Current Titles

Democracy (Second Edition)DlscriminatonCitizenshipWelfare (Second Edtion)PreedomBureaucracy (Second Edition)NationalismTrotskyismRevolution and

Counter-RevolutionPolicySocialismExchangeSocial ScienceSocial DarwinismPowerRghtsScienceUberalism (Second Edition)The StateKnshipSovereigntyDlscourseUtopianismSocial StructurePostmodernity (Second

Edition)Idcology (Second Edition)PluralismI'ascsmConservatismStructurationHace and EthnicityPostcommunismOrientalismCapitalismClassCasteEcologsrnCulturePopulismStatusMulticulturalism

Anthony ArblasterMichael Bantonl. M. BarbaletNorman BarryZygmunt BaumanDavid BeethamCraig CalhounAlex Callinicos

Open University Press

;0

, _1 ('11 i 1"'''7 , 1,"''./1..'! , {' f

IjlL/' fh ~ i i o(I

\j'''i.~'- Concepts in the Social sciences~~-,~

f\l

"1,.'vl~,

Disc-oursePeter CalvertH. K. ColebatchBernard Cricklohn DavisGerard DelantyPeter DickensKeith DowdingMichael FreedenSteve Ful/erIohn GrayIohn A. Hall and G. Iohn lkenberryC. C. HarrisIohn HoffmanDavid HowarthKrishan [(umar{os Lpez and iohn Scott

David LyonDavid McLel/anGregor McLennanMark NeocleousRobert Nisbetiohn Parkerlohn RexRichard SakwaZiauddin SardarPeter SaundersRichard ScaseUrsula SharmaMark SmithMark f. SmithPaul TaggartBtyan S. Turnerc. W. Watson

\'('\)-.l.\'-

\)-,~~

David Howarth

.~

~.~.'''~

"-,;:ir \i" ' ~.~ r: ~~t"'=;'''''='!I::~ ---, ......

::~G, ItO~ ....(\I~J']

~~. ,.)\,J _~:::~!~Ch\!~~2>,;~:/>

\i--.(

~ J

.~

(\~,; .:,1

.1ti..le----------------_ ...."""-"",..".;,,,,,,,., .':..

x Discourse

The writing of this book has inevitably and happily resulted in anumber of debts. Stuart Hall and John Hoffman made valuablecomments on my original pro posal for the book. 1 would alsolike to express my thanks to Alan Cardew, Mark Devenney, JoeFoweraker, Jason Glynos, Alejandro Groppo, David Morrice,Yannis Stavrakakis and Jacob Torfing for their comments andrefections on different parts of the book. 1 am particularly gratefulto Steven Griggs, Ernesto Laclau, Todd Landman, James Tully, andespecially Aletta Norval for reading and commenting on the en tiremanuscript. Not least, 1 would like to thank Frank Parkin, JustinVaughan and Maureen Cox for their comments and suggestions, aswell as Gaynor Clernents and Viv CracknelI at Open UniversityPress for their patience in awaiting final delivery of the manuscript.Finally, 1would also like to express my gratitude to those studentsat Staffordshire University, the ECPR Summer School, and theIdeology and Discourse Analysis programme at the University ofEssex who listened to and criticized some of the ideas and argu-ments presented here. NaturalIy, 1 take full responsibility for thelimitations of the final arguments put forward in the book.

i~;:~~(~~:!~!t!.,.~;

:f~~...:~~

.~~-.),}

~;

.'~

\1.:;:.~,

!'

'..~....,~::.'~'.J,.:.,~..~.

all

....~..il.::,}~

"1~';~~t~~

".~-. ;

lntroducton:Defining the

Concept ofDiscourse

'I'hc concept of discourse plays an increasingly significant role inrontemporary social science. Although originating in disciplines:,I/cl as linguistics and semiotics, discourse analysis has been.xtcuded to many branches of the human and social sciences. Its".lwing prominence is not only evident in the increasing numberI' srudies which use the concepts and methods of discourse analy-t,i:;, hut also visible in the widening scope of its deployment.Scholars in academic disciplines as diverse as anthropology, his-1111 Y and sociology; psychoanalysis and social psychology; cultural,I',"nder and post-colonial studies; political science, public policyun.rlysis, political theory and international relations, not to men-non linguistics and literary theory, have used the concept of dis-1 "lIr~c lo define and explain problems in their respective fields of;;llIdy.l

The reasons for this explosion of interest are complex and itisIIIi( he aim of this book to explore them in any detail. Neverthe-..;s, attcntion ought to be focused on a series of connected factors .'I'hesc include a growing dissatisfaction with mainstream positivistill'pmachcs to social science, and the weakening f its hegemony ind'plines sucl~:as political science and sociology. .It is also a1'" Iuct of the belated impact of the so-called 'linguistic turn' onIhe social sciences, and the consequent rise of new approaches such

2 Discourse

as hermeneutics, critical theory and post-structuralism following inits wake (Dallmayr and McCarthy 1977; Rabinow and Sullivan1979; Rorty 1992a; Finlayson 1999: 47-68). Similarly, the resur-gence of Marxist.theory in the West, as well as the wider dissemi-nation of psychoanalytic discourse in the social sciences, have alsocontributed to a greater pluralization of the social sciences. Finally,the emergence of a distinctive field ofdiscourse analysis within thediscipline of linguistics during the 1970s, and its subsequent take-up by practitioners in disciplines such as cultural studies and liter-ary theory, have led to a novel conception of discourse and aspecific way of conducting discourse analysis (van Dijk 1985, 1997a,1997b; Fairclough 1989, 1992; Jaworski and Coupland 1999b;Willing 1999).

This book focuses on developments in the social sciences, asthere are no general accounts of the way in which the theories andmethods of discourse analysis can be applied to this domain ofstudy and research. This is especially pertinent given that there issome scepticism about the precise epistemological status andmethodological suitability of discourse theory in the social sciences.lt is therefore crucial to show how the concepts and methods ofdiscourse analysis can be 'operationalized' in meaningful ways, thatis, it is important for discourse analysts to apply their abstracttheories and concepts to empirical research questions so as toproduce novel interpretations, and to show the 'added value' oftheir studies in understanding and explaining the social world.

The concept of discourse

In the social sciences, the proliferating 'discourse about discourse'has resulted in rapid changes to the commonsensical meanings ofthe word. For some, discourse analysis is a very narrow enterprisethat concentrates on a single utterance, or at most a conversationbetween two people. Others see discourse as synonymous with theentire social system, in which discourses literally constitute thesocial and political world. For example, Jacques Derrida (1978a:280) argues that 'when language invaded the universal problema tic... everything became discourse', while Ernesto Laclau and Chan-tal Mouffe (1987: 84) use the concept of discourse to 'emphasize thefact that every social configuration is meaningful', in which case 'thediscursiye is coterminous with the being of objects'. In short, as theconcept of discourse has been employed in the social sciences, it has

:~ 3lntroduction

acquired greater technical and theoretical sophistication, whileaccruing additional meanings and connotations.

As with other complex and contested concepts in the social sci-ences, the meaning, scope and application of discourse is relativeto the different theoretical systems in which it is embedded (Con-nolly 1993: 10--44). These theoretical systems are laden with par-ticular assumptions about the nature of the social world and theway that we attain knowledge of it. This means that if one is toprovide a proper 'grammar' of the use of the concept, one needs tobe sensitive to the various theoretical contexts in which it functions.Generally speaking, positivists and empiricists argue that dis-courses are best viewed as 'frames' or 'cognitive schemata', bywhich they mean 'the conscious strategic efforts by groups ofpeople to fashion shared understandings of the world and of them-selves that legitimate and motvate collective action' (McAdam etal. 1996: 6). Viewed as frames, discourses are primarily instru-mental devices that can foster common perceptions and under-standings for specific purposes, and the task of discourse analysis isto measure how effective they are in bringing about certain ends(Snow and Benford 1988).

By contrast, realist accounts of discourse place much greateremphasis on what they call the ontological dimensions of discoursetheory and analysis. Crucial to this ontology is the idea that thesocial world consists of an independently existing set of objects withinherent properties and intrinsic causal powers. The contingentinteraction of these objects with their 'generative mechanisms'causes events and processes in the real world (Harr and Madden1975; Bhaskar 1978, 1979; Harr 1979; Stones 1996: 26-39). Thus,in this conception, discourses are regarded as particular objectswith their own properties and powers, in which case it is necessaryfor realists 'to focus on language as a structured system in its ownright', and the task of discourse analysis is to unravel 'the concep-tual elisions and confusions by which language enjoys its power'(Parker 199?: 28). Moreover, in order to account for the specificcausal impact of these objects they need to be placed in relation toother social objects, such as the state, economic processes, and soon. In short, this approach stresses the underIying 'materialresources which make discourses possible', arguing that the 'studyof the dynamics which structure texts has to be located in anaccount of the ways discourses reproduce and. transforrn thematerial world' (Parker 1992: 1).

4 Discourse

While sharing the underlying assumptions of realism, Marxistsstress the way in which discourses have to be explained by refer-ence to the contradictory processes of economic production andreproduction. In this perspective, discourses are normally viewedas ideological systems of meaning that obfuscate and naturalizeuneven distributions of power and resources, This means that dis-course analysis has the critical task of exposing the mechanisms bywhich this deception operates and of proposing emancipatoryalternatives (Althusser 1969, 1971; Pcheux 1982; Zizek 1994).

Norman Fairclough and his school integrate a wide range ofsociological and philosophical currents of thought, including thework of Antonio Gramsci, Mikhail Bakhtin, Louis Althusser,Michel Foucault, Anthony Giddens and Jrgen Habermas, todevelop what they call critical discourse analysis (Fairclough 1989;Wodak 1996; Fairclough and Wodak 1997). For instance, Fair-clough and Ruth Wodak (1997: 259-60) use Giddens's theory ofstructuration to provide an overall sociological framework withwhich to conduct discourse analysis. Giddens's (1984) theory ofsociety differs from positivist, realist and Marxist accounts in thathe stresses the centrality of human meaning and understandingin explaining the social world. His explicitly 'herrneneutically in-formed social theory' thus places greater emphasis on the actionsand reflexivity of human agents in reproducing and changing socialrelationships. Fairclough takes up this theme of 'the duality ofsocial structure and human agency' by insisting that there is amutually constituting relationship between discourses and thesocial s.yS,tems in which they function. The task of discourse analy-sis is thus'to examine this dialectical relationship and to expose theway in which language and meaning are used by the powerful todeceive and oppress the dominated.

FinalIy, post-structuralists and post-Marxists such as JacquesDerrida, Michel Foucault, Ernesto Laclau and Chantal Mouffe putforward much more comprehensive concepts of discourse. They gofurther than the hermeneutical emphasis 00 social meaning byregarding social structures as inherentIy ambiguous, incomplete andcontingent systems of meaning. For instance, Derrida (1978a, 1982)argues for a conception of discourse as text or writing, in which al!human and social experience is structured according to the logic 01"diffrance; while Foucauldiandiscourse analysis is intent 011showing the connection between 'discursive practices' and widcrsets of 'non-discursive' activites and institutions (Foucault 19n,

;ti~h

N

I~

~~;]

~~[iJ~1:fi~~-~~f,--;n~~:JH~1,

~!g::

II

,......1

1...

\')'

-l"l

'~

~' ~~'i,

-;-"--,!

1

I Introduction s1981, 1991a). For their part, Laclau and Mouffe (1985, 1987) decon-

- struct the Marxist theory of ideology and draw upon post-struc-turalist philosophy to develop a concept o discourse that includesall the practices and meanings shaping a particular community osocial actors. In these perspectives, discourses constitute symbolicsystems and social orders, and the task of discourse analysis is toexamine their historical and poltica} construction and functioning.The approach developed in this book falls squarely into the post-structuralist and post-Marxist traditions o analysis, although 1 willdistinguish and defend it against the other perspectives.

~iAims and arguments of this book

This book has five main aims. To begin with, 1examine a number odifferent meanings and uses of the concept of discourse by concen-trating on its deployment to ever wider aspects o the social world,1 sketch out a movement that begins with a narrow and technicalconception of discourse analysis, still evident in maiostream posi-tivist and empiricist research programmes, which is then progres-sively broadened and deepened by the emergence and extension ofstructuralist thought. 1 then examine how the critique of structural-ism gives way to post-structuralist and post-Marxist approaches to.Iiscourse analysis. For obvious reasons o time and space thisJ',cnealogy is not exhaustive. Hence it does not include Paull: icoeur's (1971, 1976) hermeneutic account of discourse, nor doesI1 examine Jrgen Habermas's (1984, 1987a) theory of communi-, ;11 ivc action, though it does consider the latter's critiques of post--,fllldmalist models and does endeavour to clarify the relationshipIwlw'TIl post-structuralist and hermeneutical approaches .

:;''1'111\(1,1 pul Iorward a particular approach to discourse theory_'".1 .uuilvsi, which brings together three anti-positivistic tradi-1" ,n, ," iuicllcctual inquiry - post-structuralism, hermeneutics and1'" 1 f\l:llxi~;I\l_ More specifically, 1 articulate aspects of Michel, ,,", .1111':; ;Ippruach to discourse analysis, especially his various1'" 11o '"IIlI,.ical suggcstions and strictures, with the post-Marxist, ,,I, 101'''11 ,,1"diSCOl\fSC devcloped by Ernesto Laclau and Chantal- 1 - 'ti If,

1100;,1 I n;lllIill

6Discourse

rein to the idea that there are no constraints on social and politicalaction to the extent that 'anythinggoes'. While 1 show that thesecriticisms are wide of the mark, 1 do argue that there are certainaspects of discourse theory which are in need of further refinementand development, and 1 offer ways of extending the research pro-gramme.

Fourth, 1 consider some of the methodological devices that dis-course theorists have developed, and need to develop further, whenapplying discourse theory to empirical research objects. In particu-lar, 1 draw upon styles of research gleaned from the writings ofJacques Derrida, Michel Foucault and Ferdinand de Saussure inorder to counter the charge that discourse theorists and analysts areno more than methodological anarchists, relativists and 'armchartheorizers' .

Finally, in the course of developing these arguments, 1present aseries of empirical examples and cases designed to ilIustrate someof the advantages of employing discourse theory, as well as to indi-cate areas of possible future research. 1 also examine some import-ant theoretical questions that can fruitfully be addressed from adiscursive approach. These include theorizing the relationshipbetween structure, agency and power; the complexities of politicaIidentity and difference; the construction of hegemonic formations;the production of subjectivity and the Iogic of decision making; andthe connection between the role of identities and interests in thesocial sciences. Let us consider each of these objectives in moredepth.

A brief genealogy 01discourse

Very schematically, theories of discourse have undergone threesig-nificant transformations. Traditionally, discourse analysis is .con-cerned wth the investigation of 'language in use' and attention isfocused on the analysis of 'talk and text in context' (van Dijk 1997b:3). In this view, discourse analysis is rather narrowly construed andfocuses primarily on the rules governing connected sets of sen-tences in speech or writing. For example, speech act theory focuses00 the fact that by saying something we are also doing somethiog.When someone utters a statement such as '1 promise' or '1 namethis ship the Queen Mary', and meets their requisite 'felicity con-ditions' - in other words, they do intend to keep their promises orare authorized to narne ships - they are also performing an act. In

~~?

. ".!',:,

Introduction

the language of analytical philosophers such as J. L. Austin (1975)- and John Searle (1969), they are performing speech acts. Discourse

analysts working in this tradition have elaborated complex typolo-gies of different sorts of speech act and have tried to explain differ-ent aspects of communication, such as psychiatric interviews, bytrying to identify the intended meanings of a speaker's utteranceand the responses of hearers (Labov and FansheI1977).

In a related vein, conversation analysts drawing largely onGarfinkel's (1967) sociological method of ethnomethodology,which is the study of the way in which individuals experience theireveryday activities, endeavour to deduce from observation whatspeakers fredoing and how they are doing it (Trask 1999:57). Moreconcretely, discourse analysts such as Schegloff and Sacks (1973)have examined the organization and logic of 'turn-taking' in con-versations. For instance, their research shows that a key principIethat structures conversations is the avoidance of 'holes' and 'inter-sections' between speakers. A further aspect of this research hasconcentrated on the accepted principles that usual1y govern thelogic of 'turn-taking' in conversations, in which speakers adoptcertain 'speaker roles' and are encouraged to speak by con-ventional markers such as body language, gaze, ton e and certainritualized words. According to this particular form of discourseanalysis, these insights enable us to understand 'patterns of indi-vidual relations between interactants, individuals' positions withinIarger institutional structures, and overall societal organisation'(Jaworski and Coupland 1999a: 21).

However, in the wake of the growing centrality of structuralism,post-structuralism, hermeneutics and Marxism in the social sci-ences during the 1960s and 1970s, the concept of discourse hasbeen extended to a wider set of social practices and phenomena.In this regard, the work of Michel Foucault is particularly perti-nent. In his earlier 'archaeological' writings, Foucault (1970, 1973)stress es the way discursive practices form the objects and subjectsof discursive formations. Discourses are thus 'practices that sys-tematical1y form the objects of which we speak' (Foucault 1972:49), and they consist of historicaIly specific 'rules of formation' thatdetermine the difference between grammatical1y well-forrnedstatements and 'what is actuaIly said' at particular times and places .(Foucault 1991a: 63). As against empiricist, realist and Marxistconceptions, in which the nature of the objective world determinesthe character and veracity of discourses, Foucault argues that

7

';1

.:-;:;:~

8 Dtscourse

certain discursive.rules enable subjects to produce objects, state-ments, concepts and strategies, which together constitute dis-courses. In his later 'genealogical' writings, Foucault (1987)modifies his quasi-structuralist conception of discourse.Ratherthan describing the historical rules that make possible sets of state-ments, Foucault is now concerned with the way in which discoursesare shapedby social practices and the way they in turn shape social.relationships and institutions. Throughout, Foucault's approach todiscourse analysis ernphasizes the methodological requirements ofsuch an enterprise, and he is at pains to think critically about thedifferent strategies and techniques of conducting research.

The third phase ofdiscourse analysis, which develops partly outof Foucault's various contributions, and partly from Derridean,Marxist and post-Marxist insights, considerably expands the scopeof discourse analysis, so as to inc1ude non-discursive practices andelements, Fairclough's (1989: 25, 2000) critical discourse analysiswidens the focus of discourse theory to include the analysis of politi-cal texts and speeches, as well as the contexts in which they are pro-duced. However, discourses are still understood as the sernioticdimension of social practice, and thus remain a distinct level of theoverall social system. By contrast, Laclau and Mouffe's approach,which I shall simply call discourse theory, enlarges the scope of dis-course analysis to include all social practices, such that discoursesand discursive practices are synonymous with systems of socialrelations. In this book largue for the latter perspective, and it is toits basic contours that I shall now turno

Discourse theory

Discourse theory begins with the assumption that all objects andactions are meaningful, and that their meaning is a product of his-torically specific systemsof rules. It thus inquires into the way inwhich social practices construct and contest thediscourses thatconstitute social reality. These practices are possible because sys-tems of meaning are contingent and can never completely exhausta social field of meaning. Three basic categories are needed tounpack and elaborate upon this complex set of statements.Theseare the categories ofthe discursive, discourse and discourse analy-sis. By the discursive I mean that all objects are objects of dis-course, in that a conditiou.of their meaning depends upon

10 Discourse

discourse of social democracy and those who wanted its radicalrestructuring.

Discourse analysis refers to the process of analysing signifyingpractices as discursive forms. This means that discourse analyststreat a wide range of linguistic and non-Iinguistic material -speeches, reports, manifestos, historical events, interviews, policies,ideas, even organizations and institutions - as 'texts' or 'writings' thatenable subjects to experience the world of objects, words and prac-tices. This allows discourse theorists to draw upon and develop anumber of concepts and methods in linguistic and literary theorycommensurate with its ontological assumptions. These includeDerrida's 'method' of deconstruction, Foucault's archaeological andgenealogical approaches to discourse analysis, the theory of rhetoricand tropes, Saussure's linguistic distinctions, Wittgenstein's concep-tion of rule following, and Laclau and Mouffe's logics of equivalenceand difference (Howarth 1998: 284--8; Stavrakakis 1999: 57-9,76-8).

Three traditions

Discourse theory critically engages with the structuralist,hermeneutical and Marxist traditions of thinking. To begin with,the structuralist tradition .of thought is the major influence on thedevelopment of discourse theory. As developed by thinkers such asFerdinand de Saussure, Roman Jakobson and Louis Hjelmslev,structuralists focus on the way in which meaning and significationare a product of a system of signs. In opposition to theories of Ian-guage which assert that words and language refer to a world ofobjects, structuralists argue that meaning depends on relationsbetween different elements o a system. For instance, to understandthe meaning o the word 'rnother', one must understand relatedterms like 'father', 'daughter', 'son' and so on (LacJau 1993: 432).Meaning is thus an effect o the formal differences between terms,and not the result o any correlation between words and things, oran inherent characteristic of texts, objects or practices.

While originators of the structuralst model of language such asSaussure were convinced that it could be extended to aIl signify-ing systems and practices, they did not provide the resources tocarry out such a project. This precJuded the emergence of dis-course analysis as the original linguistic model could only beapplied to the analysis of signifiers, words, phrases, expressionsand sentences. It was thus left to later structuralist theorists such

t~

~!

-~

-%'

Introduction 11

as Claude Lvi-Strauss, Jacques Lacan, Louis Althusser andRoland Barthes to use the structural model of language to eluci-date a greater range of social phenomena. These included theexplanation of the role of myths in society, the formation ofhuman subjectivity in language, the analysis of different modes ofproduction and social formations, as well as the exploration ofvarious symbolic codes, such as cooking, eating and playing sport,that give meaning to everyday life in society.

Moreover, as the original model was extended to cover a greaterrange of social phenomena, so other writers began to identify weak-nesses in the key assumptions of structuralist thinking. In particu-lar, questions were raised about the historical construction ofsystems, -the fixed relations between elements of systems, and theexclusion of human subjectivity and agency from the social world.In endeavouring to resolve difficulties of this sort, writers such asJacques Derrida, Michel Foucault and Ernesto Laclau and ChantalMouffe were to question some of the underlying assumptions ofstructuralism, thus making possible what have become known aspost-structuralist ways of thinking.

In the second place, the discursive approach advocated in thisbook borrows selectively from the hermeneutical tradition ofinquiry. On the one hand, discourse theory stands against thosepositivist, behaviouralist and structuralist accounts of social lifethat concentrate simply on observable facts and actions, or whichdisregard everyday social meanings in favour of unconscious struc-tural laws, Instead, discourse theorists draw upon hermeneuticalphilosophers such as Martn Heidegger, Ludwig Wittgenstein,Charles Taylor and Peter Winch in order to interpret the meaningsand self-understandings of actions, rather than pinpointing theircausal mechanisms. This means that one of the major goals of dis-cursive social inquiry is to discover the historically specific rules andconventions that structure the production of meaning in a particu-lar social context.

On the other hand, discourse theorists do not endeavour touncover the underlying meanings of social practices that aresomehow concealed from actors, nor do they simply seek torecover the interpretations actors give to their practices. This sug-gests that meanings reside in social practices waiting to beretrieved or discovered by the interpreter and that once discernedcan be communicated transparently from one subject to the next.By contrast, drawing on post-structuralist theories of language,

1~~ . , -~t?5T'C'~-' ~tettt'tiW~\>;.~~(

12Discourse

discourse theorists understand meaning as an effect of 'the play ofsignifiers', and argue that the very conditions that make possiblethe transmission of meaning -language as a pre-existing system ofdfferences - also render it problematical. Moreover, rather thanlocating interpretations at either the surface or depth levels ofsociety, they seek to provide new interpretations of social practicesby situating their meanings in broader historical and structuralcontexts.

Marxisrn constitutes the third major influence on the emergenceand development of discourse theory. The distinctive aspect of theMarxist approach to discourse is the way in which ideas, languageand consciousness are regarded as ideological phenomena thathave to be explained by reference to underIying economic andpolitical processes. It is also concerned with the role of social agentsin criticizing and contesting relations of exploitation and domi-nation. However, the models of ideology in classical Marxismreduced ideologies to more determina te social processes such aseconomic production and class struggle. It was thus left to laterMarxist theoreticians such as Antonio Gramsci, Louis Althusserand Michel Pcheux to try and develop non-reductionist and anti-essentialist accounts of society and historical change. These writersstress the material and practical characteristics of ideology, ratherthan its purely mental or (mis)representational qualities. Never-theless, despte these advances, they remain imprisoned withinthe overarching assumptions o Marxist tbeory. By drawing upon .structuralist, post-structuralst and hermeneutical traditions ofthinkng, post-Marxists have begun to elaborate a relational andanti-essentalist approach to the study of discourse by pursuingpossibilities foreclosed in the Marxst tradition. It is this synthesisthat constitutes the main focus of this book, though it is far fromuncontroversial.

Critique and evaluation

The theory of discourse that I have begun to outline is strongly con-tested and has provoked considerable debate (cf Geras 1987,1988,1990; Dallmayr 1989; Mouzelis 1990; Osborne 1991; Aronowitz1992; Mouffe 1996;Sim 1998;Wood 1998). Realist, Marxist and pos-itivist critiques concentrate on the alleged idealism and textualism ofdiscourse theory, arguing that it reduces social systems to ideas andlanguage. For them, this has the consequence of neglecting the

lntroduction 13

!~ material conditions, institutions and natural constraints on the pro-duction and transformation of social meanings. Critics also allegetbat discourse theorists fall prey to conceptual and moral relativismwhich renders them incapable of making claims to truth and valid-ity, and/or of making objective value judgements about the objectsthey study. Positivists accuse discourse theorists of abandoning thesystematic collection of objective facts, and of substituting subjec-tivist and methodological anarchistic accounts of social phenomena.Similarly, behaviouralists argue that discourse theorists' concernwith meanings and language precludes a value-free inquiry intosocial and political behaviour, and entails an over-exaggeration ofideological and subjective factors.

In evaluatng discourse theory, I argue that existing critiques areunsatisfactory beca use they concentrate their attacks on the onticalrather than ontological levels of analysis. 1 show that discoursetheory does not reduce the social world to language understoodnarrowly as text or speech. Instead, it makes a useful analogybetween lnguistic and social systems, thus providing a powerfulmeans lo conduct social and political analysis. 1 also argue thatdiscourse theory's underlying ontological and epistemologicalassumptions circumvent the charge of relativism which has beenbrought by a number of commentators. However, I do not presentdiscourse theory as a complete and wholly unproblematic approachto social and political analysis. Although it opens up powerful newways of interpreting and evaluating empirical evidence, there aresome important theoretical and conceptual issues that need furtherclarifcation if discourse theory is to make a meaningful contri-bution to our analysis of the social and political world. Amongstthese are a series of methodological queries about the strategiesand styles of carrying out discourse analysis.

Applying discourse theory

An important aim of this book concerns the application of discoursetheory to specific empirical cases showing how this approach can beused to study different aspects of society and politics. As againstcharges of 'methodological anarchism' or 'epistemological ir-rationalism', which have sometimes been levelled al the approach, 1put forward styles of research that can 'operationalize' its overarch-ing assumptions about society and politics. 1also tackle the questionof applying discourse theory to empirical cases without subsuming

/j(']' I.Jf reani!

Saussure,Structuralismand Symbolic

Systems

This chapter examines the structuralist contribution to a theory.ofdiscourse. Structuralism was inaugurated by the Swiss linguist, Fer-dinand de Saussure, and is a central component of the so-called 'lin-guistic turo', which played a powerful role in our understanding ofphilosophy and the social world during the twentieth century. Theconcem with questions of meaning and signification not onlyprompted investigation into the nature of language itself, but alsoresulted in the extension of linguistic models into the social sciencesmore generally. Structuralists emphasize that all hurnan actions andsocial institutions are best viewed as symbolic systems of practice,and researchers in the social sciences have deployed the methodsand assumptions of structuralism to develop sophisticated concep-tions of social formations, and to explain events suchas revolutionsand the actions of states (see, for example, Poulantzas 1973;SkocpoI1979). In this chapter, however, 1 concentrate on the wayin which structuralisttheory provides the conceptual resources fordeveloping a viable conception of discourse, and 1 isolate the limit-ations of the original structuralist account of social systems.

Structuralism and discourse

Structuralist theoryprovides an important starting point fordeveloping a viable concept of discourse in the social sciences by

~,,,~~.,,,, "trt&;,,; ..

1

17Saussure, Structuralism and Symbolic Systems

assuming that there is a cIear anaLogy between language and socialrelationships. For instance, Jacques Lacan (1977:147) claims thatthe human unconscious is 'structured like a language', and thestructural anthropologist Claude Lvi-Strauss (1968, 1977) arguesthat social relations in 'primitive' societies can be treated as iftheywere linguistic structures. In this conception, both societies and lan-guages are perceived to share similar logical structures and fea-tures. This means that phenomena as diverse as social formations,political ideologies, myths, family relationships, texts and wrestlingmatches can all be understood as systems of related elements (seeBarthes 1973). Without entering into a detailed discussion of thedifferent types of structuralism at this stage, this means that theindividual lements of a system only have significance when con-sidered in relation to the structure as a whole, and that structuresare to be understood as self-contained, self-regulated and self-transforming entities. Hence it is the structure itself that determinesthe significance, meaning and function of the individual elements ofa systern (Piaget 1971: 5..:..16;-Uawkes-t97i:1;;.:;t8;.----- -----------

This means that structuralists do not seek to reduce socialphenomena to underlying causes or deterrninants, nor do they treatsocial reality as a random agglomeration of atomized events orfacts. Instead, they argue that apparently unrelated or inexplicableevents or processes can be made inte1ligible by reference to aformal system of relationships. The process of discovering a systemof relationships with which to delimit or individua te an essentialset of elements is predicated on a new method of analysis (de-rived from the study of language and mathematics). In brief, thismethod consists of defining social phenomena as relations betweenelements, the construction of a possible -set of permutationsbetween them, and an analysis of their actual relationships (Lvi-Strauss 1969: 84). Very simply, if we were to explain the game ofchess to a beginner, we would first identify the different pieces(king, queen, bishop), then explain their possible interactions(moving, capturing, checking), before finally observing their actualinteractions in a real game between two players.

The emergence of a .structuralist conception of society beginswith Saussure's theory of language. While he did not develop a dis-tinctive concept of discourse, he stands squarely atthebeginning ofour inquiry. The reason for this paradox is the enormous methodo-Iogical and substantivecontribution he made to the.social sciences,and the way he makes possible a discursive approach .to social andpolitical analysis. Along with writers such as Marx, Durkheirn and

14 Discourse

them under its abstract categories or giving in to naive positivismoMoreover, 1argue that within discourse theory there is a set of yard-sticks with which to measure and evaluate the plausibility and ade-quacy of the ernpirical accounts made in its name. Drawing on thephilosophy of Heidegger and Foucault, 1argue for a complex theoryof truth in which the truth and falsity of cases is relative to a frame-work of meaning (or paradigm) within which problems are identi-fied and analysed. Thus it is the community of scholars that formsthe ultimate tribunal of judgement in the social sciences and it is theproduction of paradigm studies that determines the progressive ordegenerate status of the discourse theory research programme.

Plan of the book

The chapters and structure of the book reflect its principal aims andarguments. Chapter 1 examines the structuralist model of languageand its implications for developing a discursive approach to socialand political analysis. 1 begin by introducing Saussure's structural-ist theory of language, after which 1consider the way in which Lvi-Strauss deploys this model of language to analyse social relations assymbolic systems. Chapter 2 draws upon Derrida's deconstructionof the structural model of language to argue that Saussure and Lvi-Strauss are unable to develop a coherent conception of discourse.Instead, Derrida's deconstructive methodology uncovers possi-bilities that are foreclosed by Saussure and the structuralist tra-dition, thus making possible the development of a post-structuralisttheory of discourse understood as writing or text.

Chapters 3 and 4 evaluate Foucault's different accounts of dis-course theory and analysis. 1 begin by considering Foucault'sarchaeological account of discourse as elaborated in his earlier writ-ings. 1argue that this audacious attempt to develop an autonomousand critical account of discursive practice runs aground beca use itcontradicts Foucault's stated aim of providing a purely historicaldescription of meaningless statements. Chapter 4 shows howFoucault's later approach to the study of discourse, manifest in hismethods of genealogy and problematization, provides a much moresecure basis for anaIysing the relationship between discursive andnon-discursive practices.

In Chapters 5 and 6, 1 explore Laclau and Mouffe's post-Marxisttheory of discourse. Chapter 5 examines the classical Marxisttheory of ideology and politics, and evaluates the degree to which

15lntroduction

western Marxists such as Gramsci, Althusser and Pcheux wereable to transcend the limitations of the Marxist approach. It con-eludes with a deconstructive account of the Marxist approach toideology and politics, thus paving the way for a consideration of thepost-Marxist alternative. Chapter 6 introduces and evaluatesLaclau and Mouffe's post-Marxist theory of discourse. and criticaByexplores the approach to social and political analysis that followsfrom this conception. It concludes by raising and addressing a seriesof questions in need of further clarification and investigation.

Chapter 7 concludes the book by discussing the applicationof discourse theory to actual empirical research It raises andaddresses a number of epistemological and methodological diffi-culties encountered in conducting discourse-theoretical research.These include questions pertaining to the definition of objects ofstudy, the appropriate methods and styles of discourse analysis, theso-caBed application problem raised above, and ssues surroundingthe generation and evaluation of evidence. By drawing on a rangeof existing discursive studies, 1 show how these difficulties can beovercome. The chapter concludes with a set of methodologicalguidelines for applying discourse theory.

Note1 See, for anthropology,historyand sociology(White 1978,1987;Clifford

1988;Dant 1991;Jenkins 1991;Munslow1992;Hall 1998);for psycho-analysisand social psychology(Potter and Wethere1l1987;Burman andParker 1993);for cultural, gender and post-colonialstudies (Hall 1997;Williams and Chrisman 1993); for political science, public policyanalysis,political theory and international relations (Apter 1987,1997;Dryzek 1994,1997; George 1994;Hajer 1995;Milliken 1999;Torfing1999);and for linguisticsand literary theory (Coulthard 1977;Fowler1981;Jaworskiand Coupland 1999b;Williams1999).

1'" byvbh 'i't-""*m;~.,.:;~

ll Discourse

H!'nd. SlItlssurc stresses the rol e of social systems in understandinghmnnn societies. This focus contrasts starkly with the centralityuscribed to individuals, events, facts or evolutionary processes inpositivist, empiricist and speculative approaches to the social sci-ences (CuIler 1974: xii). Instead, Saussure emphasizes the sharedsystems of signs which make up our natural languages. Words,symbols and other forms ofcommuncaton require ashared set ofnorms and rules that human beings learn and internalize. This isal so true, he thought, for other social practices and activities, suchas playing games, voting in an election, and engaging in civil dis-obedience. AIl these practices acquire their meaning, and are thusmade possible, if both participants and observers (such as socialscientists) agree or disagree upon a set of institutionalized rules. Inshort, rather than treating social phenomena as discrete and iso-lated entities, Saussure stresses the overall social contexts in which-actions take place and are rendered intelligible.

Saussure's theory of language

In his Course in General Linguistics, Saussure provides a distinctiveobject oflinguistic theory. He begins by distinguishing between syn-chronic and diachronic linguistics: the former perspective considerslanguage as a system of related terms without reference to time;whereas the latter dimension refers to the evolutionary develop-ment of language over time (Saussure 1974: 81). Saussure separatesthese two aspects and privileges the synchronic dimension. How-ever, this does not mean that Saussure disregards linguistic change,as he argues that it is only beca use language can be viewed as a com-plete system 'frozen in time' that linguistic change can be accountedfor (Culler 1976: 35-45). Without the synchronic perspective therewould be no means for charting deviations from the norm.

Saussure's (1983: 15) second major contribution can be capturedin his deceptively simple, though revolutionary, proposition that'language is a system of signs expressing ideas'. Language as asystem of signs, or langue as he names it, consists of the necessaryset oflinguistic roles that speakers oflanguage must adhere to if theyare to communicate meaningfully. This 'sum of word-images storedin the mindsof all individuals'is rgorously contrasted with 'speech'or parole, which refers to individual acts of speaking (Saussure 1974:13-15). In distinguishing language and speech, Saussure (1974:13-14)is separating: '(1) what is social from whatis individual;and

,.'i::: '_:.II.i i"--I'"..IIII1I1I1 III.~,,ric

~1

1:.~tii{'&~j1::'~j'i'

.~,f

1i:le'f!

1

ii

Saussure, Structuralism and Symbolic Systems t9(2) what is essential from what is accessory and more or less acci-dental'. In other words, each individual use of language (or 'speech-event') is only possible if speakers and writers share an underlyingsystem of language. Just as anyone who wishes to play chess mustlearn the roles of chess before being able to play properly, so anyonespeaking and writing a language must assimilate the system oflin-guistic roles that make up the language.

The basic elements of a language for Saussure are signs. Signsunite a sound-image (signifier) and aconcept (signified). Thus thesign cat consists of a signifier that sounds like lkat! (and appears inthe written form as 'cat'), and the concept of a 'cat', which the sig-nifier designates. A key principIe of Saussure's (1974: 68) theoryconcerns jhe 'arbitrary nature of the sign', by which he means thatthere s no natural relationship between signifier and signified. Inother words, there is no necessary reason why the sign cat is associ-ated with the concept of a 'cat'; it is simply a function and conven-tion of the language we use.

However, Saussure (1974: 65) does not assert that the functionof language is simply to name or denote objects in the world. Thisnominalist conception of language would assume that languagesimply consists of words that refer to objects in the world (seeWittgenstein 1972; Harris 1988). Such a view would imply a fixed,though ultimately arbitrary, connection between words as names,the concepts they represent, and the objects they stand for in theworld. According to Saussure, however, meaning and significationoccur entirely within the system of language itself. Even objects donot pre-exist concepts, but depend on language systems for theirmeaning. As Jonathon Culler (1976: 22) notes, the English word'cattle' has undergone significant conceptual change, meaning atone point property in general, then being restricted to four-footedproperty, beforefinally acquiring the sense of domesticatedbovines. Simlarly, different languages do not simply denote objectsin different, but perfectly translatable terms. For instance, Lvi-Strauss (1972: 17-22) uses the French words bricoleur and bricolageto account for the way in which myths are constructed. However,as there are no equivalent terms in English, it is difficult to providea direct translation of them, and they tend to be left in their orig-inal French formo In this sense, it can be argued that the words inlanguages articulate their own sets of concepts and objects, .ratherthan actng as labels for pre-given objects (Holdcroft 1991: 11-12,48-50).

l'tr:r~f~El4z!;, ....

20 Discourse

In its most radical form, therefore, the arbitrary nature of the signhas revolutionary implications. Saussure is not only claiming that itis arbitrary which signifier is connected with which signified, he alsoargues that there is no really existing property that fixes the signi-fied (Culler 1976: 23). Both are fixed solely by their relationships toother signifiers and signifieds in a particular language. This culmi-nates in a relational and differential conception of language, ratherthan a realist or essentialist perspective. Language comprises asystem of linguistic and conceptual forms whose identities are notfixed by reference to objects in the world, but by their internaldifferences (Sturrock 1979: 10). For instance, 'mother' derives itsmeaning not by virtue of its reference to a type of object, butbecause it is differentiated from 'father', 'grandmother', 'daughter'and other related terms.

This argument is captured by two further principies of his theory.They are that language is 'form and not substance', and that lan-guage consists of 'differences without positive terms' (Saussure1974: 113). Tbe thesis that language is 'form and not substance'counters the idea that the sign is just an arbitrary relationshipbetween signifier and signified. This proposition suggests that signssimply connect signifiers and signifieds, but are still discrete andindependent entities. However, this would be to concentra te solelyon what Saussure calls signification, which captures the way inwhich signifiers literaIly signify a particular concept. Such a viewdisregards the fact that for Saussure (1974: 114) language is also 'asystem of interdependent terms in which the value of each termresults solely from the simultaneous presence of the others'. Toexplain the paradox that words stand for an idea, but also have tobe related to other words in order to acquire their identity andmeaning, Saussure introduces the concept of linguistic value. Heintroduces the concept via a number of analogies. For instance, hecompares language to a game of chess arguing that a certain piece,say the knight, has no significance and meaning outside the contextof the game; it is only within the game that 'it becomes a real, con-crete element ... endowed with value' (Saussure 1974: 110). More-over, the particular material characteristics of the piece, whether itbe plastic or wooden, or whether itresembles aman on a horse ornot, do not matter. Its value and function are simply determined bythe rules of chess, and the formal relations it has with the otherpieces in the game.

Similarly, linguistic value is doubly determined. On the one hand,

.""' _IOIIIIIIiIII lIiIiIIi_iiIII .!'!'iIS!f~,

Saussure, Structuralism and Symbolic Systems 21

a word represents an idea (entities that are dissimilar), just as apiece of stone or paper can be exchanged for a knight in chess. Onthe other hand, a word must be contrasted to other words that standin opposition to it (entities that are similar), just as the value androle of the knight in chess is fixed by the rules which govem theoperation of the other pieces. In sum, the value of a word is notdetermined merely by the idea that it represents, but by the con-trasts inherent in the system of elements that constitute language(langue). Here Saussure (1974: 115) compares words with anexample of value drawn from econornics by considering the case ofa coin such as a pound. The pound signifies the amount of moneythat can be exchanged for a commodity such as a bus ticket.However, tbe value of the pound depends on the other denomina-tions o money that make up the monetary system or differenlsystems of currency such as dollars. Thus a pound is worth one-tenth of the value of a ten pound note, and might be equivalent to1.7 US dollars depending on the exchange rate at a given point intime.

These reflections culminate in Saussure's (1974: 120) final theor-etical principle that in language there are only 'differences withoutpositive terms':

Whether we take the signifiedor the signifier,language has neitherideas nor sounds that existed before the linguisticsystem, but onlyconceptual and phonic differences that have issuedfrom the system.The ideaor phonicsubstancethat a signcontains isof lessimportancethan the other signsthat surround it.

However, this stress on language as apure system of differences isimmediately qualified, as Saussure argues that it holds only if thesignifier and signified are considered separately. When united intothe sign it is possible to speak of a positive entity functioning withina system of values:

Whenwecompare signs- positive tenns - with each other, we can nolonger speak of difference; the expressionwould not be fitting,for itapplies only to the comparing of two sound-images,e.g. [ather andmother, or two ideas, e.g. the idea 'father' and the idea 'mother'; twosigns,each having a signifiedand signifier,are not different but onlydistinct. The entire mechanism of language, with which we shall beconcernedlater, isbased on oppositionsof thiskind and on the phonicand conceptualdifferencesthey imply.

(Saussure 1974:121)

22Discourse

Thus in the formal and relational theory of language that Saus-sure advocates, the ideruity of any elernent is a product of the differ-ences and oppositions established by the elements of the linguisticsystem. Saussure charts this conception at the leveIs of signification- the relationships between signifiers and signifieds _ and withrespect to the values of linguistic terms such as words. Spoken sig-nifiers are differentated at the phonemic Ievel, Thus the phonerneIp/ is defined in opposition to other phonemes, such as Ib/ or 11/,inwords such as 'pet', 'bet' or 'let'. Apart from the substitution of otherphonemes, the phoneme Ip/ can form words by being combined withother phonemes to make words such as 'pot', 'pit' and 'pat'. EquaUy,written signifiers, or graphemes as linguists call thern, are distin-guished by the oppositions they establish with other letters in ourwriting. As long as we can distinguish between the letter 'd' andother letters, we can read and make sense of writing. Similarly, tounderstand the meaning of a term such as 'rnother' requires us tounderstand the system of family terms - 'father', 'daughter', 'grand-mother' and so forth - against which the concept of 'rnother' can bedistinguished. The same would be true of colour terms, in which theidentity of 'pink' is only possible when distinguished from 'red','blue', and 'brown' for example (cf. Wittgenstein 1977).

Languages thus comprise differences and relationships. Thedifferences between signifiers and signifieds produce linguisticidentities, and the relationships between signs combine to formsequences of words, such as phrases and sentences. In this regard,Saussure (1974: 122-7) introduces a further conceptual divisionbetween the associative and syntagmatic 'orders of values' in Ian-guage. These two orders capture the way in which words may becombined into linear sequences (phrases and sentences), or the wayin which absent words may be substituted for those present in anyparticular linguistic sequence. For example, in the sentence 'The calsat on the mat' each of the terms acquires its meaning in relationto what precedes and folIows it. This is the syntagmatic ordering oflanguage. However, others can substitute for each of these terms.'Cat' can be replaced with 'rat', 'bat or 'gnat' and 'mat' can bereplaced with 'carpet', 'table' or 'floor'. This is what Saussure calIsthe associative ordering of language and is derived from the way inwhich signs are connected with one another in the memory (Harris1988: 124).

According to Saussure, the principIes of associative and syntag-matic orderng are evident at alI levels of language. This ranges

il

Saussure, Structuralism and Symbolic Systems

rom the combination and association of different phonemes intowords to the ordering of words into sentences and discourse. ThusSaussure is able to analyse relations within different levels of lan-guage, and he is able to analyse the relations between differentlevels, while still employing the same basic principIes he enunciates.This constitutes the structuralist dimension of his theory, and pro-vides one of the key reference points for developing a structuralistmethodology in the social and human sciences. It is to this endeav-our that we now turno

23

i~~~.~

Society as a symbolic system

Claude Lvi-Strauss (1968: 33) extends structurallinguistics to thesocial sciences in his attempt to develop a structural analysis ofanthropological phenomena. He presents four ways in which struc-turallinguistics can renovate the social sciences. It shifts attentionfrom the study of conscious linguistic phenomena to their under-lying unconscious infrastructure; focuses on the relations betweenterms, rather than treating them as independent entities; introducesthe concept of a system of elements; and aims to discover generallaws by either induction or logical deduction. In so doing, Lvi-Strauss (1987: 15) applies Saussure's linguistic modeI to the study ofsocieties understood as complex symbolic orders: 'any culture maybe looked upon as an ensemble of symbolic systems, in the frontrank of which are to be found language, marriage laws, economicrelations, art, science and religin'. He thus seeks to reveal theunderlying structures and relationships of human thought andexperience that constitute social reality (Coward and Ellis 1977:14-15).

Lvi-Strauss's project is premised on two assumptions. The first isthat there are in principle 'deep' structures underlying the variouspractices in any society, which can be pinpointed and studied (Lvi-Strauss 1968: 87). Moreover, even though they lie below surfacephenomena, they are manifest in language, myths, systems of classifi-cation such as totemism, cooking, dress codes, manners and customs(De George and De George 1972: xxiii). Even more ambitiously, healso assumes that it is possible to discern a common underlyingstructure of relationships for all societies. This is because there exist'rules of transformation' tbat enable researchers to find correlationsand equivalences amongst seemingly disparate symbolic phenorn-ena, just as it is possible to loca te cornmon grammatical structures

]

~:Ji'jiti~

~l

r~~o;;~J."~ri'eKitn'ft . f" t" '@Lr' "@--~ffBrrt;'T~-e3f$~~:;j;.:,\

24Discourse

amongst different languages (Lvi-Strauss 1977: 18-19). In order toevaluate rus structuralist approach and to examine the emergence ofLvi-Strauss's particular conception of discourse, let us consider hisaccounts of totemism and myth respectively.

Totemism and myth

Lvi-Strauss's analyses of phenomena such as totemism and mythare of particular interest, because they enable us to examine twosubstantive aspects of his structuralist account of society, while alsocharting the evolution of his distinctive concept of discourse.Generally speaking, totemism refers primarily to a unique relation-ship between an animal or plant (the totem) and individuals andgroups living in a particular society. Totemism also signifies certainprohibitions governing the relationships between human beingsand their totems, as well as specic customs and practices organizedaround totems. In some cases, totemic beliefs suggest that certaintotems are the ancestors of specific groups of people (Sperber 1979:30).

One of Lvi-Strauss's chief aims in his analysis of totemism wasto dispel the 'totemic illusion', which he argued formed the con-ventiona wisdom of his time. This approach suggested thattotemism was specific to 'primitive peoples' who have no capacityfor abstract thought. Thus totemism served purely practical needs,or functioned as a source of religion, and represented no more thana stage of human development on the journey to civilization. Lvi-Strauss (1969: 71) shows this iIIusion to be an attempt by Westernanthropologists to establish a boundary between 'primitive' and'civilized' societies based on the idea that 'primitive' societies are'closer to nature' than more advanced Western societies.

Lvi-Strauss's alternative account centres on the universalhuman propensity to classify and organize the worId in a meaning-fui fashion. He examines the way in which different societies useand arrange totemic phenomena for particular purposes, and seeksto 'decode' their signifying power for people living in thesesocieties, as well as those who are studying them (Pool e 1969: 16).In contrast to those searching for an essential definition oftotemism or for a discrete phenomenon 'discovered' in nature(essentialism or realism), Lvi-Strauss (1969: 79) asks why and howhuman beings identify with animals or plants, and why these naturalphenomena are used to distinguish different groups within society.

'aussure. Structuralism and Symbolic Systems 25

l lc also wishes to ascertain what these constructions tell us about(he human mind and experience in general.

These two objectives require him to go below the surface ofpotentially misleading facts and ideological illusion so as to locatea rational and systematic set of relationships that order phenomenain particular ways. Instead of imposing a false totemic essence onsocial phenomena, and then seeking to explain them causally, Lvi-Strauss (1969: 84) begins by decomposing totemism into differentsystems of symbolic relations. He then argues that animals andvegetables are selected and used by primitive peoples not becausethey resemble human beings in any way, but because they consistof systems of-differential units, just as human societies consist ofdifferent kinship groups,

Totemic representations are thus the means of linking or corre-lating these two systems of difference: 'it is not the resemblances butthe differences, which resemble each other'. That is to say, on the onehand, 'there are animals which differ from each other', as they eachbelong to a distinct species and, on the other hand, 'there are men... who also differ from each other' in that they occupy differentpositions within society. Consequently, the 'resemblance presup-posed by so-called totemic representations is between these twosystems of differences' (Lvi-Strauss 1969: 149-50). Moreover,these totemic representations do not express or refiect pre-existingdifferences, but serve to constitute them. Primitive societies are nottrying to establish correlations between groups and animals, but usethe differences between animals to distinguish themselves. As Lvi-Strauss (1969: 161-2) suggests, 'natural species are chosen notbeca use they are "good to eat" but because they are "good tothink" '.

In this sense, totemic classification provides the means for humangroups to distinguish themselves from one another. Rather than aunique characteristic of primitive societies, Lvi-Strauss seestotemic representations as the function of a universal desire to clas-sify and organize the world - to establish differences between andwithin groups. Just as 'nation-states' adopt particular flags andanthems to differentiate themselves, so primitive peoples choosespecific plants and animals to mark them out as unique and singu-lar.

Lvi-Strauss's (1968: 208) structural explanation of myths - fan-tastic and repeated stories that provide sacred or religious accountsfor the origins of the natural, supernatural or cultural world -

26 Discourse

explores the basic similarity apparent in the great variety of myth-ical tales across the world's known societies. He argues that mythscannot be understood discretely, nor as they are told in the varioussocieties in which they occur. Instead, just as Saussure posits anunderlying system of langue beneath the contingent acts of speak-ing, Lvi-Strauss argues that myths have to be understood inrelation to the series of differences and oppositions that existbetween their constituent elements.

However, the analogy with Saussure's linguistic theory is notexact as Lvi-Strauss (1968: 209) introduces a third level of lan-guage to account for myths. Whereas parole and langue correspondto non-reversible and reversible time respectively, in which speechis the contingent articulation of words at a given time and languagethe ever present system that makes speech possible, myths consti-tute a more complex leve! of language that combine the propertiesof both parole and langue. This is because myths are not just toldat particular times and places, but also perform the universal func-tion of speaking to all people in all societies (Lvi-Strauss 1968:210).

Myths are therefore not to be confused with speech or language,as they belong to a more complex and higher order, and their basicelements cannot be phonemes, morphemes, graphemes orsememes. On the contrary, they have to be located at the 'sentencelevel', which Lvi-Strauss (1968: 211) calls 'gross constituent units'or 'mythemes', and he endeavours to explore the relations betweenthese elements. In his analysis of particular myths, these units areobtained by decomposing myths into their shortest possible sen-tences and identifying common mythemes, which are numberedaccordingly. This will show, as he puts it, that each unit consists ofa relation to other groups of mythemes.

However, so as to distinguish myths more properly from otheraspects of language, and to account for the fact that myths areboth synchronic (timeless) and diachronic (linear) phenomena,mythemes are not simple relations between elements, but rela-tions between 'bundles' of connected elements. After they havebeen differentiated and correlated together, Lvi-Strauss (1968:211-12) can then analyse myths at the diachronic and synchroniclevels. He can observe and establish equivalent re!ations thatoccur within a story at different points in the narrative, while alsobeing able to characterize the bundle s of relations themselves andthe relations between them. In this way, he can consider myths

27

~\l')~J

.usussure, Structura/ism and Symbolic Systems

hoth as surface narratives and as an underlying set of timelesslyrclated elements.

Having clarified these methodological issues, Lvi-Strauss's(1968: 230) overall objective is to explore the function of myths indifferent societies, and to establish their role in human thought andsocieties more generally. This means he must examine the differentsorts of relationships that exist between groups of connected rnythsso as to discern the 'rules of transformation' that regulate therelations between thern, In other words, he must discover the time-less and universal structures that govern the endless production andrnodification of myths in different societies and cultures, as theyendeavour to make sense of human existence.~i

~.1

~

Deconstructing structuralism

Structuralism has made an important contribution to our under-standing of language and social systems. Saussure's theory of lan-guage highlights the decisive role of meaning and signification instructuring human life more generally. More particularly, Lvi-Strauss's extension of Saussure's linguistic model to wider sets ofsocial relationships and practices means that society itself can beunderstood as a symbolic system. Rather than assuming society tobe the outcome of individual interactions, or the teleologicaldevelopment of the human spirit, or a product of the underlyinglaws of economic production, attention is focused on the changingset of signs and codes that make possible different social practices.In this vein, structural Marxists such as Louis Althusser (1969),tienne Balibar (1970), Nicos Poulantzas (1973) and ManuelCastells (1977) rethought the classical Marxist model of society, inwhich the economic base determined the ideological and politicalsuperstructure, by developing a conception of society as a relatedset of elements or instances. These ideas provide powerful concep-tual resources for exposing the weaknesses of essentialist, positivistand naturalistic accounts of society, while making possible an inno-vative method of conducting social and political analysis. They alsocreate the means for developing a distinctive concept and theory ofdiscourse.

Nevertheless, there are a number of problems with the classicalstructuralist model. By stressing the way social systems determinesocial meaning, it runs the risk of replacing the humanism of exist-ing approaches with a new form of essentialism based on the

..i1-~i

~I

~~..&1~.

f~

~i

't}.

~1

::i.t' . ",',.. ~.',. :~~!~~/~;

28 Discourse

primacy of a sta tic and complete structure. This assumption makesit difficult to provide an adequate account of the historicity of socialsystems, as well as the role of social agents in bringing such changeabout. To put it in different terms, Saussure and Lvi-Strauss'sintended revolutionary ambitions are often blunted by their ownarguments and assumptions. However, instead of throwing thebaby out with the bathwater, this calls for a deconstructive critiqueof structuralism that draws attention to its weaknesses with a viewto exploring possibilities that are closed off by the dominantargumentative logics, which organize the structuralist paradigm(Derrida 1981a: 6-7).

The limits of Saussure

Saussure's linguistic model challenges many of our commonsensicalbeliefs about language. He subverts our assumption that wordssimply denote objects in the world, or that they represent or expressour 'inner ' thoughts, or that there is a fixed connection between thewords we use and the ideas they convey. Instead, he argues that thelinguistic sign is arbitrary, that language is form and not substance,and that language comprises differences and not positive terms.The potential implications of Saussure's programme are a decisiveoverthrowing of the view that language is a transparent system ofsigns which we use unproblematically to eommunicate our ideas,and a startling subversion of our ability to be 'authors' of our ownintentions and desires.

However, these potentials are not fully realized. Four major criti-cisms can be levelled at Saussure's writings. He presents a homol-ogous relationship between the signifier and the signified, whiehtogether constitute the sign; he fixes the meaning of signs in a neces-sary fashion; he retains the notion of an autonomous subject of lan-guage; and he is unable to develop a viable conception of discourse.Let us consider these criticisms in more detail.

First, Saussure's (1974: 67) conception of the sign enforces a rigidseparation between signifier and signified, and proposes a one-to-one correspondenee between these two aspects of the signoInsteadof an intermingling between the 'material' signifier (sound-image]and the 'ideal' signified (concept), sometbing that should followlogically from Saussure's view that language is purely formal andrelational, he splits the two aspects into two distinet entities andthen recombines them in the sign itself. This separation implies that

~,s

!L ..J,i

saussure, Structuralism and Symbolic Systems 29

there can be a signified without a signifier, and vice versa. On theone hand, this opens the way for ideas to be located in the mind andthus to pre-exist language, in which case it is only when ideas have1.0 be expressed or comrounicated that they will have to be rep-rcsented in language by signifiers. On the other hand, such a viewsuggests that signifiers are only 'material' or 'sensible' entities thatare entirely devoid of conceptuality or ideality, in wheh case talkof the 'materiality of the signifier' results in a naive materialism(seeBennington 1993:26-31).

These paradoxes undermine Saussure's claims that language isform and not substance, and consists of differences without posi-tive terms. Toobeginwith, he splits the signifier and signified on thegrounds that one is substantial and the other conceptual, eventhough it follows from his own theory that if one tries to loeate apure signified one only finds other signifiers. Finding out themeaning of a word in a dictionary, for instance, can only be accom-plished by being able both to reeognize words ('signifiers') and tounderstand the meanings ('signifieds') of other words. Similarly, theonly way of distinguishing between different signiers is by recog-nizing differences that are themselves not material but ideal. Thedistinction between 'bat' and 'cat' is not just a case of a differeneebetween "b' and 'e'; it is also determined by what these words mean.In short, the logical conclusion of Saussure's own theoretical systemis to blur the very distinetion between signfier and signified to thepoint that there is no strong separation at al!.

In addition, Saussure privileges one sort of substance (speech or'sound') over another (writing or 'imagc') within the realm of thesignifier itself. Thus, not only does he seprate signifier and signi-fied, he inadvertently privileges the phonic substance (speeeh)oyer the graphic substance (writing). This introduces a furthermaterial difference within language and provides Derrida (1976;1981a) with an important entry-point for his deconstructivereading of Saussure. As Derrida points out, Saussure's privilegingof speech over writing is consonant with the priority attributed tothe voice and reason (logos) within Western metaphysieal thoughtsinee its inception. Writing is thus presented as secondary tospeecn, and both elements of the signifier are regarded as inferiorto the signified. As 1 shall argue in the next chapter, Derrida'sdeconstructive reading reverses these binary oppositions anddevelops a new eonception of writng (arche-writing), which doesnot privilege speech or coneepts. In sum, these difficultes point to

~'~;

~

~~!;{.;

30 Discourse

the ambiguous role of the sigo in Saussure's theory, and in Westernthinking more generally. The sign is there to represent ideas, whichin turn designate objects in the world. That is to say, while oneaspect of Saussure's writings challenges the dominant tradition ofthinking by emphasizing the fact that language is a self-enclosedsystem of differences and by attributing equal importance to thesignifier and signfied, he nevertheless privileges concepts and thehuman mind.

Second, Saussure fixes the meaning of signs in a necessaryfashion by arguing that the identity of the sign is a product of theoverall system of linguistic values. In other words, while Saussurestresses that elements in a language are relational and thus depen-dent on one another for their meaning, he presents the overalllinguistic system as closed and complete. This means that theunderlying systematicity of the linguistic system establishes themeaning of each term in a language (Benveniste 1971: 47-8). Thisresults in a new form of structuralist essentialism, in which thesystem of differential elements is regarded as a fully constitutedobject, rather than things, words or individuals. In short, Saussurefocuses on language as a product, rather than a process of produc-tion. He do es not account for the active construction and historicityof structures, and does not consider the possibility that structuresor systems may be ambiguous or contradictory.

Third, Saussure retains the idea of an autonomous subject of lan-guage standing outside the linguistic system His theory revolvesaround the central role he concedes to language users - human sub-jects - which pre-exist and are external to the linguistic system andhe presents the human speaker as the key agent or mechanismwhich links the sign to ideas and then finally to 'reality'. His writ-ings are thus replete with references to the 'hurnan mind' and the'psychological states of speakers'. For instance, in his discussion ofthe methods of synchronic analysis, he argues that '[sjynchrony hasonly one perspective, the speakers', and its whole method consistsof gathering evidence from speakers; to know to what extent a thingis a reality, it is necessary and sufficient to determine to what extentit exists in the minds of speakers' (Saussure 1974: 90). In this way,Saussure violates his desire to privilege form over substance, andcompromises the sharp boundary he wishes to draw betweenlangue and parole. According to Saussure, the formal and essentialsystem of language ought to be independent of any contingentspeech acts performed by individuallanguage users. This provides

Saussure, Structuralism and Symbolic Systems 31

hirn with his distinctive object of linguistic theory. However, thecentral role attributed to the human mind and language users blursthis fundamental division, and consequcntly many of the inferencesSaussure wishes to draw.

Lastly, Saussure stops short of developing an adequate concep-tion of discourse. While he argues that discourse comprises lin-guistic sequences greater than a single sentence, he does notprovide the tools for its analysis in structural terms. This is beca useSaussure (1974: 125) cIassifies sentences and systems oi sentencesas instances of parole and not langue, thus restricting his theory tothe false opposition between language as a total system of signsand speech as a product of the 'individual freedom' of each lan-guage usej. This theoretical decision means that the constructionof sentences and the relations between sentences are attributed tothe spontaneous creativity of individual speakers, thus fallingoutside the remit of a formal structural approach. Discoursecannot therefore be analysed as a regular system ofrelated and dif-ferential units, as Saussure banks upon an all-powerful conceptionof human subjectivity. Moreover, as discourse theory in the socialsciences is primarily concerned with the examination of changingand contested systems oi discourse, this makes it difficult to em-ploy Saussure's theory without modifying some of its key assump-tions.

The sum total of these conceptual difficulties results in anambiguous nheritance. Saussure's advances towards a relationaland non-essentialist theory of language, which can be analysedindependently of individual speech acts, run aground because heassumes that both the sign and the human subject/mind can beviewed as fully constituted entties or objects. This objectivism rein-forces the static picture of signification and meaning presaged inSaussure's prvileging of the synchronic over the diachronic dimen-sions of language. Language is thus seen as a total system of differ-ences which, albeit temporarily, is fixed; the idea of language andsgnification as an endless and indeterminate production of mean-ings is thus broached, but ultimately foreclosed. As more radicalcritics such as Derrida argue, this is the product of Saussure's inser-ton into the tradition of Western metaphysical thinking that priv-leges the role of human reason and thought over and above thecontingencies of language use and location. We must now considerthe implications of these aporas for a structural analyss of dis-course in the social sciences more generally.

. ~-_ "" ...; . ':'"" ;"';'.:......."" "." ~""-"-"'"'''''''''''''';'-:'':;;'~A''''

32 Discourse

The paradoxes of Lvi-Strauss

Lvi-Strauss's work presents us with further paradoxes, as heimposes rigid structural models on empirical phenornena, which heregards as intrinsically complex and historically contingent. Tobegin with, he extends the linguistic model to the social sciences bystressing the symbolic character of social interactions and struc-tures. However, he does not reduce social relations to a closedsystem of language, as he acknowledges their contradictory and his-torical character (Levi-Strauss 1987: 17-18). Instead, by focusing onthe way in which human subjects impose meaningful structures andcategories on their world of objects and relations, he providespowerful resources for criticizing naturalistic and positivistic con-ceptions of society. His fascinating interpretations of totemism andmyth thus provide vital insights for our understanding of contern-porary constructions of identity and difference, whether of anetbnic, nationalist or gendered formo

On a methodological level, Lvi-Strauss provides a sophisticatedset of tools for the analysis of social relations, and reflects criticallyon the problems arising from the study of social phenomena inthese terms. Finally, and crucially for the aims of this book, Lvi-Strauss is harbinger of a distinctive concept of discourse for thehuman and social sciences. The introduction of mythemes as theconstituent elements of myths differs from Saussure's distinctionbetween langue and paro/e, which is confined to the sub-sententialleve!. Along with purely formal accounts of language elaborated bywriters such as Louis Hjelmslev (1963) and Roman Jakobson(1990), this makes possible a distinctively structuralist analysis ofdiscourse, as Saussure's structuralist theory of language can beextended to the analysis of discourses at the wider symbolic level,Ths approach s taken further by post-structuralist thinkers such asDerrida and Foucault when developng their deconstructive andarchaeological accounts of discourse.

Nevertheless, just as Saussure retreats back into classical con-cepts and assumptions that undermine his first principIes, so Lvi-Strauss exhibits many of the closures of structural analysis. Porinstance, at one level, his theory of myth stress es the endless andcontingent construction of myths in society: 'There is no real endto mythological analysis, no hidden unity to be grasped once thebreaking-down process has been completed. Themes can be splitup ad infinitum' (Lvi-Strauss 1994: 5). This infinite construction ofmyths is replicated at the level of human subjects, for just as myths

noussure, Structuralism and Symbolic Systems 33

.~J

are not fixed by any complete and underlying system, so subjectscannot be certain of their own identities. Thus they are not unifiedentities standing outside the ever changing world of mythcal dis-course, which helps to form them, but are subject to the vicissitudesand changcs undergone by myths .

However, at other points in his text, Lvi-Strauss (1994: 10)offers the hope of a total and universal account of myth:

Throughout, my intention remains unchanged. Starting from ethno-graphic experience, 1 have always aimed at drawing up an inventory ofmental patterns, to reduce apparently arbitrary data to some kind oforder, and to attain a level at which a kind of necessity becomes appar-ent.

Moreover, he argues that myths are caused 'by the mind thatgenerates thern' and 'by an image of the world which is alreadyinherent in the structure of the mind' (Lvi-Strauss 1994: 341). Theidea of ahuman mind that is able to generate and comprehend anendless series of myths has important cOl1sequences for his con-ception of social structure. Instead of changing structures that con-tinuously transform meaning, they are split into two separatelevels. On the surface level, Lvi-Strauss (1994: 21) locates therichness and variety of empirical phenornena. Beneath the surface,however, he detects a sta tic and essential structure, which com-prises the oppositions and correlations between the basic elements

of the system.Moreover, these elements and subjects are analysed as fully

enclosed objects, rather than contradictory processes of construc-tion. This is in line with the formal structuralist method that Lvi-Strauss uses, and his tendency to reduce the truth and falsity ofphenomena to questions of method (Derrida 1978a: 284). Themethod itself is predicated on establishng relationships betweenpre-existing and fully formed objects. Thus his method consists ofthree basic operations:

(1) define the phenomena under study as a relation between two ormore terms, real or supposed;

(2) construct a table of possible permutations between these terms;(3)take this table as the general object of analysis which, at this level

only, can yield necessary connections, the empirical phenomenonconsidered at the beginning being only one possible combinationamong others, the complete system of which must be recon-structed beforehand.

\f

W!

jJ~-

.~

}

(Lvi-Strauss 1969: 84)