what is discourse and why analyse it_图文

1What is discourse and why

analyse it?In 1996 a spokesperson for British Telecom (BT), the UK’s largest phone company,launched a campaign to improve the nation’s communication skills, explaining that‘since life is in many ways a series of conversations, it makes sense to be as good

as we possibly can at something we tend to take for granted’ (quoted in the Guardian,

30 December 1996). Analysts of spoken discourse do not usually share BT’s goal ofmaking people ‘better’ at talking: they begin from the assumption that people are,with few exceptions, highly skilled users of spoken language. But most would probablyagree that ‘life is in many ways a series of conversations’, and that talking is ‘somethingwe tend to take for granted’. When linguists and other social scientists analyse spokendiscourse, their aim is to make explicit what normally gets taken for granted; it is alsoto show what talking accomplishes in people’s lives and in society at reference to ‘linguists and other social scientists’ in the last paragraph

is meant to underline the important point that working with spoken discourse is

an interdisciplinary enterprise: among those who may be engaged in it are anthro-pologists, linguists, philosophers, psychologists, sociologists, students of the mediaor education or the law. A commonly used academic label for what these variouspeople are doing, and one which I will use myself throughout this book, is discourseanalysis. But while it is useful in many contexts to have this generic label available, itdoes need to be remembered that ‘discourse analysis’ is an umbrella term, allowingfor considerable variation in subject matter and approach. For instance, I should makeclear straight away that discourse analysis is not exclusively concerned with spokendiscourse: in principle it can deal with socially situated language-use in any channelor medium. Discourse analysts may work with written data, or data from signlanguages of the deaf, and some analysts work with textual graphics and images aswell (e.g. Kress and van Leeuwen 1996).

There is also a distinction to be made between analysing discourse as an endin itself and analysing it as a means to some other end. Some discourse analysts –including many of those whose disciplinary affiliation is to linguistics – are primarilyconcerned to describe the complex structures and mechanisms of socially situatedlanguage-use. They ask questions like ‘how does turn-taking work in conversation?’,or ‘does the form of a question affect the form of the answer?’ or ‘why do peoplemisunderstand one another?’. They study talk because they want to know about some linguists, and many researchers who are not linguists, are more interestedin the idea that ‘life is in many ways a series of conversations’, which implies that

8WORKING WITH SPOKEN DISCOURSEpeople’s talk can be a source of evidence about other aspects of their lives. Thoughthey may not be studying discourse as an end in itself, many sociologists and socialpsychologists, or researchers in education, in cultural studies and media studies, adoptmethods which produce discourse data. Interviews, focus group discussions andethnographic studies using participant observation all involve verbal interactionbetween a researcher and research subjects, and/or between research subjectsthemselves. At least some of the analysis carried out by researchers who choose thesemethods will involve listening to talk, transcribing it, and reflecting on its meaningand significance.

As we will see in more detail later on, there are many different varieties ofdiscourse analysis, and there is a certain amount of argument about their relativemerits. My aim in this book is to be as inclusive as possible: though I write as a linguistmyself, I have tried to design my account of what discourse analysis is, and does, tobe helpful to readers across a range of academic disciplines. That means trying tocover the spectrum of approaches they are likely to encounter, making clear what thesimilarities and differences are, and bringing out the distinctive contributions madeby different the title Working with Spoken Discoursesuggests, this book will not beconcerned with allforms of discourse, but only with spokendiscourse. In theIntroduction I gave one reason for this choice: talk forms the data many socialresearchers turn to discourse analysis to help them interpret, and they are often unsurehow to approach this data in a suitably ‘analytic’ way. Arguably, this is less true

of written discourse. Anyone who has been educated in a highly literate society

will have developed, not only the ability to read and write, but also some ability tothink analytically aboutwritten texts. For instance, many school students have hadsome experience of learning how to do ‘close reading’ of literary texts: they have hadtheir attention drawn to the structure of a poem or to the existence of competinginterpretations of its meaning. By contrast it is much less likely that they have everbeen taught to approach ordinary talk – or any kind of spoken language – in the samesystematic way. Similarly, most people acquire in the course of their schooling

an extensive metalanguage (‘language about language’) with which to describe thestructures of writing: terms like letter, comma, sentenceand paragraphbelong to thismetalanguage. They rarely possess a parallel metalinguistic apparatus for discussingthe structures of spoken language: as we will see in more detail in Chapter 3, mostpeople do not realize the extent of the differences between writing and speech.

The above reference to metalanguage reminds us that issues of terminologyand definition tend to loom large in all academic enterprises; discourse analysis is noexception. I have already used several different terms – conversation, talkand [spoken]discourse– for what might appear to be much the same thing. Do all these terms,however, really mean the same thing? The question would be unlikely to arise in thecontext of an ordinary, non-academic exchange, but an important part of being‘analytical’ is being able to reflect on and ask questions about the conceptual frame-works and vocabulary we take for granted in everyday life. So at this point I want tolook more closely at some of the key terms that are relevant to discourse analysis,beginning with the apparently straightforward term conversation.

WHAT IS DISCOURSE?9CONVERSATION, TALK, DISCOURSEACTIVITY(Note: this activity is more interesting to do in a group, and it is especially interesting if thegroup includes speakers of more than one language.)As quickly as possible, list all the words you can think of that are used to describe differentkinds of talk in each of the languages/varieties you know. Now examine your list more closely.•How would you define each of the terms you have listed (for the benefit of someonelearning the language, for example)? Is each one distinct from all the rest or is thereoverlap? Is there any disagreement in the group about the definition of certain terms?

•What dimensions of contrast (e.g. formal v. informal, serious v. non-serious) seem to beimportant in distinguishing different kinds of talk?•If different languages/varieties are represented in your list, do they all make similardistinctions?•How many of the kinds of talk you have listed could you also describe as ‘conversation’?If that term applies better to some cases than others, why do you think that is? If it isinapplicable in some cases, what makes it inapplicable?In ordinary usage, conversationusually refers to spoken rather than written language.I say ‘usually’ because recently an interesting exception has become noticeable:

people who regularly interact with others via the internet, for instance in ‘chat

rooms’, sometimes refer to what they are doing as ‘talking’ or to their exchanges

as ‘conversations’, though the medium is actually written language. The term ‘chatroom’ makes an explicit parallel with a certain kind of informal conversation, namelychat. If we are being analytical, these usages might prompt some questions. Areinteractivity(the fact that on-line exchange can involve a relatively rapid ‘back

and forth’ between participants) and spontaneity(the fact that contributions to

chat-room exchanges are typically composed without much planning or editing)more salient characteristics of what we call ‘conversation’ than the channel or mediumof interaction? Is the actual language people produce in chat rooms more similar toface-to-face speech than other kinds of written language?

We (that is, English speakers, though the same thing is true for speakers ofmany other languages too) have quite a large vocabulary for distinguishing differentkinds of talk. We can describe interactions in terms of their tone, level of formalityand subject-matter using terms like argument, blether, chat, discussion, gossip. We candescribe spoken language events in terms of their setting, context or purpose usingterms like interview, debateand seminar. Is conversationjust a generic term thatsubsumes all the others, or does it cover only some of the possibilities? Is a seminara ‘conversation’? Is the talk I have with my doctor when I visit her surgery ‘conver-sation’ in the same way the talk I have with her if I run into her at the supermarket is

10WORKING WITH SPOKEN DISCOURSE‘conversation’? The activity on p. 9 is intended to encourage you to think about yourown understanding of what ‘conversation’ is, and more generally what different kindsof talk are recognized by language-users in the community or communities youbelong point the activity might illustrate is that conversationin English has both‘generic’ and more ‘specific’ uses. It is generic in the sense that we can use it to describea relatively broad range of different kinds of spoken interaction. More specifically,though, it seems most ‘natural’ to apply it to interaction which is characterized byinformality, spontaneity and egalitarian relationships between the participants

(if your boss asks you to come and have a ‘conversation’ about your punctuality, youtend to suspect euphemism, or irony). Certainly, for me as an English-speaker it seemsmore natural to use the word conversationin connection with ‘chat’ or ‘gossip’ thanfor a seminar or a medical consultation. Each of these has features of conversation,but intuitively I feel it is not the prototypical case.

In this book I want to consider many kinds of spoken interaction, and toexamine the similarities and differences among them. Therefore, when I discussspoken interaction in a generic sense I will use two other words in preference toconversation: one is talkand the other is discourse. Just to complicate matters, one ofthese terms is more generic, or at least more inclusive, than the other. Talk, to statethe obvious, refers only to spoken language-use, whereas discourse, as I have notedalready, can refer to language-use in any channel or medium. But that is not the onlydifference between the two terms. Discourseis evidently a more ‘technical’ term thantalk. And like a lot of technical terms, it is also ‘contested’ – that is, it has generateda lot of debate among scholars about what it means and how it should be used. Infact, the term discourseis notorious for the arguments surrounding it and theconfusion it can cause. A major source of potential confusion is that the meaning ofthe term tends to vary quite significantly depending on the academic discipline andthe theoretical preferences of the person who uses it. The range of meanings discoursecan have in academic discussion is an issue that needs to be clarified sooner ratherthan later. So before I go any further: what is discourse?LANGUAGE ‘ABOVE THE SENTENCE’ AND LANGUAGE‘IN USE’The most straightforward definition of discourseis the one often found in textbooksfor students of linguistics: ‘language above the sentence’. Of course, that is not at allstraightforward unless you understand some basic assumptions in linguistics, so letme spell them sts treat language as a ‘system of systems’, with each system having

its own characteristic forms of structure or organization. For instance, the soundsystem of a language (its phonology) does not have the same kinds of units, or thesame rules for combining them, as the grammatical system of that language. As yourunits get larger (e.g. words are larger than sounds and sentences are larger thanwords), you metaphorically move ‘up’ from one level of organization to the next.

WHAT IS DISCOURSE?11If discourse analysis deals with ‘language abovethe sentence’, this means it looks forpatterns (structure, organization) in units which are larger, more extended, than of the earliest discourse analysts, the linguist Zellig Harris (1952),

posed the question: how do we tell whether a sequence of sentences is a text1– thatis, the sentences relate to one another and collectively form some larger whole –

as opposed to just a random collection of unrelated bits? The answer to that

question, Harris thought, would make clear what kind of structure exists ‘above thesentence’. Texts would have this structure, whereas random collections of sentenceswould y, language-users do routinely interpret sequences longer than a sentenceas texts in which the parts combine to form a larger whole. Consider the followingexample, a simple text produced by a child, which is discussed in a famous article bythe conversation analyst Harvey Sacks (1972).The baby mommy picked it obvious instance of ‘structure above the sentence’ in this example is

the pronoun it, which is anaphoric(referring back). It comes in the second sentencebut it refers to something mentioned in the first: ‘the baby’. A reader or hearerautomatically takes it that the ‘it’ which the mommy picked up must be the baby, andnot some previously unmentioned object like a rattle or a banknote. The pronoun

is a cohesivedevice, tying the two sentences together, and cohesion is a property oftexts.

But there is more to say about what makes this sequence work as a text.

For instance, it is natural to read it as a narrative, in which the sequence of events inthe text mirrors the sequence of events in the reality being reported: the baby criedand thenthe mommy picked it up. Indeed, in this case we are likely to infer not merelysequence but causality: the mommy picked up the baby becauseit cried. The way weprocess the text as a narrative implies that we are following a general procedure fordealing with structure ‘above the sentence’: where A and B are sentences, we assumethat A followed by B means ‘A and thenB’ or ‘A and consequentlyB’.

But there is a problem with Zellig Harris’s proposal about distinguishing textslike the child’s story from random collections of sentences which are not texts: weseem to have a strong tendency to apply the principles just described to any sequencewe are confronted with, however bizarre. Michael Stubbs (1983: 93) quotes a radioannouncer who once said:

Later, an item about vasectomy and the results of the do-it-yourself

competition.

12WORKING WITH SPOKEN DISCOURSEWhy does this raise a smile? The default assumption is that the parts of the announce-ment on either side of the conjunction andrelate to one another in the same way asthe two parts of the baby/mommy sequence, and so we reason that the announcermust be referring to a do-it-yourself vasectomy competition. However, I am confidentthat most of us immediately go on to reject this interpretation. We recognize that

the announcer did not intend the structural relationship that is implicit in theorganization of his discourse. But this recognition does not come from ponderingthe details of the announcer’s language, and deciding that, contrary to our initialassumption, his utterance has no structure. Rather, we realize that the scenario wehave conjured up by applying the usual procedure is implausible and ridiculous: noone would organize, or enter, a competition in which men performed vasectomieson themselves. In other words, we take account not simply of the linguistic propertiesof the announcement, but also (and in this case more significantly) of what we knowabout the world.

Real-world knowledge is also relevant to the interpretation of the baby/mommy text. There is no purely structural reason why we have to take it that

the mommy who picked the baby up is the baby’s own mother, since the child doesnot specify that by using a possessive pronoun, referring only to ‘themommy’.Nevertheless I would bet that most readers did make that assumption. The text followsa familiar script whereby babies cry and are picked up by their mothers to stop themcrying. It is imaginable that a crying baby might be picked up by a total stranger whowas, however, the mother of some other baby; but that would not be many people’sfirst guess.

I am suggesting, then, that we make sense of discourse partly by makingguesses based on knowledge about the world. If that is accepted, then arguably thedefinition of discourse as ‘language above the sentence’, and of discourse analysis asa search for structure at a level higher than sentence structure, is not adequate. Thatdefinition suggests that single sentences and texts have a similar kind of organization:the difference is one of scale. But is that really the case? To be grammatical, a sentencemust contain certain constituents in a certain order: it is conformity to structuralrules that makes the difference between grammatical sentences and ‘word salad’ (like*stood boy the on up chair a– the asterisk is a linguist’s convention for denotingungrammatical sequences). But our ability to decide whether and how discoursemakes sense appears to involve much more than quasi-grammatical generalizationsabout what can go with or follow what.

It might also be asked whether the characteristic features picked out bydiscourse analysts have much to do with the size of the units being analysed – the factthat they are larger than a sentence.2Henry Widdowson (1995) has pointed out thata ‘text’ can in fact be smaller than a sentence. He observes for instance that the legendLADIES on the door of a public lavatory is a text, as is the letter P which is used inBritain to indicate a space for parking cars. A single word or letter cannot have‘structure above the sentence’. So what makes these examples texts? Widdowson’sanswer is that in the contexts he is concerned with, each of them is intended to conveya complete message. Of course, what we take that message to be does depend on thecontext, and once again, its interpretation relies on real-world knowledge that is not

WHAT IS DISCOURSE?13contained in the text itself. Looking up the word ladiesin a dictionary would not,

on its own, make clear what message it conveys when written on a door. (Someonewho spoke English but was unfamiliar with the concept and etiquette of publiclavatories might think it meant ‘there are ladies behind this door’.) A great deal ofgeneral knowledge and contextual information has to be brought to bear on even themost banal texts we encounter if those texts are to serve their communicative purpose.A distinctive feature of discourse analysis, as opposed to the study of syntax (sentencestructure), is its overt concern with what and how language communicates when itis used purposefully in particular instances and contexts, and how the phenomenawe find in ‘real language’ (implicitly contrasted to the idealized, made-up examplesentences most often discussed by analysts of syntax) can be explained with referenceto the communicative purposes of the text or the interaction. From this standpointa better definition of discoursethan ‘language above the sentence’ might be ‘languagein use’: language used to do something and mean something, language produced andinterpreted in a real-world context.

Deborah Schiffrin (1994) suggests that the two definitions of discourseI havejust outlined correspond, roughly speaking, to two important currents or tendenciesin twentieth-century linguistics. One is formalismor structuralism: an interest in theabstract form and structure of language. The other is functionalism: an interest

in what language is used to do. But Schiffrin goes on to point out that treating this

as an absolute distinction would be an oversimplification. Because its meaning

is so dependent on context, discourse is not amenable to ‘pure’ formalist sely, functionalists have always been concerned with form as well as are interested in how the two are connected, suggesting that language has acertain kind of formal organization because of the purposes it is designed to serve.

Most discourse analysts who locate themselves within the academic disciplineof linguistics are concerned with both form and function, though the balance betweenthese concerns may vary. But not all discourse analysts are linguists, and not all woulddefine their goals in terms of improving our understanding of language as such. Manysocial scientists (including, in fact, some linguists) are more interested in discourseas a source of evidence or insight about social life and social relations. Their questionsare not like Zellig Harris’s, primarily about the way language works. Rather they usediscourse analysis as a qualitative research method for investigating social phenom-ena: sexual harassment, attitudes to the monarchy and youth subcultures are amongthe topics that have been investigated in this way.3But investigators doing this kindof work often adopt a definition of the term discoursewhich differs from the ones wehave examined so far. That alternative definition is now sufficiently influential acrossdisciplines to merit more detailed consideration.

POWER, KNOWLEDGE AND PRACTICE: DISCOURSE(S)AND THE CONSTRUCTION OF SOCIAL REALITYI mentioned above that discourse analysis is an increasingly popular qualitativeresearch method in social science. The word qualitativein this context contrasts with

14WORKING WITH SPOKEN DISCOURSEquantitative. Discourse analysis is an alternative to using standardized instrumentslike questionnaires, which yield statistical data. A researcher who wants to find outusing quantitative methods what people do in their leisure time might ask a sampleof subjects to fill in the same questionnaire and then produce a statistical digest oftheir responses (e.g. ‘50% of women under 35 reported shopping was their mainleisure activity’). A researcher who decides to use discourse analysis as a methodwould be more likely to spend time talking in depth to a sample of the people s/he isinterested in, encouraging them to explore the subject in their own way and in theirown words. The researcher would record subjects’ talk, transcribe it and analyse it –not in order to make statistical generalizations, but in order to point out recurringthemes in the way people talk about leisure activities.

This method is sometimes criticized on the grounds that it will not tell usaccurately what people reallydo in their leisure time. The criticism is not withoutvalidity, but a discourse analyst might point out that it could equally be made of thequestionnaire/statistical analysis approach. When people answer a researcher’s ques-tions, whether in a face-to-face interview or by completing a written form, they areconstructing a certain representation of themselves for the researcher’s benefit: theymay be telling the researcher what they think s/he wants to hear or what they wouldlike her/him to believe. This is another version of the process of self-construction thatgoes on in ordinary talk, which is always produced with an eye to the situation andthe person(s) to whom it is addressed. Arguably it is an unavoidable element of allcommunicative acts: people simply do not answer questions, in any situation, withoutfirst making some assessment of who is asking and why. (As a simple illustration ofthis point, recall the last time a doctor or nurse asked you how many units of alcoholyou consume in a week. Was your answer affected by your assumptions about whatuse the nurse or doctor planned to make of the information? Was your answeraccurate and truthful? Would you have given just the same answer to your mother,your best friend, or a prospective employer? What is your attitude to the questions

I am asking now – for instance, if you don’t drink alcohol, are you offended by myapparent assumption that all my readers do? Would that affect your response to me,if this were a conversation?)Researchers who favour discourse analysis over supposedly more ‘objective’methods argue that paying attention, not merely to whatpeople say but to howtheysay it, gives additional insight into the way people understand things. It is less aboutcollecting facts than about studying interpretive processes. Such researchers may alsoargue that analysing ‘real’ talk does a better job than standardized instruments

of capturing the messiness of real life, and to that extent could be seen as more ratherthan less ‘accurate’. Giving people a multiple-choice questionnaire obliges them tochoose one option from a set constructed by someone else: they check box A, andthat makes them look as if they are committed to A while rejecting B and C. Yet whenpeople talk it often becomes clear that matters are more complex than that: they don’tdismiss B and C out of hand, and they have their doubts about A. Standardizedinstruments produce an impression of certainty and consistency which is, arguably,misleading. Another advantage that might be claimed for discourse analysis is that itgenerates data by getting people to engage, or observing them while they engage, in

WHAT IS DISCOURSE?15an activity – talking – which is normal and familiar to them, rather than asking themto undertake an unusual or artificial task. Life may or may not be ‘in many ways aseries of conversations’, but it is in no way a series of box-checking exercises.

Social researchers who do discourse analysis often want to make the point thateven when we talk ‘in our own words’, these words may not actually be ‘ours’ at all,in the sense that they are not original or unique to any one individual. As one analyst,Jay Lemke, has put this point:We speak with the voices of our communities, and to the extent that we have individualvoices, we fashion them out of the social voices already available to us, appropriatingthe words of others to speak a word of our own. (1995: 24–5)Within any community there is a finite range of things it is conventional or intelligibleto say about any given concern. When people talk about shopping, or drugs, or theroyal family, what they say will be drawn from the community’s repertoire of thingsit is possible to say rather than representing some unique perspective on the is not to suggest that people never say anything novel or unexpected, or that theydo not have ideas of their own. But language-using is an intersubjectiverather thanpurely subjective process: a ‘voice’ that is wholly individual runs the risk of beingincomprehensible. Hence Lemke’s point that individuals’ ways of talking are formedusing resources that are shared with others in their communities. Discourse analysiscan be seen as a method for investigating the ‘social voices’ available to the peoplewhose talk analysts social researchers today would argue that people’s understandings

of the world are not merely expressed in their discourse but actually shaped by theways of using language which people have available to them. Another way of puttingthis is to say that reality is ‘discursively constructed’, made and remade as people talkabout things using the ‘discourses’ they have access to. Evidently, the word discoursein this formulation is not being used in the way linguists typically use it, to mean‘language above the sentence’ or ‘language in use’. An obvious difference is that thelinguist’s discoursehas no plural, whereas social theorists often talk about plural usage reflects the influence of the philosopher and cultural historianMichel Foucault, who defined discoursesas ‘practices which systematically form theobjects of which they speak’ (Foucault 1972: 49).

To see what Foucault meant, let us consider the case of ‘drugs’. The word drugsmight seem to name a pre-existing ‘real-world’ category (of substances that affect

the mind and body in certain ways). But if we think about the way the word is mostoften used, it becomes evident that it does not simply denote allsubstances that havecertain effects: depending on context, it denotes either those which are medicinal,

or those which are used non-medicinally and are also illegal. Caffeine, nicotine andalcohol are clearly mood-altering substances, but if we hear a report on ‘drugs’ we donot immediately think of coffee, cigarettes and beer. We certainly do not think ofcoffee-bean importers as drug traffickers or of tobacconists as drug dealers. What thissuggests is that our category of ‘drugs’ (in the non-medicinal sense) has been formedthrough a particular set of practices: legislation (making some substances illegal),

16WORKING WITH SPOKEN DISCOURSEpolicing (trying to prevent breaches of the law and to catch people who do break it),the practices of the courts (where stories are told about why people have broken thelaw and decisions are made about how to deal with them), of social and charitableagencies (which try to reduce the harm caused by drugs), schools (which practise‘drugs education’) and the media (which report on ‘the problem of drugs’). Buying,selling and using illegal substances are also practices relevant to the understanding of‘drugs’ as a category, though fewer people are involved in these practices comparedwith the numbers exposed to education or media so many practices and agencies involved, not surprisingly there aremultiple ‘discourses’ on drugs. We may be working with the same category, but

we can discuss it in different ways. For instance, there is a ‘law and order’ discoursein which drug-use is discussed as a crime, committed by people who are ‘bad’. Analternative discourse is ‘medical’: people who use drugs are sick, and need treatmentrather than punishment. There is a ‘social’ discourse in which drug-taking arises fromdeprivation and hopelessness. In contrast to these negative ways of talking, there isalso a discourse in which drug-using is defined as a recreational activity, enjoyedwithout ill effects by the majority of those who engage in it. Another positive discoursesuggests that using drugs may help people attain greater spiritual awareness.

Each of these ways of talking about ‘drugs’ has a history, but in some practices(and many discussions) they are not kept distinct. Drugs education, for instance,typically aims to persuade young people that they should not use drugs (it is unhealthy,illegal and dangerous), but some programmes also discuss the idea that drug-use ispleasurable, on the grounds that the appeal of drugs must be acknowledged if youngpeople are to take warnings about the dangers seriously. Some programmes assumethat many or most people will experiment with drugs, and aim to teach them how tominimize the risks involved. So drugs education may mix, in various proportions,elements of the ‘law and order’, ‘medical’ and ‘recreational’ discourses. Together, thevarious ways of discussing drugs and the practices that go along with them form

a network of concepts and beliefs that set the agenda for debate and define what weperceive as reality on this subject. This is what theorists mean when they say thatreality is ‘discursively constructed’.

It might be asked what the sense of the word discoursethat I have beendiscussing has to do, specifically, with language. Recall Foucault’s definition, quotedabove: although he calls discourses ‘practices’, he goes on to say that they ‘form theobjects of which they speak’. The link between practice and speaking (or moregenerally, language-use) lies in Foucault’s concept of ‘power/knowledge’. In themodern age, Foucault points out, a great deal of power and social control is exercisednot by brute physical force or even by economic coercion, but by the activities of‘experts’ who are licensed to define, describe and classify things and people. Definition,description and classification are practices, but they are essentially practices carriedout using language. Words can be powerful: the institutional authority to categorizepeople is frequently inseparable from the authority to do things to them. Thus forinstance, experts define mental health and mental illness, and on the basis of theirdefinitions, individuals can be classified as mentally ill and detained in psychiatricinstitutions. Experts produce definitions of good or adequate ‘parenting’, and parents

WHAT IS DISCOURSE?17who do not meet the minimum standard may have their children taken away fromthem. Experts elaborate a concept of ‘intelligence’ and devise ways of measuring it(such as IQ tests); this may have real-world consequences for individuals’ educationand employment prospects.I have explained the sense of discoursethat comes from the work of Foucaultbecause this usage of the term is now quite common, and students of discourseanalysis in a variety of disciplines are likely to encounter it. However, it should notbe supposed that all social researchers who adopt discourse analysis as a method arecommitted to the ideas of Foucault, or those of any other theorist. Some sociologistsand social psychologists use discourserather as some linguists do, to mean ‘languagein use’. There are varying views on whether and to what extent social reality is‘discursively constructed’: you do not have to believe in the discursive constructionof reality to regard what people say as a source of insight aboutreality.

As I pointed out earlier, though, any researcher who sets out to investigatesome aspect of reality by studying discourse will end up with data in the form oflanguage. And it is easy to underestimate the complexity of those data. As practisedusers of at least one language, researchers may be tempted to assume that it requiresno special expertise to interpret linguistic data – that this is simply an extension

of our ordinary, everyday behaviour as participants in verbal interaction. But that

is at best only partly true. Being able to do something yourself is not the same as beingable to analyse it from the outside. Discourse is not pure content, not just a windowon someone’s mental or social world; it has to be considered as discourse, that is, as aform of language with certain characteristics which are dictated by the way languageand communication work. It is not only linguists who can benefit by paying attentionto the ‘how’ as well as the ‘what’, the form as well as the content of people’s sely, linguists have something to gain by attending to other social scientists’insights into what discourse does, or what social actors do with it.

SUMMARYThis chapter has been concerned with the meaning of the term discourseand the goalsor purposes of analysing it. The view of discourse analysis taken here and throughoutthis book is a ‘holistic’ one, which acknowledges that discourse analysis is severalthings at once. It is a method for doing social research; it is a body of empiricalknowledge about how talk and text are organized; it is the home of various theoriesabout the nature and workings of human communication, and also of theories aboutthe construction and reproduction of social reality. It is both about language andabout life.

Part I continues, though, not with these grand abstractions, but with someconcrete, practical considerations. The first requirement for any kind of discourseanalysis is a body of data to analyse. In the next chapter we will look at the optionsand problems involved in collecting spoken discourse data.

18WORKING WITH SPOKEN DISCOURSESUGGESTIONS FOR FURTHER READING ABOUT

‘DISCOURSE’A book whose purpose is to ‘unpack’ the complex term discourseis Sara Mills’sDiscourse(1997). A shorter survey of various tendencies in contemporary discourseanalysis is provided by the editors’ Introduction to The Discourse Reader(Jaworskiand Coupland 1999). This volume also includes an edited extract from MichelFoucault’s The History of Sexuality, ‘The incitement to discourse’, which givessomething of the flavour of what Foucault and his followers mean by the term.

For a more traditionally ‘linguistic’ perspective on discourse and discourse analysis,a good source is the second chapter of Deborah Schiffrin’s textbook Approaches toDiscourse(1994), which is titled ‘Definitions of discourse’. Teun van Dijk’s editedtwo-volume Discourse Studies: A Multidisciplinary Introduction(1997) is a usefulreference source for both linguists and others.

NOTES1In the next few paragraphs the word textoccurs several times, and readers may findthemselves wondering if it is just a synonym for discourse. In fact that is a disputed writers use the two terms more or less interchangeably (this is how I am using themin this section); some refer to spoken discoursebut written the difference is

one of medium); others (e.g. Widdowson 1995) make a more theoretical fly, Widdowson argues that textis the linguistic object (e.g. the words on a page in abook, or the transcript of a conversation) whereas discourseis the process of interaction/interpretation that produces meaning from language. In speech discourse comes first, andproduces a text; in writing text comes first, and readers produce discourse from it.

As will be discussed in more detail below (Chapter 3), the sentence is in any case essentiallya unit of writtenrather than spoken language.

For readers who want to follow up any of these examples, Kitzinger and Thomas (1995)is a discourse analytic study of sexual harassment; Billig (1992) is a study of discourse aboutthe (British) monarchy; and Widdicombe and Wooffitt (1995) is a study of youthsubcultures using the method of Conversation Analysis.

23