Language: Studics in Et!momcthodology. New York: lrvington. 1

Popper, K. (1963) Conjectures and Refurations. London: Routledge and Kegun Paul. Sacks, H. (1984) 'Notes on Mcthodology', in J. Atkinson and J. Hcritage (eds), Strucrurc.1·
Poner, J. {1982)' ''Nothing so Practica[ as u Good Theory'': Thc Problematic Application of ofSocial Action: Swdies in Conversmion Analysis. Cambridge: Cutilbridgc University
Social Psychology', in P. Stringer (ed.), Corifroming Sociallsmes. London: Acudemic Press.
Potter, J. (1984) 'Testability, FlerJbility: Kuhnian Values in Scientists' Discourse Conccrning
~- . 1

Sacks, H., E. A. Schegloffand G, A. Jefferson {19'74) 'The Simplest Systematics for the
Theory Choice', Philosoplry of the Social Sciencel', 14: 303-30. Organization of Turn-Taking in Conversation', J..nnguage, 50: 697-735.
Potter, J, ( 1986) 'Reading Repertoires: A Preliminary Study of Sorne Procedurcs Scientists Suid, E. W. (1978) Oriellfa/ism. New York: Pantheon Books. 1

Use to Construct Rcudings', mi meo under submission, University of St. Andrews. Said, E. \V. (1981) Covering Islam: How Ihe Media and E:rpert.\' DetJrmine How IVe Sce
Potter, J. (1987) 'What is Rcflexive about Discourse Analysis? The Case of Reading tite Rest of rhe World. London: Routledge and Kegan Puul. 1
Rcadings', in S. Woolgar (ed.), Tt1ming tlw Pages ofScience: Knoll'/edge and Refle.tivity. Sampson, E. E. (1983) 'Deconstructing Psychology's Subject', Jo mwl of Mind and
London and Bever[y Hills, Cnliforniu: Sage. Be!zlJI~·our, 4: 135-64. 1

~Ppttcr, J. and L Litton {1985) 'Some Problems Underlying the Theory of Social Reprcscnta- De Saussurc, F. (1974) Course in General Linguistics. London: Fonfuna.
tions', Brirish Joumal of Social PsycJwlogy, 24: 81-90. Schcgloff, E. A. (1968) 'Sequencing in Conversational Openings', .-lmery·can Amhropologisr,
Pottcr, J. und M. Mulkay {1982) 'Making Theory Uscfu!: Utility Accounting In Social 70: 1075-95. 1

Psychologists' Discoursc', Fundamema Sciemiae, 34: 259-78. Schcgloff, E. A. and H. Sacks (1973) 'Opening Up Closings', Semiotica, 7: 289-327.
Potter, J, and M. Mulkay ( 1985) 'Scientists' Intcrvicw Tnlk: Interviews as a Tcchnique for Schlcnker, B. R. (1980) /mpri!S.I'ion Mww¡.:enwnt: '1711! Self-Conce¡n.¡Sncial ldemiry aml
Revcaling Participants' Interpretativc Pmctices', in M. Brcnner, J. Brown and D. Cantcr lnterpersorwl Relarions. Monterev, California: Brooks/Cole.
(eds), 77w Research lnterview: Uses and Approaches. New York: Acadcmic Press. 1
Schlenker, B. R. and B. W. Darby (19-81) 'Use of Apologies in Social P[edicaments', Social
Potter, J. and S, Rcicher {1987) 'Discourses of Community and Conflict: The Organizution Psychology Quarterly, 44: 271-8.
of Social Categories in Accounts of u "Riot" ', British Jau mal of Social Psyclw/ogy, 26. Schlesinger, A. M. Jr (1965) A 17wu.mnd Days: Jo/m F. Kennedyin tlw White House. London:
')!; .....-.._:;~otter, J., P. Stringer und M. Wetherell {1984) Social Texts and Comext: Literawre and Deutsch. 1

~)• Social Psychology. London: Routledge nnd Kegan Paul. Schuman, H., G. Knlton and J. Ludwig (1983) 'Context and Contiguitylin Survey Qucstion-
Quine, \V, V. O. nnd Ullian (1970) 111e Web of Belief. New York: Rundom House. naires', Public Opinion Quarterly, 47: 112-5.
Ravcl1., J, R. {1971) Sciemijic Knowledge and its Social Prablems. Harmondsworth: Pcnguin. Sdmtz, A. ( 1972) 77w Pht:IW/1/I!IUJiogy of tlw SocicJ/ World, London: Hcincmunn.
Reeves, \V. (1983) Brltish Roela! Discourse: A Study of Brirish Political Discourse abour Scott, M. B. and S. Lymun (1968) 'Accounts', American SociologicaliReview, 33:46-62.
Race and Race~Related Mauers. Cambridge: Cambridge Uníversíty Press. Seurle, J. R., F. K..iefcr and M. Bicn.visch (eds) (1979) Smdies in Semamics and Pragmarics.
Rcicher, S. and J. Pottcr (1985) 'Psychologicul Theory as Intergroup Pcrspective: A Dordrccht: Reidel. 1

Compnrative Anulysis of "Scientific" and "Lay" Accounts ofCrowd Events', Human Semin, G. R. nnd A. S. R. Manstead (1983) Tite Accoumabiliry oj¡l Conduct: A Social
Relations, 38: 167-89. Psychological Analysis. London: Academic Press.
Rogcrs, C. (1961) On Becoming a Person. Bastan: Houghton Mifflin. Shnpiro, M. J. (1981) lnnguage and Polirical Understanding. Ncw Ha¡en: Yale University
Rony, A. O. (1976) 'A Literary Postscript: Characters, Pcrsons, Selves, lndividuals', in
A. O. Rony (ed.), 17re ldentities of Persons. Berkeley: University of California Press. Sha:;r~~·M. (1983) 'Liternry Production as a Politicizing Prnctic~'. Paper prescnted
Rony, R. (1978) 'Philosophy as a Kind of Writing', New Literary History•, 10: 141-60. nt the American Political Science Association Conference, PaJmbr House, Chicngo,
Rosch, E., C. B. Mervis, W. D. Gray, D. M. Johnson and P. Boyes-Braem (1976) 'Busic September. 1

Objects and Natural Cutegories', Cognitive Psychology, 8: 382-4-39. Shupiro, M. J. and M. G. Banham (1983) 'The Manipulation of cdncepts in Reporting
Rosenthal, R. and R. L. Ro~now {eds) (1969)AmJacr in Behavraural Researc/1. Ncw York: lntcmational Conflicts'. Paper presented ut the lntemational Socíety ofPolitical Psychology
Acndemic Prcss. Conference, Oxford University, July.
Rosier M. {1974) 'Asking Silly Qucstions', in N. Armistcud {ed.), Reconstmcring Social Shnrrock, \V. and R. Watson (1984) 'What's thc Point of ''Rescuing Motives'' 7' British
Psychology, Harmondsworth: Penguin. Joumal of Sociology, 35: 435-51.
Rothban, M. {1981) 'Memory Processes und Social Beliefs', in D. L. Hamilton (ed.), Shcars, R. und l. Gidley (1982) Storm 0111 of Africa! Auckland, New Zcaland: Macmillan.
Cognirive Processes in Srereoryping and Jmergroup Behaviour. Hillsdale, N. J.: Erlbaum. Shen.vin, C. \V. and R. S, Isenson, {1967) 'Project Hindsight', Scie(Jce, 156: 1571-7.
Ryle, G. (1949) The Concepr of Mind. London: Hutchinson. Shottcr, J. (1981) 'Telling and Reporting: Prospectivc and Retrospectivc Uses of Sclf-
Sabini, J. and M. Sil ver ( 1981) 'lntrospection and Causal Accounts', Joumal of Personality Ascription', in C. Antaki {ed.), 71ze Psycholog)' of Ordinal]• Er~Janations of Social
and Social Psyclwlogy, 40: 171-9. Behaviour. London: Academic Prcss. 1

Sabini, J. and M. SU ver {1982) Moralities of Everyday Lije. Oxford: Oxford University Press. Shotter, J. (1984) Social Accmmwbiliry and Selj1wod. Oxford: Blackwell.
Sacks, H. (1972) 'An lnitiallnvestigation ofthe Usability ofConversational Data for Doing Shottcr, J. (1985) 'Social Accountability and Self~Spccitication', iry K. J .. Gergen and
Sociology', in D. Sudnow (cd.), Studies itl Soda/ lntL'mctirm. Ncw Ynrk: Free Pm~s. K. E. Duvis (cds), 77w Social Ccms1rur1irm oftlw Pa.\'tJ/1. Ncw York: Spnnger-V!!rlag.

1
i
202 Discoursel and social psychology References 203
1

Sigall, H. and N. oJtrove (1975) 'Bcautiful but Dangerous: Effects of Offendcr Physical Tcdeschi, J. T. and M. Riess (1981) 'Verbal Strntegics in Imprcssion Management', in C. Antak.i
Attractiveness and rhc Nnture of thc Crime on Juridic Judgment', Journal of Per.wnality (ed.), 7f¡e PsychologyofOrdinary Explanations oJSocial Behavtour. London: Acndemic
and Social PsycJJoiogy, 31: 410-14. Press.
Sinclair, J. McH. arid M. Coulthnrd (1975) Towards an Analysis of Discourse, London: Tet\ock, P. E. (1984} 'Cognitive Sryle nnd Po\iticnl Belief Systems in the B~itish House of
Oxford Universi6r Press. Commons', Joumal ofPersonality and Social Psycl!ology,'46: 365-75.
Slonmn, A. (1976) ·¡what are the AÍm~ of Science?', Radical ~hi!osopi,'Y, 1~: 7-17 .. Tet]ock, P. E. andA. S. R. Mnnstead (1985) 'Impression Mnnngement Versus Intrn~Psychic
Smith, D. (1974) 'T~e Social Construcuon ofDocumentary Renilty, Socwlogtcal/nqwry, Ex.planations in Social Psychology: A Uscfu1 Dichotomy?', Psychological Review, 92:
44: 257-68. 1 59-77.
Smith, D. (1978) 'K is Mentally Ill: The Anatomy of a Factual Account', Sociology, 12: Thompson, J. B. (1984) Studies in tlw T11eory of /deology. Cambridge: Po\iry.
23-53. 1 Totman, R. (1980) 'The lncompletencss ofEthogen!cs', Europeanloumal ofSocial Psychology,
Smith, D. (1982) •fhe Active Test'. Pnper prescnted at the Tcnth World Congress in 10: 17--41.
Sociology, Mcxibo City. Trew, T. (1979) 'Theorynndldcology ntWork', in R. Fowler, B. Hodge, G, KressandT. Trew
Smith, D. (1984) 'Tfxtua\ly Mcdiated Social Organizntion', lntemational Social Sciencc (cds), Language and Control, London: Routledge nnd Kegan Paul.
Joumal, 36: 59-¡75. Trilling, L. (1974) Sincerity and Authcntidry. London: Ox.ford University Press.
Smith, J. (1981) 'Sclf and Experiencc in Mnori Culture', in P. Heelas and A. Lock (eds), Tumer, C. F. andE. Krnuss (1978) 'Fallible Indicators oftheSubjective Slate ofthc Nation',
Indigenous Psyc)JOlogies. London: Acndemic Press. American Psychologist, 33:456-70.
Snyder, M. (1981) ·dn the Self·Perpctuating Nnture of Social Stereotypcs', in D. L. Hnmilton Tumcr, J. C, ( 1981) 'Sorne Considcrntions in Gcncrnlizing Experimentnl Social Psychology',
(ed.), Cognitive frocesses in Stereotyping and lntergroup Behaviour. Hillsdnle, N. J: in G. M. Stephenson andJ. H. Da vis (eds), Progress inApplied Social Psycho/ogyVo!. l.
Erlbnum. 1 London: Wilcy.
Snyder, C. R .• R. U. Higgins nnd R. J. Stuch.')' (1983) Excuses: Masquerades in Search Ungnr, S. (1981) 'The Effects of Slatus nnd Excuse on Interpersonnl Reactions to Deviant
ofGrace. New York: Wlley. Behnviour' ,-Social Psyclwlogy Quarterly, 260-3.
Stringer, P. (ed.) (1982) Confronting Social/ssues Vals 1-2. London: Acndemic Press. Van der Vlist, R. {l982) 'Social Psychological Theory and Empiricnl Studies of Prnctical
Stringcr, P. (1985) ·~:Vou Decide Whnt Your Title is To Be nnd [Rend] Write to thnt TitJe', Problems', in P. Stringer (ed.), Confronting Sociallssues: Applicarions ofSocial Psychology
in D. Bnnnistcr ed.), /ssues and Approaches in Personal Construct T11eory. London: Vol. l. London: Acndemic Press.
Acndemic Press. Wann, T, W. ( 1964) Bchm•iourism and Phenomcnology. Chicngo: Univer.;ity ofChicngo Press.
Stubbs, M. (1983) biscourse Analysis. Oxford: Blnckwell. Watson, R. (1978) 'Catcgorizntion, Authorizntion nnd B\ame~Negotintion in Conversation',
Sturrock, J. (1986) 'srmcturalism. London: Paladín. Sociology 12: 105-13.
Suleiman, S. R. nndll. Crosmnn (eds) (1980} T11e Reader in rhe Text: Essays on Audience Watson, R. (1983) 'The Presenlntion ofVictim nnd Motive in Discourse: The Case ofPolice
and lnterpretation, Princeton: Princeton University Press. Interrogations and Interviews', Victimo/ogy, 8: 31-52.
Swartz, S. and L. swb (1982) 'Hnrré's Social Being and the Connection with Structurnlism', Wntson, R. and T. Weinberg (1982) 'Interviews and the lnternctionul Construction of Accounts
Joumal for the heary of Social Behaviour, 12: 201-12. ofHomosexunl Jdentity'. Social Analysis, ll: 56-78.
Sykes, G. M. ando.¡ Matzn (1957) 'Techniques of Neutrnlizntion: A Theory ofDelinquency', Webb, E. J., D. T. Campbel\, R. D. Schwnrtz nnd L. Sechrest(1966) Unobtrusive Measures:
American Sociol'ogical Review, 22: 664-70. N011·Reacrive Rescarch in the Social Sciences. Chicngo: Rand McNnlly.
Sykes, M. (1985) ·biscrimination in Discourse', in T. A. van Dijk (cd.), Handbook of Wcblcy, l. A. (1986) 'Professional Wrestling: The Wor\d of Roland Barthes Revisited',
Di:;course Analy~~is Vol. 4. London: Acndcmic Press. Semiorica, 58: 59-8 J.
Tnjfe\, H. (1969) •Cognitive Aspects of Prcjudice', Joumal of Bio:;ocial Sciences, !, Weiss, C. H. nnd M. J. Bucuvnlns ( 1980) Social Science Research and Decision Making. Ncw
Supplement Mo~. No. 1, Biosocial Aspects of Roce, 173-91. York: Columbia University Press.
Tajfel, H. (1972) 'E{periments in a Vacuum' '.in J. A. 1srncl und H. Tujfcl (eds), 1he Comext Wetherell, M. S. (1983) 'Socio·Psycho\ogicnl nnd Litcrnry _Accounts of Femininity', in P.
of Social P:;ychology. London: AcademJc Press. Stringer (ed.), Confronting Social/ssues Vol. 2. London: Acndemic Press.
Tajfel, H. ( 1981) ·sfcial Stereotypes nnd Social Groups', in J. Tumer nnd H. Giles (eds), Wetherell, M. S. (1986) 'Linguistic Rcpertoires and Liternry Criticism: New Directions for
lntergroup Behaviour. Oxford: Blackwell, aSocial Psychology ofGender', in S. Wilkinson (ed.), Fcminist Social Psychology. Mil ton
Tajfel, H. (1982) H1Jman Groups and Social Categories. Cambridge: Cambridge University ,. ·t _ Keynes: Open University Press.
Prcss. ! r.;.,~ -~11. M. S.,.R. McFadyen, J. Potterand.S .. Rothwell (1986)
~:lí)Yemece __ 'Categorie-siriDisC(iUf'Se'.,--.....
1
Tnjfel, H. nnd J. Fqrgas ( 1981) 'Social Cntegorization: Cognitions, Vnlues nnd Groups', -. mimeo, University of St. .Andrews~----. __ . ~---
in J. Forgas (e&), Social Cognition. London: Acndemic Press. ~he~~[
tvr:-s:·ñnd fPOtt~r (1986) 'Maj-~~it).!Group Representntons of "RñCeReTiitions''
Tajfel, H. nnd A. wftkcs (1963) 'Classificati'on and Quantitntive Judgement', Brir(sh Jo.~[IJO.{ in New Zealnnd'. Paper presented at the British Psychologicnl Society, Social Psychology
of Psychology, 54: 101-14. Section Annual Conference, University of Sussex., September.
Tannen, D. (1984) Caherence in Spoken and Written Discourse. Norwood, N. J: Ablex. '~~,: ~Wethcrell, M. S. nnd J. Pottcr, (forthcoming a) Mapping the Úlnguage of Racism.
Tnylor, S. E. (!98d 'A Cntegorization Appronch to Stercotyping', in D. L. Hamilton (ed.), Wethcrell, M. S. and J. Potter (forthcoming b) 'Narrative Characters and Accounting for
Cognitive Proce.~.ws in Stereoryping and lntergroup Behaviour. Hillsdule, N. J.: Erlbaum. Violence', in J. Shotter and K. E. Gergen (eds), Texts ofldentity. Greenwich, Conn.: JAI
Press.
204 Discourse and social psyclwlogy

Wetherell, M.s., H. Stiven and J. Potter (1987) 'Unequal Egalitnrianism: A Preliminnry


Study of Discourses Conceming Gender und Employment Opportunities', British Jo umal
of Social Psychology, 26: 59-71. Index
Wicker, A. W. (1969) · Anitudes Versus Actions: The Relntionship ofOven and Behnvioural
Responses te Attirude Objects', Jaumal of Social Jsmes, 25:41-78. . .
Wicker, A. W. (1979) An Introduction to Ecological Psyclwlogy. Monterey, Callfomm: Ability, 86 , Anscombe, E., 180
Brooks/Cole. Abruham, C., 41, 173 Answers, S2-5, 87, 89
Wierler, L. (1974a) 'TeUing the Code', in R. Tumer (ed.), Ethnomethadology. Hannondsworth: Absence, 2S, 31 Anthropology, 6, 57 1

Penguin. Acct!nt, 38 Anti·Apnrtheid, 111 l


Wieder, L. (1974b) úmguage and Social Reality. The Hague: Mouton. Acceplances, 82-3, 86 Anything goes, 159
Wilder, D. ( 1978) 'Perceiving Persons as a Group: Effects on Attriblllions of Causality und Accountability, 106 Apologies, 76-7
1

Beliefs', Social Psychology, 1: 13-23. Accounts, 1, 74-81, 84-94, 122, 177-8 Appeal to 1

Wilder, D. A. (19S6) 'Social Categorization: Implications for Creation and Reduction of (see also discourse) accident, 76 1

Intergroup Bias', ín L. Bcrkowitz (ed.), Advances in E-cperimental Social Psychology of nccounts, 4, 182--4 fi1cework, 78 1

Vol. 19; New York: Academic Press. as uccuratc/inaccurate descriptions of higher nuthority, 78
Williamson, J. (l97S) Decoding Advertisements. London: Boyars. mental states, 178-80 loyallies, 76 1

Wittgenstein, L. (1953) Philosophicallnvestigations. Oxford: Blackwell. conventionnl charucter, 85 mental elements, 76


Wittgenstein, L. {19SO) Remarks on the Philosophy of Psychology Vals 1-2. Oxford: effcctiveness, 77-9 mitigating circumsinnces, 78
BlackwelJ. f:~ctua1, 1SI-2, IS3 natural drivcs, 761
Woolgar, S. {19S0) 'Discovery: Logic and Sequence in a Scientific Text', in K. Knorr· of self, 111-5 rctribution, 78
Cetina, R. Krohn and R. Whitley (eds), 11w Social Process ofScientific lnvestigati!Jiz. Accounting utilitarianism, 78 1
Dordrccht: Reidel. device/strnlegy, 107, 113-4 vnlues, 78
Woolgar, S. (19S2) 'lrony in thc Social Studics of Scicnce', in K. D. l<norr·Celina nnd for error, 149-53, 155, 161, 169 Applause, 47, 87, 175.

M. Mulkay (eds), Science Observed: Perspecrives on the Social Swdy of Science. London for freedom and constraint, 75, 91-2 App1ication 1

and Beverly Hills, California: Sage. Accusations, 87-93, 130, 170 of disc-ourse anniySis, 174-5
Woolgar, S. {ed.) (1987) Tuming tlze Pages ofScience: Knowledge and Rejlexiviry. London Achievement motivation, 41 of knowledge, 1861
and Beverly Hills, California: Sage. Acknowlcdgcmcnts, 77 Arbitrnriness of the sign, 25
Woolton, A. (1977) Dilemmas of Discourse. London: Allcn Unwin. Actions, 57-S, 61-2 Architecture of conversmion, SI
Wowk, M. T. (1984) 'Blame Allocation, Sex and Gendcr in a Murder Interrogation', Women 's cxpressive versus prncticnl, 60 Argyris, C., 40 1

SIUdies lmemationol Forum, 7: 75-82. versus movements, 57 Ashmore, M., 30, 168, 183, 184
Wynne, A. {1987) •Accounting for Accounts of Multiple Sclcrosis', in S. Woolgar (~d.) semiological character, 58, 63, !03-4 Asscssments, S2
Tuming tlze Poges of Science: Knowledge and Rejlexiviry. London and Beverly H11ls, unitary meaning, 63 Atkinson, M., 30, 47, 83, 87-92, 153,
California: Sagc. . Adams, J., 186 162, 166, 170, 175, 187
Yearley, S. (l9Sl) 'Textual Pcrsuasion: The Role ofSocial Accounting in thc Construct10n Adcnosine triphosphate, 146-7 Atkinson, P., 30, 40
of Scientific Arguments', Philosoplzy of tlze Social Sciences, 11: 409-35. Adjacency pairs, 81-6, 88, 92-4, 13S Altitudes, 1, 32, 25, 47, 43-55, 122,
Yearley, S. (1984a) Science and Sociological Practice. Milton Keynes: Open University Press. central organization of conversation, S3 142-6, 157, 160, 178, 179, 184
Yenrlcy, S. (1984b) 'Proofs and Reputntions: Sir James Hall and the Use of Classification typcd, 82 definitions, 43-4 1

Dcvices in Scientific Argument', Earth Sciences History, 3: 25--43. Acsthetics, 184 prcdicting behavio~r, 53-4
Ycarley, S. (1985) 'Vocabularies of Freedom and Resentment: A Strawsonian Perspective African National Congress, 6 traditional 11pproachcs, 43-6
on the Naturc of Argumentation in Scicncc and the Law', Social Studies of Science, Aggression, 36, 59-62, 112, 113, 136 as undcrlying entit es, 46, 49-50,
15: 99-126. Agncs, 126 53-4, 164
Zimmennan, D. H. (197l) 'Thc Prncticalities of Rule Use', in J. Douglas (ed.), Understorzding Agreements, 84 Attributions, 91. \40, 142, 146
Everyday Lije. London: Routledge and Kegan Paul. A\len, V. L., 118 Austin, J., 14-IS, 74-6, SI
Zimmermnn, D. W1d M. Pollner (1971) 'The Everydny World as a Phenomenon', in J. Doug1as Allport, G., 43 Authcnticity, 97
(cd.), Understanding Everyday Lije. London: Routlcdge and Kegan Pnul. Althusscr, L., 109 Azjen, L. 53-4, 178
Zuckennan, H. A. (1977) 'Deviant Behaviour and Social Control in Science', in E. Sagarin Anchoring, 123, 132, 141-2, 145-6 (see
(ed.), Devionce and Social Clzange. London and Beverly Hil\s, California: Sage. a/so social representations) Bainbridgc, L., 178
Anger, ISO Balzac, H. de 1S7
Animul behaviour, 63 Banham, M. G., 187
Anonymity, 79 Bnr-Hillel, Y., 23
i
206 Discourse and social psycho/ogy Index 207
1
Barjonct, P. E., 1~9 inductive errors, 129 Commonsense assumptions, 30, 42 Caurse in General Línguistics, 24
Bnrncs, B., 23, 79, !59 occasioned use, 133 Communication, 9 Courtroom discourse, 87-93, 156, 170,
Bnrthes, R., 26-8\ 31, 187 over-inc\usion, 132 Community, 132-7, 138, 167 162
Bnumeister, R. F.!, 37 prefonned entitics, 120-1, 136 nttributes, 134-6 Cownrd, R., 104, 109-10
Bny of Pigs, 1831 process of assignment, 119 policing, 135-6 Craft skills, 148
Buzerman, C., 64, 183 puzzle, 131-2 relations, 134-5 van Crannch, M., 138
Benttie, G., 81 Categorization, 116-37, 138, 143 repertoire, 138, 149, 160 Crnne, D., 65
Becker, H., 19 nnalytic, 29, 33, 41, 44-6 variable npplication, 134 Crisis in social psychology, 2-3, 40
Bchnviourism, 12 14, 95 consequences of, 116 Competence, Criterin for understnnding, 179-80
Belicfs, 17, 179-~0 nexibility' 116 linguistic, 10-12, 56-8 Criticism, 68
Berelson, B., 167¡ mechanical process, 118-21, 137 social, 56-60, 73, 177 Crosmnn, I,, 187
Berger, P. L., 142 perceptual effects, 116-7, 119-21 , Concepts, 25 Cross-cultural compnrison, 104-6
Best, E., 104 1 133, 136-7 see also signifieds Cuff, E. C., 126, 182
Billig, M., 53, 1:p-2, 137, 144, 167 of physica1 world, 117 Conditiona1 rclevance, 82-3 Culler, J., 26, JO, 164, 167
Biochemistry, 146-56, 160, 169-70, 175 suppressing varinbility, 35, 39, 40-2, Confirmation through exception, 69-71, Cultural
Bio1ogy, 181 1 123-4, 144, 164 170 conventions, 27
B1nmings, 37,48 54, 79-80, 89-91, simplifying, 117 Conflict, universnls, 104
116, 126-pO, 132, 136-7 ns social accomplishment, 53, 77, linguistic, 6, Culture, 9, 25, 27
no blame nccdunts, 86 116, 120-1, 126, 156 political, 6
Bogdnn, R., 42, ~3 of social world, 118 Connotntion, 5 Dnhrendorf, R., 98-9
Bourhis, R. Y., 38 Cattell, R., 96 Conservative Party, 139-42 Darby, B. W., 77, 79, 65
Bowcr, G. H., 71 Causal Consistency/inconsistency, 34, 37-8, 54, Dnvidson, J., 85
Boxing, 27-8 constrnints on behaviour, 40 122, 126, 156, 163-5, 166-70 Da vis, K. E., 92, 95
Brnnnigun, A., 64 lnws, 56, !59 as indicator of validity, 34, 163 De Cario, N., 56
Brur.il, D., 166 1 Causes nf action, 75 as variable statcs, 38 Delny, 80, 82, 84-6, 93, 166
Brenkdowns in normal conduct, 74 Ccn.:mony, 29 of behnviour, 40 infcrentinlly rich, 85
Brenner, M., 34,1163 Chullenges, see nccusutions Constative utternnces, 15-17 Delivcry, 84-6
Brewer, M., 119r 125 Chnlmcrs, A., 79 Constitution, 50-2 Denial of
Brondbem, G., 26 Churncter types, 96-7, 102, 111-5 Contcnt unnlysis, 41, 167, 173, 175 injury, 76
Brown, D., 33, &o Chemiosmosis, 146-7 Contcxt, 10, 14, 18, 23, 29, 33-5, 37-8, intent, 78
Brown, G., 6 1 Chomsl.-y, N., 9-14, 26, 57, 58 41,45-50,54, 56, 72-J, 79-61, victim, 76
Brunswik, E., 7l Chua, B. H., 166 126, 133, 156, 164, 169 volition, 78, 91
Bucuvnlus, M. J ·¡· 174 Cicourcl, A. V., 40 of discourse, 147-9, 157 Denzin, N. K., 63
Buildings, 26 1 Cirilo, R. R., 7 formal versus informal, 147-9 Derrida, J., 31
Burenucratic model of thought, 121-2 Clnrke, D., 56, 138 Contingent repertoire, 149-55, 160, 169 Descriptions, 2-4, 6, 15-17, 21, 23, 28,
Button, G., 13 1' Clnssrooms, 87 bnsic principie, 149-50 JO, 33-4, 36, 44-5, 46, 51-2,
C\osings, 138, 171 Contrast struetures, 47 56, 61, 64-5, 69, 72, 90, 92,
Cannell, C. F., 163 Code, 9 Conversation, 20, 23, 29, 38, 59, 116, 116, 123-4, 126-7, 1Jo, m;
Cantor, N., 119-20, 122, 132
1
Coding, 41, 66, 161, 167, 173-4 166, 174 140, 143, 156, 160
Casey, N., 13 1
cydical process, 167 Conversation analysis, 7, 30, 74, 80-94, Devinnce, 20
Cntegories, 1, 2, 15, 6, 19, 20, 52, Cognition, 7, 9, 10, 146 102, 103, 136, 160, 161, 170-1, van Dijk, T., 6-7, 121
1
116-37, 141, 143, 179 (see also Cognitive 184, 167 Dialect, 38
membersiÜp) consistency, 37 The Convict Code, 19-23, 30, 72-3 Dickens, C., 96.,..7
base for infeiences, 126, 129-32 dissonunce, 36-8 ns persunsive device, 22, 72 Dimensions of judgement, 43, 46, 52
bound uctivit!cs, 129, 132, 136 psycho\ogy, 119, 140, 145 open flexible texture, 22 Disagreements, 84
duplicativc o}gnnization, 129, 131 processes, 116,120, 137,.1~7¡,, as exp1nnation, 19-20, 22 Disclnimers, 48, 54, 76-8, 180
cxemplify wÓrld views, 127-8 178-60, 185-6 Coo1ey, C. H., 96 Discourse, 6-7, 46, 116
fhed slructuie, 120, 136 reductionism, 56, 121, 142, 157 Coulter, J., 178, 179, 180 nction oríentntion, 28-9, 48, 54, 93,
hiernrchical Órganization, 119-20, Collett, P., 58 Coulthard, M., 6, 29 126, 134, 160
125, 136! Collins, H., 79, 159, 173, 175, 181 Counsel, 87-92 as social behaviour, 14
'
208 Discourse and social psychology 1/ndex 209

consequences of, 18, 21, 28, 34-6, Drew, P., 30, 33, 83, 86, 87-92, 130, Fcstinger, L., 37, 98 Barré, R., 2, 29, 40, 42156-61,63, 73,
43, 46, 55, 64, 91, 137, 146, !33, 170 Feyernbend, P., 158, 159 79, 95, 102, 103, 104-5, 106-7,
164, 168, 170 Duncan, B. L., 36 Fishbein, M., 53-4, 178 138, t59, 173, m, 179, 180
construclive, 2-3, 6, 23, 32-6, 41, Fiskc, J., 26, 184 Harris, R., 11
43, 51, 52, 55, 64, 67, 72, 128, Ecological Football fans, see soccer fans Hartlcy, J., 26, 184
146, 152, 157, 160-1, 162, 164, psychology, 63 Porgas, J., 117,138 Heath, S., 109
169, 175, 178, 180, 181, 182--4, validity, 79 Foucauh, M., 6-7,31, !09 'Hebdidge, 0., 26
(see also constitution) Edelmrm, M., 187 Fractures in social arder, 75 Heelas, P., 103
ns dependent variable, 173 Elgin, P., 161, 186 Frnnkfurt School, 109 Heider, F., 92
as function, 14-8, 23, 28, 32-6, 38, Ellis, J., 109 Frederickson, C. H., 7 Henriques, J., 95, 101, )03, 104, 109
41, 45, 49, 54, 57, 61, 67, 72, Elms, A. C., 2 Freedom fighters, 5 Heritage, J., 18, 28, 47,j73, 81, 82, 86,
74, 79, 93, 110, 114-5, 122, 126, Empiricist repertoire, 149-51, 153-5, Free will, 75 87, 153, 162, 166, 187
136, 145, 157, 160-1, 164, 169 Functions of discourse, see discourse Herzlich, C., 139, 144 ¡
168-70, 174, 176, 178, 180, 185 basic principie, 149-50 Fuzzy sets, 119-20, 136 Hesse, M., 158 1

specific versus global, 33 Envy, 180 Hewitt, J. P., 48, 76-71


conscious versus unconscious, 34, Etcetera clause, 22-3, 72 Hcwstone, M., 142, 144
Garfinkel, H., 18-19, 22-3, 75, 126,
153 Ethnography, 30, 40
156, 168, 179 H~~hMencrgy physics, 751
direct versus indirect, 33 Ethnomethodology, 1, 18-23, 28-30, 32, HlJnCk, 5 ,
intentional versus unintentional, 34 41, 75, 81, 102, 116, 126, 128-9, Gnrvey, C., 81 1
Hinde, R., 63
literal, 61-2, 64, 67, 160 (see also 143, 153, 168, 173, 179-80 Gendcr, 126, 128, 130, 156
History of science, 70-l., 158-9 1

truth) Ethogenics, 42, 56-64-, 173, 177 Genernliznbility of findings, 40


Hodge, R., 184, 186 1

and mentnl states, 177-80 central hypothesis, 57 Generative


Hollway, W., 109 1

organization, 46-7, 54, 64, 67, 69, Ethology, 63 grammar, 10, 13, 14
Ho1sti, O. R., 41, 167,1173
71-2, 80, 137, 164, 174, 178, Eva1uation, 3, 5, 30, 51, 54, 1\6, 134, mechanisms, 56
Gergen, K., 2, 40, 63, 95, 101-3, 106, Honour, 60 1

179, 185, 186 138-9, 149, 175 Hotrodders, 127-8


pathway to actions, belicfs, evcnL'i, Evolutionary theory, 70 108-9, 178
How ToDo TIJings With IVorcü, 16
34-5, 41, 49, 57-62, 160, E:r.:aminutiun, 87-92 Di Giucomo, J. P., 139, 144
l·lunmnist, 95, IOH
Gidley, l., 111
1

179-80 E:r.:cuses, 74-80, 85, 91-2, 113 thcories of sclf, 99-100, 101, 102,
pcrsuasive, 33, 41, 47, 61, 92, 122, 187 Experimenter effcct, 185 Gilbert, N., 6-7, 35, 64, 65, 68, 136,
107
138, 146-55, 156, 160, 162, 169,
1

primary psychologícal rcality, 81 ExperimenL'i, 2, 36-40, 56, 65. 74, 77,

~~;~~~~~:~~~~:;2
and the 'real' wor!d, 177, 180-2, 184 79-80, 85-6, 93, 116-8, 158, 170-1, 172
as tapie, 34-5 (see a/so 159, 168, 172, 275-6, 185-6 Giles, H., 7, 38
topic/resource distinction) field, 77, 79-80 Goffman, E., 19, 76, 82, 98 111114
Discourse analysis, JI, 14, 23, 30, 32-8, gencrulizability of findings, 40, Gombrowicz, W., 98
Hypothcsis testing, 41
42-5,93,53-7, 61, 64-5, 67, naturulism, 79-80 Orad y, K., 187
69,71-3, 116-7, 122, 124, 126, as social uccomplishments, 175 Grammar, 11, 13, 14, 24, 57
lckes, W., 178
136, 138, 145, 146, 157, 178-87 Extraversion, 96-7 Grentbatch, D., 47, 187
ldeuiization, 10, 11, 13~ 18, 29,
application of, 186-7 Extreme case formulations, 47-8, 91 Grectings, 57-8, 80, 82
decontextunlization 10, 18, 28
definitions, 6-7 Eysenck, H. J., 96-7 Grice, M. P., 30
regularization, lO 1

majar components, 32-5 Eysenck, S. G., 96-7 Gross categorization, see categorization
stundardization, 10
Group dynnmics, 183
ll

principal tenet, 33 ldentity crisis, 97


process of, 46,65-7, 111, 123, 133, Facts, 18, 52, 63, 72, 92, 143, 154, 161 Gusfie1d, J., 161, 183 1

ldcology, 118, 140, 163, 187


146-7, 153, 155, 168-9 as construction, 92 Immigrants, 45-54 1

self-referential, 183-4 Falsifiability, see tcstability Halliduy, M. A. K., 7 coloured, 43-5 ¡'
and social psychology, 185-6 Fanshe\1, D., 29 Halliday, Q., 133 Polyncsian, 46, 49-p3
Discrediting, 114-5 Farr, R., 139 Hnmilton, O., 118, 121 lmprcssion manngcmenl, see self-
Dispreferred seconds, 84-6, 93 Fnshion, 26 Hamlyn, D. W., 99 . prescntation 1

organization of, 84-6 Fenture matching, 119, 133 Hammersley, M., 30, 40 lndexicality, 23, 41, 1Df. 145, 168
Documems, 11, 40, 42,64-5, 133, Felicity conditions, 16-7, 29 Hansard, 163 Indígenous psychologies, 103, 104-6
147-8, 161-3, 166, 168, 172--4 Femininity, 109-10 Haruri, J. V., 187 lnfinite rcgress, 182 1

1
210 Discour1e and social psychology lndex 211
!

lnsertion sequencJs, 82 Kinder, D. R., 40 Lymnn, S., 74, 76, 130 Miller, G. A., 175
lnsu!ts, 58 / Kintch, W., 6-7 Lytt1eton, H., 186 Milis, C. W., 76, 130, 180
Integrntive complexity, 173 Kiss ass, 19 Milroy, L., 7

:~::~~~~~'~ 180 Knowledge application, !55 McCabe, C., 101 Mind, 107-8, 113, 178
1
Krampen, M., 26 McClelland, D. C., 41 Mischel, W., 97, 119-20, 122
distinctivcness~ 38 Krnuss, E., 40 McConnhay, J ..B., 53 Mitigations, 128, 130, 132
psychology, 71 Kreckel, M., 166 MacDonell, D., 6 Mixon, D., 39
lmennission, 1661 Kress, G., 186 McFadyen, R., 123 Molecular theory of gases, 70
lnterpretntive 1 Kroger, R. O., 56 McGuire, W. J., 2, 43 Moscovicí, S., 81, 138-146, 157
procedures, 128, 146, 153-5, 173 Kuhn, T. S., 65, 158, 159 McHoul, A. W., 187 Mostyn, B., 41, 167
genernte-pfoblems, 153, 155 Kulka, R. A., 92 McKin1ay, A., 68, 138, 139, 145, 152, Motives, 5, 19, 21-2, 37, 48, 71,
sociology, 173 !55 113-4, 126, 130-1, 177, 180
repertoires, t:b-9, 146-57, 164, Lnboratory research, 79 Manstead, A., 37, 74, 76-8,90-1 Mulkay, M., 6-7, 22, JO, 35, 42, 64,
!69-71 1 Labov, W., 7, 29, 163 Maoris, 58, 104-6, 123-6, 128 65, 68, 87, !36, 138, 146-55,
definition,ll49 Lakatos, l., !59 cosmology, 105 156, !59, !60, 162, 163, 164,
uses nnd functions, 149 Lnkoff, G .. 107 Marsh, A., 43-7, 53 !69, 170-1, 172, 175, !83, !84
reseurch, 31 1
Lalljee, M., 55 Marsh, P., 40, 56,59-62, 64, 65, 66-7, Multiple partial indicators, see
Inter-rnter relinbility, 173 Language, (sce also discourse) 72 triangulation
lnterviews, 1, 40146, 49, 59, 123, 133, abstrnct system, 14, 18, 28 Mnrx.ist analyses, 109 Multiple sclerosis, 184
145, 147-8, 153-5, 162-5 acquisition, 12, 81 Masculinity, 109 Murder confession, 130-2
confrontntion~l arena, 164 device, 12 Mas1ow, A. H., 100 Myers, G., 64
comparnbility¡or responses, 163 historien! origins, 81 Material rcality, 180-2 Myth, 26-7, 31
craft skill, 165 medium for action, 9, 173 Mntson, F., 95
schedule, 165j as nnming process, 25 Matza, D., 76 Narrative, 3, 5, 52, 60-1,71, 135-6
Intonntion, 10, 3~-9, 166 social nature of, 18, 28 Mead, G. H., 81, 98 National Health Service, 174-5
Jntroversion, 96-¡7 subject!object property, 107 Meals, 24 Natural sciencc, 56, 70, 147, 174
lnvitations, 83, 85-6, 88, 122 La langue, see underlying system Meaning, 25-7, 31 idealized view of, 159
1

Ironizntion, 42, 181 Latour, B., 159 of actions, 27-8, 57-8, 63 social scicncc as copy of, 56
Isenson, l. S., ti4 Law, J., 23, 159 contex.tunl, see indcxicality Naturalistic research, 79, 162, 176
Israel, J., 2 1
Lay ex.plnnations, 2, 20-1, 34, 51-2, 79, equivalence, 45 New literary forms, 183-4
140, 144-5 versus force, 17 Newspapers, 5, 59-60, 62, 64, 116, 133,
Jackson, S., 291 Lnyton, E., 174 of utterances, 23 !63, 174, !87
Jncobs, S., 29 1 Lee, J., 130 as accomplishment, 23 Newton, 1., 11
James, W., 1081 Legal discourse, 156 taken-for-grantcd, 27 Ncw Zealand/New Zealanders, 46, 48-52,
Janis, I., 183 1
Levinson, S. C., 7, 28, 29, 30, 33, 80, Mehan, M., 87 58, 104, 111, 1!3, 114, 123, 164
Jnyyusi, L., 129 1 130, 133 81, 83, 85, 86, 153, !55, 162, Membcrship Nisbett, R. E., 177-8
Jefferson, G., 8q, 166 171, 172 cntegorics, 44, 116, 122, 128, 130 The Normal, 7
Johnson, M., 107 Linguistic categorization devices, 128, 131 Normative expectations, 82
Johnson-Laird, ~·· lO philosophy, 1, 14, 74-5, 179-80 consistcncy rule, 128, 130-1 Northern Ireland, 88
Jones, E., 92 1
repertoires, see interpretntive Memory, 179-80
Jurors, 18-9 1
rcpertoires Mental Objcctilication, 141-2, 145 (see also
Justilicntion, 3, 52, 61, 65, 69-72, Linguistics, 6, 7, 138 illness, 3-5 social reprcscntations)
75-so, s.\-5, 90, II6, 108-9, Litcrary studies, 6, 8, 31, 187 stntes, 45, 177-80 Objects of thought, 43, 46, 50-4
130, ¡sol Liternture 26, 103 Mentnlistic terms, 179-80 Observation, 57-8, 62-4,
Litton, l., 42, 52, 138, 142, 144 Merritt, M., 86, 172 canvcntions for interpreting, 66
!
Kuhn, R. L., !63 Lock, A., 103, 104- Merton, R., 65 theory ladeo, !58
Kallen, D. B. P!.
174 Lodge, D., 99 Metaphors, 91, 107-8, 112, 134, 138, Occnsion, sce contex.t
Offcrs, 82, 86, 93
Kcllcy, H. H., ~2 Lodge, P., 159 145, 149
Kenncdy, Prcsidfnt, 183 Luckmann, T., 142 Micro-sociology, see sociology Opcn-cnded interviews, see interviews
Kcss\er, J. 8., Q2 Lyons, J., 10, l03 Milgram, S., 39 Opcn souls doctrine, 177-8
!
212 Discourse and social psycltology 1 lndex 213

Operationnlization of variables, 79 Play dialogue, 13 Rcbuttals, 87-92 explanatory, 21,59,172-3


Opinion polls, 40, 43-4, 48, 139-40, A Plea for Excuses, 74 Records, see documents generative, 11-2 1

158, 168, 176 Plummer, K., 40, 42, \63 Reeves, W., 45, 53, 121, 163 identification of, 56, lss
varintions in wording, 40 Polanyi, M., 175 Reference, 2 indctcrminncy of npP1ication, 22-3
contrndictory responses, 40 Police, 111-2, 113, 115 Reflexivity, 2, 21-3,28,72-3, 177, infonnnl, 19 1

limitations on responses, 40 Politcness, 32-3, 80 182-4 intcrprctntive, 58-9 1

interpretation of responses, 40 Political science, 187 as topic/strntegy, 183-4 officiul, 20 1

Opinions Pollner, M., 20, 143, 182 Refusals, 85, 87 paradigmatic, 24 ¡'

as discursive acts, 48 Pomernntz, A., 47, 68, 91 conventionnl formal, 87 prescriptive, 24


flexible, 40 Popper, K., 65, 158, 159 Refutability, .sec testability procedural, 150 1

tailored to context, 40 Positivism, 2, 3, 14, 158, 172 Reichcr, S., 114, 133, 138, 184 regulnlive, 58-60, 68, 72
Ordinnry Post·structuralism, 31, 102 Rcificntion, 42, 106, 113, 181, 187 scqucncing, 29 1

rensoning, 126, 128-32 Poner, J., 35, 42, 52, 53, 65, 68, 72, 96, Repnir, 75 strntegic use, 68-9
ta1k, 11-3, 18-9, 29, 64, 80-8, 140, 102-3,111,112,114, 133,138,139, Rcplication, 159, 172, 183 ns symbolic resourcds, 73
162 (see a/so discourse) 142, 144, 145, 146, 150, 152, 155, Requests, 32-3, 76 syntngmatic, 24 1
as joint achievement, 13 156, 161, 162, 164, 167, 168, 174, indircct, 33 as temp1ntc, 19, 22-t3, 69, 73
constitutive pnrt of action, 21 183, 184, 187 Reseurch trnnsformnlionn1, 121
as explanatory resource, 20 Power, 108-10 programmcs, !59 Thc Rules of Disonler, i56
multiconsequential, 21, 28, 30 Pragmntics, 7 qucstions, 160-1 Ry1e, G., 179 1

multifonnative, 21, 28, 30 Preference structure, 81, 83-92 rcport, 172-4


Sabini, J., 178, 179, tsp
1

conventionnl represenlation, 13 as normntive ranking, 83 Resource, see topic/rcsourcc distinction


The Orient, 187 Prejudice, see racism Response, 30 S¡¡cks, H., 13, 80-2, 126-9, 162, 171
Orne, M. T., 37 Pronouncintion, scc intonation Rcsponsibility, 77 Su;d, E., 187 !

Ostrove, N., 92 Protestors, 111, 112-5 Restriction, 35, 39-40, 42 St Clair, R., 7, 38 1
Overlnp, 82, 84, 166 Prototypes, 119-26, 128, 132-3, 136, 141 Revolution in conception of scicnce, St Pnul's riot, 133, 144-5
Psycholinguistics, 9, 11, 13, 18, 28-9 158-9 Snmplc
Pnin, 27, 58 Public speaking, 47, 160, 173, 175 Rhctoric, 38, 52, 70, 73, 91, 121, 144, se\ection, 161-2
Parliament, 133 155, 173, 182, 187 single texts, 161
Pnrker, 1., 104, 109, 139 Qualitntive research, 42 functional, 61-3, 72 size, 161
Parole, 31 Quantification, 39 Ricss, M., 76 Snmpson, E. E., 95, 10¡1, 109
Participant Qucstionnnires, 40, 47, 116, 145 Rims, 63, 88-90, 111, 133-6, 138, 162 Snussurc, F. de, 24-7, 81
observation, 19, 173 contradictory responses, 40 versus disturbnnce, 63, 135, 184 Scapegonting, 76 1

orientation, 153 Queslion wording, 40 Ritual chase, 60 Scnrman Tribunal, 88-~


Pnrticularization, 121-2 Questions, 82-4, 87-9, 165, no Rack music, 26 Scheg\off, E. A., 80, 82, 171
Pauses, .see deluy Quine, W. V. 0., 65 Rogcrs, C., 100 Schemnta, 7 1

Pecheux, M., 6 Roles/Role thcory, 21,97-9, 100, 101, Schlenker, 8., 76, 77, 79, 85
Performance, 56-7, 73 Race theories, 44 102, 107 Schlesinger, J., 183 1

linguistic, 10-14, 57 Racism, 36, 43, 48, 52, 118, 121-3, Rorty, A., 97 Schuman, H., 40 l
Performative 130, 139-40, 187 Rorty, R., 187 &hutz, A., 63 ,
utternnces, 15-7 Railwny timetnble, 25-6 Rosch, E., 119, 136 Scientists' 1

verbs, 17, 28 Rating scnle, 43-5, 48, 50, 52, 145 Rosenthnl, R., 37 confercnces, 65-71, 1163
Per1s, F., 100 one versus multi·dimensionnl, 53 Rosicr, M., 40 discourse, 3, 64-73,. 146-57, 160,
Personality inventaries, 96-7 Rationality, 113 Rosnow, R. L., 37 169-72, 175, 18~
Pcters, R., 180 Ravetz., J. R., 175 Rothbart, M., 121 bnsic principie, 150, 155
Pcttigrew, T. F., 132 Reading for gist, 168 Royal Ulster Constabulary, 88-90, 92 as hard case, 641
Philosophy, 6, 14, 18, 23, 29, 180, 181, Readings, 5 Rules, 10, 56, 62, 64, 103, 122, 179 reudings, 155 i
182 Realist brenking, 20 model talk, 155 1
of langungc, sec linguistic philosophy model of lnnguage, 34-6, 42, 72, as cognitive resources, 57-8 Jetters, 155 1

of mind, 109 134, 169, 182 as rcprcsentations, JO, 23 1

Nobcl Prizc speeches, 87, \62


of sciencc, 65, 72, 158-9, 172 principie, 101, 103 dcrivcd from tnlk. 56 rule use, 56, 64-711

1
214 Discot!rse and social psychology 1ndex 215
1

Scott, M., 74, 76, 130 Se:< roles, 98 representntions, l. 138-46, 155-7, Sturrock, J., 28
Scripts, 7 ! Sexual rclntions, 11 O 178, 179 Subject/object dichotomy, 181
Searle, J., 18 1
Shapiro, M., 187 coherence, 138, 145-6 Subjectivity, 109
Sears, D. 0., 49 . Sharrock, W., 180 code for communicntion, 138, 140-3 Submissive display, 60
Secord, P. F., 2, 40, 56, 79, 177 Shears, R., 111 cognitive versus linguistic, 145 Suleimnn, S. R., 187
Selective readin~, 35, 40, 42 Sherwin, C. W., 174 consensual, 142-5, 156 Swanz, L., 58
Sclf, 1, 5, 9, 31j, 95-115, 122, 181 Shotter, J., 81, 95,·103, 106, 107, constructive, 141-2 Swartz, S., 58
nlienUied, wp 178 distinguishing groups, 138, 141-3, Sykes, M., 45, 76
authentic, 100, 107 Signll, H., 92 146, 156
us centre of~xpertence, 101, !02, \05 Sign, 25 elcmenlS, 139-40, 145 Tnjfcl, H., 2, 117-8, 120, 129, 132, 186
in conflict, 97,
' 98, 107 Signification, 26 figurative nucleus, 139 Tannen, D., 6
as c~~~~;ny~historically relntive, second ltvel, 26-8 homogenizing force, 141 Taperecording, 163, 166
Signifieds, 25-6 organizing beliefs and nttributions, Taylor, S. E., 118
definitive de~criptions of, 95-6, 102 Signifiers, 25-6 138 Tnylor, S. J., 42, 63
and discoursb, 102-115, 178, 179, Silver, M., 178, 179, 180 originare in internction, 140 Tedeschi, J. T., 76
Sinclnir, J. McH., 6, 29 operntionnlization, 144 Teenage drivers, 127
180 1

grammntical and metaphoricnl, 106-8 Skinner, B. F., 14 problems with, 142-6 Television, 26, 174, 187
as honest soul, 96-7, 114 Sloman, A., 159 irreducibly social, 140, 146 news, 87, 133, 162
and ideology:, 108-10, 113-4- Smith, D., 161, 182, 187 structure, 126 Territory, 59
nnd intergrotlp conflict, 111-5 Smith, J., 104-5 texts, 1-6, 8, 32, 42, 182-3, 187 Terrorists, 5, 30-1
looking glasJ sclf, 98 Snitch, 19 comple:<, 186 Testnbility, 65-73, 159, 172
Mnori imag~ of, 104-6 Snivcler, 19 self-referentinl, 183-4 versions of, 66-71
phenomeno!Ógicnl significnnce, 107 Snydcr, C. R., 74 Socin!ization, 98 Tetlock, P., 37, 173
and politicallprnclice, 104 Snyder, M., 121 Socialist C/Jallenge, 135 Te:<t,
public nnd prívate, 99-100 Socccr Sociolinguistics, 7 comprehension, 7
romantic imrlge, 99-100 fans, 56,58-61, 64--7, 160 Sociology, 6, 8, 40, 42, 63, 76 mental organizntion, 7
scientific stutly of, 95 violence, 59-61, 63, 71-2 of science, 65, 72, 159, 172, 174 memory for, 7
nnd sincerityj. 98-9 Social South Africa, 111 Thnt's how it is response, 152
as sociully determined, 97-9 cognition, 121-2, 128, 132-3 Speaker changeover, 81, 87 Theory
thentricnl imhge, 97-9 comparison, 78, 98 Spenker's choice, 56, 65-72, 155, 167, 183
trnditionnl m'odels, 95-101, 115 competence, see competence creativity, ID, 11-4 contingent and empiricist accounts

V1olent, lllt5
1

constructivism, 95, 101-2 intuitions, 10-11, 13-4 of, 152


Westem models of, 105-6 desirnbility, 185 Speech rules of, 65-71, 152
in womcn, tilO differentintion, 185 nccommodation, 7, 36, 38 versus practice, 144-5
Self~nctunlization, 100 influence, 92-3 ncts, 14-8, 23, 28-30, 32, 61, 102 Thompson, J. B., 187
Self~blnming, 7i knowledge, 57 indirect, 29, 61, 172 Tapie, 87
Sclf~control, 112, 115 underlying action and accounlS, errors, ID Topic/resource distinction, 20, 30, 143,
Self-fulfilment, ~8. lOO 57-9 sounds, see signifiers 173
. 68 '
Sclf-prn!se, 1
perception, 36, 52, 178 Springbok rugby tour, 111-5, 164 Total institutions, 19
Self-presentation, 33, 37-8, 60, 97, psychology, 2, 3, 6, 7, 9, 18, 29, 31, Status, 77, 79, 81, 92-3 Totman, R., 58
108-9, '115 . 32, 34-9, 41-3, 53, 56, 63-5, Stereotypes, 36, 52-3, 118 Traditional social science, 19-20, 23, 32,
vnrintion in, 37 74-6,79, 92-3, 116-7, 119-20, consequences of cntegorization, 54-6, 136, 158
Semantics, 184 ! 123, 127, 129, 132, 138, 142-3, 118-23, 132 TrniLS/Trait theory, 96-7, 98, 99, 100,
Scmantics of triYin, 26 146, 157, 169, 173, 175, 178, Stirrer, 113-4 101' 102, 114
Semin, G., 74, p6-8, 90-1 179, 183, 185-6 Stokes, R., 48, 76-7 Trnnscript, 11, 12, 13, 30, 49, 65, 67, 81,
Semiology, l, 9 23-31, 32, 58, 102,
,
of discourse, 14 Strangers, 80 133, 147, 161-6, 168, 172, 174
1

103, 1'"1 European, 138 Stream of consciousness, 108 as constructive activity, 165
Semiotics, see semiology of persunsion, 187 Stringer, P., 35, 174, 183 readability, 166
Scqucntinl disco:urse, 6, 13, 46, 80-l, Jaborntory, 39, 185 Structuralism, 6 speed, 166
83, 88, 92-4, 161, 171, 184 non-cognitive, 178-80 Stubbs, M., 6, 163, 165 word processed, 166
216 Discourse and social psychology

Trnnsfonnntion of responses, 45-6, 53 Video, 59, 118


Transsexual, 126 Vigncues, 77, 79, 85, 93
Trew, T., 136, 187 Violence, see nggression, soccer violence
Trinngulation, 62-4 Van der Vlist, R., 186
Trilling, L., 96-7, 103 Vocnbulnries of motive, 76
Tripp, D., 184
Truth Wnnn, T. W., 95
and fnlsity, 14-7, 62-4,92-3, Warrnnting, see justificntion
149-55 Watson, R., 127, 129, 130-l, 180
will out device, 153-5, 171 Webb, E. J., 162
Tum organizntion, '82, 184 Wcbley, I. A., 28
Tumer, C. F., 40 Weinberg, T., 127, 130
Tumer, J. T., 79 Weiss, C. H., 174
Typologies Well, 84-5
of accounts, 76-8, 90, 93 Wetherell, M. S., 35, 53, 103, 111, 112,
of cars, 127-8 123, 126, 138, 156, 162, 167,
187
Ullian, J., 65 Whitelnw, Lord, 135
Underlying system, 24-31 Wicker, A. W., 53, 63
Ungar, S., 77-80, 86 Wieder, L., 19-23, 28, 30, 72-3, 182
Use versus mention, 144-5 Wilder, D., 118, 120, 132
Wilkes, A., 117-8
Validity, l, 34, 169-72 Willíamson, J., 104
coherencc, 169-70 Wilson, T. D., 177-8
fruitfulness, 169, 171-2 Witness, 87-92
new problcms, 169, 171 Wittgenstein, L., 81, 179
participants' orient.ntion, 169-71 Wooden presentution, 80
Variation in discourse, 5, 32-43, 45, Woolgar, S., 159, 161, 183, 187
49-50, 53-4, 63-4, 67, 67-9, Wootton, A., 22
109, 123-6, 136, 144, 147-8, Wowk, M., 130, 131-2
156, 164-5, 168-70, 181, 185 Wrcstling, 27-8, 58
between and within accounts, 67, 152 Wynne, A., 184
suppression of,.35, 39-43
organizntion of 67-9 Yenrley, S., 30, 136, 152, 156, 159, 187
Verification, 14 Yu1e, G., 6
Versions, 4, 6, 33-4, 51, 63, 66, 73, 92,
122, IJO, 137, 175, 182-3, 186 Zimmerman, D., 20, 22, 143
conflict between, 49-50, 68, 127, 135 Zuckermun, H. A., 65

S-ar putea să vă placă și