To contents page

To Psychiatric Activism Page

Chapter 5

Beautiful and inexorable systems:

The discourse of discourse analysis

So much depends


a red wheel


glazed with rain


beside the white


- William Carlos Williams, Red wheel barrow

Jonathan Swift tells of a country where the inhabitants for the sake of unequivocal communication resolve to use objects rather than words, each object corresponding to a particular concept. The problem is that intellectuals soon find themselves burdened down with the weight of their ideas while their less intellectual (but more brawny) rivals are able to support arguments of considerable complexity. God, as we see around us every day, is not subject to the same constraints in terms of either brains or brawn, and has consequently allowed his or her vocabulary to grow to universal proportions.

There is however a certain moral ambiguity in coming to understand the everyday solidities of our existence as mere hieroglyphs in a more profound system of discourse; an ambiguity which is perhaps present in one form or another in all attempts at turning things into talk or talk into things. On the one hand the mundane is imbued with meaning - God speaking in the exact juxtaposition of wheel barrow, rain water and (white) chickens; on the other that which was substantial, immediate and particular is devalued - a mere token which derives currency from its place in an abstract system, but in itself is worthless.

However, the disconnection between language and reality which is the basis of Swift's satire is, according to Benjamin Whorf (1956)(1), by no means ubiquitous. He claims that "the idea, entirely unfamiliar to the modern world, that nature and language are inwardly akin, was for ages well known to various high cultures whose historical continuity on the earth has been enormously longer than that of Western European culture" (p. 249).

Whorf, who is unjustly remembered for helping to formulate the discredited Sapir-Whorf hypothesis (that, crudely put, individuals in certain cultures are unable to think of certain concepts because their language does not encompass these), was given to seeing the "skull beneath the skin"(2) of human discourse. He speaks of "the PREMONITION IN LANGUAGE of the unknown, vaster world - that world of which the physical is but a surface or skin, and yet which we ARE IN, and BELONG TO" (emphasis in original, p. 248). Whorf describes language as follows:

It is as if, looking at a wall covered with fine tracery of lacelike design, we found that this tracery served as the ground for a bolder pattern, yet still delicate, of tiny flowers, and that upon becoming aware of this floral expanse we saw that multitudes of gaps in it made another pattern like scrollwork, and that groups of scrolls made letters, the letters if followed in proper sequence made words, the words were aligned in columns which listed and classified entities, and so on in continual cross-patterning until we found this wall to be - a great book of wisdom! (p. 248)

It is said that William Durant, the founder of the General Motors empire, cribbed the well-known Chevrolet symbol from the wallpaper of a motel room. It is debatable who experienced the more intense 'epiphany upon gazing at wallpaper' (Whorf sitting at his desk, or Durant reclining on his motel room bed), but we do know that Durant, at least, realised his epiphany in the form of the Chevrolet motorcar while Whorf never got beyond talk.

The relation between Whorf's sublime speculation and Durant's functional machine is the same as that between the exquisitely wrought but apparently ineffectual 'talking cure' of psychoanalysis and the crude but apparently efficacious psychopharmacology which has now largely replaced it. However, just as psychoanalysts' talk about talk must ultimately emanate from mindless electro-chemical activity taking place in their brains, brain biologists' talk about real things must inevitably be constrained by the 'prison house of language' within which they are forced to conduct their investigations.

After demonstrating that there are definite rules for generating English-like syllables, Whorf observes:

It is as if the personal mind, which selects words but is largely oblivious to pattern, were in the grip of a higher, far more intellectual mind which has very little notion of houses and beds and soup kettles, but can systematize and mathematize on a scale and scope that no mathematician of the schools ever approached ... And now appears a great fact of human brotherhood - that human beings are all alike in this respect. So far as we can judge from the systematics of language, the higher mind or "unconscious" of a Papuan headhunter can mathematize quite as well as that of Einstein; and conversely, scientist and yokel, scholar and tribesman, all use their personal consciousness in the same dim-witted sort of way, and get into similar kinds of logical impasse. They are as unaware of the beautiful and inexorable systems that control them as a cowherd is of cosmic rays (p. 257).

To decode the 'beautiful and inexorable systems' of language, to achieve communion with the 'higher, far more intellectual mind' which steers our thinking, and which in the end is 'inwardly akin' to the physical realities of our being - this is the vision of mystics such as Whorf. Others, while sharing his conviction that we are caught in a relentless but invisible linguistic grip, are less convinced that when the code of language is finally broken we shall find it to be a 'great book of wisdom'. Rather than a font of wisdom, language is suspect - at best a vulgar plagiarism of the Durant variety, at worst a methodical conspiracy to naturalise and legitimate particular relations of power.

The critique of the new biological psychiatry presented in Chapter 4 belongs to this latter category, attempting as it does to show that there is something shady about the way biopsychiatrists use and are used by their discourse. In what follows I present an overview of the work of some of those who have articulated "the belief that underneath what is said and done in modern Western states, there is something disreputable waiting to be unmasked" (Minogue, 1989, p. 139). Starting with a brief recapitulation of the role of language in critiques of psychiatry, I successively broaden the focus to include medicine in general, linguistics, post-structuralism and discourse analysis in social psychology.

'Stammered, imperfect words without fixed syntax': The language of psychiatry

As already suggested in the preface, the idea of madness has always been closely linked to that of linguistic disorganisation (cf., Berenbaum, 1992; Gilman, 1983, 1988; Lidz, 1968). Biological psychiatrists may ascribe the bizarre speech of schizophrenia to brain dysfunction while antipsychiatrists may frame it as the communication of unacceptable ideas in an unusual idiom, but all are agreed that there is something special and different about the way psychiatric patients talk, a strangeness which has been duly inscribed in the DSM diagnostic criteria for schizophrenia. Yet, as Kleinman (1988) observes, "the entire cultural apparatus of language, symbols, and interpretations is a source of great ambivalence for the contemporary psychiatric researcher." (p. xi) In part this is due to modern biopsychiatry's general distaste for philosophical and ideological discourse, but in part it may stem from an awareness that not only the mentally ill, but also those who attempt to heal them, can become linguistically entangled.

The danger occurs when researchers turn away from patients, and start recording the healers' talk. Just as patients construct delusional systems, so "the silent master builder, psychiatry ... constructs the house of language, metaphor, and culture in which the drama of parents, spouses, friends, and other social control agents coping with emotionally troubled individuals takes place" (Light, 1982, p. 33). Light describes how

Residents learn to characterize the whole patient by his or her diagnosis, so that the patient does not have paranoid schizophrenia but is a paranoid schizophrenic. This is a fundamental change from medical diagnosis and the rapidity with which residents incorporate this perspective is startling (p. 40).

It could be argued that other physicians do much the same thing, as in surgeons talking among each other of "the ruptured spleen in bed 103", but it does appear as if in psychiatry the linguistic identification of the patient with her diagnosis is more frequent and pervasive.

Virtually every aspect of psychiatry is susceptible to this kind of switch of research interest from patient to doctor. Rather than the aberrant speech of mentally ill women, one may investigate the way psychiatrists talk about women, discovering for instance that despite the fact that the majority of psychiatric patients in almost all settings are women, they are almost invariably referred to in the abstract as 'he' (Allen, 1986). Similarly in researching the history of psychiatry one may change from a literal marshalling of the facts of the Zilboorg and Henry (1941) variety, to a realisation that:

The gauze of language, woven on a loom of convention by people whose concerns were different from ours, inevitably distorts our vision of past reality. To understand anything at all about the history of madness, we must examine first the patterns formed in the records themselves and the people and institutions that created them ... In other words, historians of insanity do not in the first instance study the insane at all: they study observations of the insane (MacDonald, 1987, p. 209-210).

Perhaps the most far-reaching and subtle critique of the strangeness of psychiatric language was that invented by Foucault (1967) which served as inspiration for the historical account of readmission given in Chapters 2 and 3. Paired with Foucault's conception of the need for a 'Great Confinement' which arose with the Enlightenment, is that of a radical break between the languages of madness and reason. Since its inception, mad-doctoring has been concerned with talking about rather than to mad men and women; and rather than champions of humane treatment of the mentally ill, figures such as Pinel and Tuke further entrenched this tradition:

As for a common language, there is no such thing; or rather, there is no such thing any longer; the constitution of madness as a mental illness, at the end of the eighteenth century, afford the evidence of a broken dialogue, posits the separation as already effected, and thrusts into oblivion all those stammered, imperfect words without fixed syntax in which the exchange between madness and reason was made. The language of psychiatry, which is a monologue of reason about madness, has been established only on the basis of such a silence. I have not tried to write the history of that language, but rather the archaeology of that silence (Foucault, 1967, p. xii-xiii).

Later Foucault (1980) would say of this enormously influential 'archaeology of silence', and of his subsequent works:

I am well aware that I have never written anything but fictions. I do not mean to say, however, that truth is therefore absent ... One 'fictions' history on the basis of a political reality that makes it true, one 'fictions' a politics not yet in existence on the basis of a historical truth (p. 193).

Gordon (1980) explains what appears from the viewpoint of conventional 'histories of the past' as a lack of concern with historical veracity thus:

We can say that the object of Foucault's critique is the status of the present. If Foucault poses a philosophical challenge to history, it is not to question the reality of 'the past' but to interrogate the rationality of 'the present' (p. 242).

Whether one is willing to accept this kind of justification or not, the danger for psychiatry once it declares itself willing to enter the domain of language is that it may be overcome by the onslaught of the likes of Foucault who writes "faster than we can read him" (Minogue, 1989, p. 138)(3).

'To penetrate the veil while retaining its hallucinatory quality': The language of Medicine

Although perhaps a special case, psychiatry is hardly the only social or professional institution or scientific discipline to be made the target of linguistic criticism. In this section the critique of psychiatry is contextualised within a broader critique of medicine, while in the following sections the context is extended even further to encompass critical language studies in general.

As in psychiatry, any study of the role of language in medicine is likely, at least in the first place, to focus on the linguistic deficiencies of the ill person, rather than those of the physician. In an early study of this sort Redlich (1945) asked 25 patients to define 60 medical terms(4). His rather predictable findings were that:

Two-thirds of the 25 patients knew too little about medical matters, their illnesses, and the implications of their illnesses. A small group possibly knew 'too much', but their knowledge was rather erratic, poorly integrated, and often quite irrational. Both groups might be helped considerably by sensible information (p. 447).

This kind of study embodies what Barthes (1973) called "a terror which threatens us all, that of being judged by a power which wants to hear only the language it lends us" (p. 46).

However, as with psychiatry, the way in which physicians themselves use language soon enough became the target of research. A typical example is Anspach's (1988) study of the language of case presentation. While the ostensible purpose of a case history is informational, in fact "it is an arena in which claims to knowledge are made and epistemological assumptions are displayed, a linguistic ritual in which physicians learn and enact fundamental beliefs and values of the medical world" (p. 357). Anspach identified various features of medical language in case presentation (such as using the passive voice and account markers to emphasise the subjectivity of patients' accounts), pointing out, for instance, that: "Physicians 'note,' 'observe,' or 'find'; patients 'state,' 'report,' 'claim,' 'complain of', 'admit,' or 'deny' " (p. 368).

A very prolific area of research with regard to language and medicine concerns the interaction between doctor and patient. Here too, the doctor is often cast as the villain. Hauser (1981) summarised the literature as follows:

Two themes are interwoven and frequently alluded to in the studies of language ... The findings describe physicians as (1) narrow in their sensitivity to patients' feelings and more subtle requests for help, and (2) withholding in their disclosure of relevant medical information (p. 114).

Many of these sorts of studies (e.g. Fisher & Todd, 1983; Marshall, 1988; Mishler, 1984; West, 1984) substantiate their findings by means of highly detailed sample transcripts of medical interviews, complete with paralinguistic information such as chair noise, uhm, hh, hm hm, and uh:m, and time indications like 1'25', but fail to indicate to what extent the identified phenomena are representative of the sample as a whole. There are, however, some exceptions, as in West's (1983) study in which she reports that of 773 questions in her transcripts of 21 doctor-patient exchanges, only 9% were patient-initiated. Patients were also more likely to respond to questions (98%) than were doctors (87%).

Criticism of this sort, although perhaps unpalatable, is in fact useful to medical practitioners, for instance in providing suggestions on how to reduce misunderstandings between doctor and patient. Although through the centuries doctors have perhaps always given more weight to what they learn from listening to the heart or palpating the abdomen, they are not unaware of the importance of conducting their verbal investigations in such a way as to obtain the most accurate information possible, and modern textbooks of clinical medicine place a strong emphasis on how to 'take a patient's history' (Butchart, 1998). However, despite its potential utility linguistic research poses a serious threat to medicine in that, at least implicitly, it tends to invalidate the bodily realities which is medicine's reason for existence. The implication is that language not only provides a pathway to the non-linguistic reality of the patient's illness, but that the illness is itself in some sense constituted in language. As Mishler (1981) observes: "The implications of constructivism are profound and far-reaching because its theorists propose that reality is constructed through human action, and does not exist independently of it" (p. 141). Some attempts to deal explicitly with medicine as a social construction are briefly reviewed below.

In the post-communist world it is easy to forget the academic prestige until recently accorded Marxist analyses. Although perhaps often obfuscatory, materialist critiques of fields such as medicine helped to refocus attention away from a purely technical, individualising approach to disease, discovering its origins instead in political and economic iniquities. Early Marxist thinkers such as Engels, Virchow and Allende did much to trace poor health to class oppression, economic underdevelopment and imperialism (Waitzkin, 1981), but more than the political economy of medicine, Marxism also tackled its ideological (and therefore linguistic) presuppositions.

Althusser (1971) in particular helped to steer Marxism away from an exclusive focus on the economic to the ideological reproduction of capital and labour in Western economies. According to Althusser modern Western democracies are kept in place not only by Repressive State Apparatuses (government, administration, police, courts, prisons), but also by Ideological State Apparatuses (churches, schools, the family, the press, the medical profession), because "the reproduction of labour power requires not only a reproduction of its skills, but also, at the same time, a reproduction of its submission to the rules of the established order, i.e. a reproduction of a submission to the ruling ideology for the workers, and a reproduction of the ability to manipulate the ruling ideology correctly for the agents of exploitation and repression, so that they, too, will provide for the domination of the ruling class 'in words' " (p.127-128). Althusser along with other post-Marxists emphasised the importance of mere words:

Why does philosophy fight over words? The realities of the class struggle are 'represented' by 'ideas' which are 'represented' by words. In scientific and philosophical reasoning, the words (concepts, categories) are instruments of knowledge. But in political, ideological and philosophical struggle, the words are also weapons, explosives or tranquillizers and poisons. Occasionally, the whole class struggle may be summed up in the struggle for one word against another word. Certain words struggle amongst themselves as enemies. Other words are the site of an ambiguity: the stake in a decisive but undecided battle (p. 24).

Apart from those inspired by Marxists ideas, there has also been a proliferation of other critical approaches to medicine since the 1970s. These include Kleinman's (1988) cross-cultural psychiatry; the 'New Cross-Cultural Psychiatry' (Littlewood, 1990; Littlewood & Lipsedge, 1987); critical medical anthropology (Lazarus & Pappas, 1986; Scheper-Hughes, 1990; Singer, 1989, 1990; Singer, Baer & Lazarus, 1990); Foucaultian accounts of medical history (Armstrong, 1995; Butchart, 1996, 1998); and Taussig's (1980, 1987) socialist-anarchist engagement with the healing process in which he attempts to "penetrate the veil while retaining its hallucinatory quality" (Taussig, 1987, p. 10).

Despite bitter in-fighting among the different approaches, they have in common a critique of biomedicine which draws attention away from disease as a physical reality, to the ways in which it is socially constructed. There is considerable variation in the degree of conviction with which the constructionist agenda is pursued, with Butchart (1998), for example, accusing all non-Foucaultian constructionist approaches to medicine of somehow still preserving a space for the 'real' world and 'real' diseases. Methodologically, constructionist methods in medicine most commonly rely for data on archival material and participant observation, the latter ranging from Kleinman's professional respectability to Taussig's emersion in South American revolutionary politics. Analysis typically takes the form of scholarly explication.

A constructionist orientation to medicine and psychiatry is part of a much wider shift towards a social understanding of language and a linguistic understanding of society. The constructionist idea that reality is in some sense a facsimile of language (rather than the other way around) has its roots in antiquity, and it would be impossible to give a definitive account of how it has come to occupy such a prominent place in academic thought, but an attempt is made to trace a few of its origins in brief outline below.

'Neither difficult nor contentious': The language of linguistics

The linguistic origins of structuralist and post-structuralist critiques of modernity are usually traced from Saussure, who in conceptualising language as a system of differences without any positive terms (in which signifier and signified are arbitrarily related), glimpsed, like Whorf, the possibility of a linguistic order more basic than, and prior to, the apparent solidity of 'houses and beds and soup kettles'. From Saussure the argument is taken via other structuralists in linguistics (Jacobson), anthropology (Lévi-Strauss), politics (Althusser) and 'semiotics' (Eco and the early Barthes) - all intent on mapping the systems of meaning which produce society and subjectivity - and on to figures such as Lacan and Foucault and their more unequivocally post-structuralist brethren (Derrida, Kristeva, Deleuze & Guattari, the later Barthes, Buadrillard) who forsake the promise of an eventual scientific blueprint of the social superstructure for the more immediate pleasures of intellectual guerilla warfare. In this narrative, continental philosophers are concerned to show how the discourses of modernity have carved up reality for us in advance, starting with the ubiquitous binary opposition between objective facts and subjective experience, while their Anglo-Saxon counterparts, by contrast, are usually shown to be obsessed with the idea of language as a representation of reality, concentrating (as in Swift's satire) on how words are used to substitute for ideas and things.

It nevertheless seems worth maintaining a space of legitimacy for lesser figures from the English-speaking world who, albeit in a small way, contributed to the constructionist approach to language(5). One such is J.L. Austin who, like Saussure, never published his magnum opus, but had it reconstructed from a series of lectures, delivered at Harvard in 1955(6). Austin's opening remark - "What I shall have to say here is neither difficult nor contentious; the only merit I shall claim for it is that of being true, at least in parts" (p. 1) - is richly ironic given the polemical content of the lectures and the greatly diminished role they accord issues of truth in the study of language. The impetus for Austin's work came from a concern with the descriptive fallacy, or what he termed the constative fallacy. Starting with a distinction between constatives (which are true or false) vs performatives (which are happy or unhappy), Austin gradually worked to the view that in general all utterances have both happiness/unhappiness and truth/falsehood. Although many of Austin's examples illustrating the shaping function of language involve ceremonial acts ("I name this ship the HMS Bounty"), he firmly established the principle that language should be seen in its social context where it not only describes (truly or falsely) a pre-existent reality, but acts to constitute social reality (happily or unhappily).

J.R. Searle (1969), a student of Austin's and later professor of linguistics at Berkeley, did much to systematise and formalise Austin's work and to extend it to everyday contexts. In his book on Speech Acts, he asked the question which in one way or another also plagued Swift, Whorf, Foucault and the various students of language, medicine and psychiatry discussed above:

How do words relate to the world? How is it possible that when a speaker stands before a hearer and emits an acoustic blast such remarkable things occur as: the speaker means something; the sounds he emits mean something; the hearer understands what is meant; the speaker makes a statement, asks a question, or gives an order? (p. 3)

Although he did not pretend to be able to answer the question, Searle argued with Austin that in getting closer to an answer linguists should take the minimal unit of communication not as the word or sentence, but rather as the speech act.

Speech acts, also called 'locutionary acts', are not as easily defined as words or sentences (Searle's, 1969, p. 16, rather vague definition is "the production or issuance of a sentence token under certain conditions"), but numerous taxonomies of speech acts have been proposed (e.g. Searle, 1976; Hancher, 1979; Stiles, 1981) and it has been suggested (Fashold, 1990) that there may be a nested hierarchy, ranging from specific speech acts (such as jokes), through speech events (such as conversations) to speech situations (such as a party). Although rather insipid compared to the more sociopolitically aware continental attempts at delineating the localities and technologies of discourse, it is evident that, at some level, the impulse is the same. The crucial difference would appear to be that the Anglo-Saxon speech-act theorists still think of the individual subject as a relatively unproblematic entity who goes about emitting 'acoustic blasts' to achieve certain concrete ends, while the structuralist and post-structuralist position is founded upon the recognition that the individual subject is itself both produced by and productive of the system of 'acoustic blasts'.

Fashold (1984, 1990) reviews various aspects of linguistics which have been influenced by speech act theory, including the ethnography of communication, linguistic pragmatics and discourse analysis. The latter ("possibly the field within sociolinguistics that has undergone more research activity in recent years than any other"; Fashold, 1990, p. 65) to an extent overlaps with a methodology of the same name in social psychology (which usually traces its roots not to speech act theory but to structuralism and post-structuralism). Before discussing discourse analysis in more detail, the following section presents an overview of the larger social contexts within which such methodologies have come to flourish.

'A bit like a whale': Postmodernity

Foucault (as presented in Parker, 1989a) claimed that Western discourse since the Renaissance can be divided into epistemes lasting roughly 150 years each. These were the Renaissance period from about 1500 to the middle of the seventeenth century when all attention was directed at recovering the true voice of God; the Classical Age lasting until the end of the eighteenth century which was characterised by an obsession with rationalism, natural science and mechanism; and the modern period which created the individual subject and which is possibly now being replaced by postmodernity. In this view both psychiatry and psychology are still largely in thrall to modernity "in which the world is experienced by people as tied together by stories of humanized science, progress, and individual meaning" (Parker, 1989b, p. 2). This modernist belief in progress "promises to release us from modern times while actually shackling us to them" (p. 12).

Postmodernity is both a critique of modernity and a condition of existence. As a critique of modernity, Ermarth (1992) describes it as follows: "Across a broad range of cultural manifestations a massive reexamination of Western discourse is under way: its obsession with power and knowledge, its constraint of language to primarily symbolic function, its ethic of winning, its categorical and dualistic modes of definition, its belief in the quantitative and objective, its linear time and individual subject, and above all its common media of exchange (time, space, money) which guarantee certain political and social systems" (p. 6-7).

As a condition of existence, postmodernity decentres the individual subject and instead gives priority to the text(7). Human subjectivity, if it exists at all, finds its expression in a shifting zone of intertextuality. Table 5.1, adapted from Brooker (1992)(8) details some of the contrasts between modernity and postmodernity. Where modernity believes in the possibility of a unifying synthesis, postmodernity playfully exposes as sham the apparent coherence in scientific or political programmes, works of art and texts of all sorts.

Table 5.1 Characteristics of Modernity versus postmodernity (adapted from Brooker, 1992)


Modernism Postmodernism


Purpose Play

Design Chance

Hierarchy Anarchy

Mastery Silence

Art Object/Finished Work Process/Performance/Happening

Distance Participation

Synthesis Deconstruction

Presence Absence

Centering Dispersal

Genre/Boundary Text/Intertext

Semantics Rhetoric

Depth Surface

Narrative Anti-narrative

Master code Idiolect

Origin/Cause Trace


Structuralism and post-structuralism (and the 'method' of deconstruction) represent a kind of thinking which has only become possible with the advent of the postmodern era. In its search for the underlying structures of meaning which operate regardless of individual intentions, and its insistence that "individuals do not speak language but that language speaks through them" (Tallis, 1989, p. 20), structuralism repudiates what Shotter and Gergen (1989) call the 'single dominant text' of modernity, which tells of the centrality and sovereignty of the individual. In its contention that truth is "a product, not a discovery, of the method that produces it" (Berman, 1989, p. 46), post-structuralism aligns itself with the postmodern idiom which thrives on chance, anarchy and play and relies on gadgets such as Derrida's sous rature (placing under erasure) to signal that what is being said and the way in which it is said is merely a temporary device to move the discussion forward. "Thus, we must use the terms that we believe to be inaccurate and inappropriate, under erasure, in order to reveal their status as useful, necessary and wrong" (Sampson, 1989, p. 7).

Postmodern academic discourse, whether of the structuralist or post-structuralist variety, is of course not without its critics. Cox (1989) calls it "an astonishing exhibition of coyly 'technical' nouns and adjectives, falsely dramatic verbs, and sentences that have lost track of their insides" (p. 73) and is at best willing to admit that it might be a "species of poetry". Minogue (1989) also complains of the lack of clarity and decries the excessive piety with which the texts of 'Continental Gurus' such as Sartre, Lukacs, Bloch, Gramsci, Habermas, Derrida, Lacan and Foucault are treated.

Foucault (1980), demonstrating both the convoluted prose and the poetic charm that Cox speaks of, admits:

For my part, it has struck me that I might have seemed a bit like a whale that leaps to the surface of the water disturbing it momentarily with a tiny jet of spray and lets it be believed, or pretends to believe, or wants to believe, or himself does in fact indeed believe, that down in the depths where no one sees him any more, where he is no longer witnessed nor controlled by anyone, he follows a more profound, coherent and reasoned trajectory (p. 79).

Behind the displeasure with post-structuralist or structuralist style lie substantive concerns about the subjectivity and extreme abstraction of the methods employed, and the radical constructionist conclusions drawn from them. Given Foucault's (1980) admission that he 'fictions' history, why should we believe that a new discursive episteme suddenly came into being just at the time that Pinel freed (or did not free) the insane? Should we put equal store in another writer of fiction's claim that "in or about December, 1910, human character changed" ( Virginia Woolf, quoted in Brooker, 1992, p. 5)? As Tallis (1989) cuttingly observes: "With only an infinitely pliable logic and their intuitions to guide them, the structuralists' journey into or away from truth is quite unfettered" (p. 26).(9)

A related problem is how seriously to take claims that reality is constructed in or from language. Cox (1989) is of the opinion that "just as we are unlikely to mistake a mime's self-conscious artistry for a plausible argument against our ability to speak, so we are unlikely, once we discover the artificiality of the Derridean method, to find in it a plausible argument for the referential inadequacy of language," (p. 66) while Tallis (1989) states:

No one would wish to challenge the obvious truth that language is implicated in the construction of reality. What is at issue, however, is the extent to which reality is intra-linguistic and language the agent or medium in virtue of which reality is structured or constituted; more particularly, the radically nominalist assumption, common to many ... critics, that the traffic is all one way: that language structures reality but reality does not influence the structure, the system of differences, that is language (p. 13).

Anglophone lucidity is not however in itself an adequate antidote for continental grandiloquence. As Berman (1989) observes, "all this enviable clarity yields no more consensus than does the most vexatious and cumbersome prose of philosophers elsewhere; and it certainly cannot (and some say it is not supposed to) yield 'truth'. Issues are never resolved, only perpetually reopened" (p. 46).

What is needed, perhaps, is for the theory to be fortified with a leavening of concrete demonstration. Although the works of some poststructuralists (such as Foucault, Barthes and Baudrillard) draw on richly detailed historical and cultural material it often appears as if this material is forced into predetermined theoretical patterns. Other poststructuralist works (such as by Derrida, Delueze and Guattarri) seem curiously empty and self-referential. In the words of Cox (1989): "These salt-flats of abstraction inspire one with a new respect for all the beautiful specifics of culture, specifics of which deconstruction takes notice only while trying to shove them into its theory" (p. 73).

'The knowledge that one seeks to disinter': The language of discourse analysis

The term discourse analysis refers to something both very specific and very nebulous. As a social constructionist approach introduced into social psychology in a 1987 book and British Journal of Social Psychology article (both with Jonathan Potter as first author), discourse analysis has a quite specific identity, at least in terms of the leading players: the group of British social psychologists around Potter, Wetherell, Reicher and, more recently, Burman and Parker; with Van Dijk and his associates as somewhat more distant European relatives.

As Parker (1989b; 1992) describes it, discourse analysis rides on the back of a series of 'crises' in social psychology periodically announced by academics (Billig, Gergen, Hare, Sampson, Shotter, Urry) disillusioned with the artificiality and triviality of traditional social psychological methods and topics. Tracing its roots to the structuralist and poststructuralist movements outlined above, discourse analysis shares in their agenda, particularly in the drive "to displace attention from the self-as-entity and focus it on the methods of constructing the self" (Potter & Wetherell, 1987, p. 102). Exactly how it differs from these or from other current approaches going by the name discourse analysis is not however always clear. The term has been current in sociolinguistics for a considerable time (Cicourel, 1980), where it is also used quite loosely to refer to approaches as diverse as speech act theory and conversation analysis.

Van Dijk (1987a, 1990) points out that discourse analysis, by virtue of its diverse origins and wide applicability, will of necessity be a cross-discipline, involving linguistics, sociology, social psychology as well as law, history and political science, all of which are "beginning to recognize that texts, documents, talk or other discursive practices constitute the central object and data of their fields" (Van Dijk, 1990, p. 6).

While it is easy to dismiss discourse analysis in psychology as faddish and ill defined, there can be no doubt that it has made a strong impact on how language is viewed in the discipline. A volume published just more than a decade ago and purporting to deal with 'language and social psychology' (Giles & St Clair, 1979) illustrates the freshness of the ideas discourse analysis brought into the discipline. In the introductory essay, Giles (1979) justified social psychology's role in language studies as follows: "If we are going to understand why individuals acquire, use and react to language and its varieties in the way they do, we require a greater understanding of the dynamics of attitudes, motivations, identities and intentions, that is, social psychological phenomena" (p. 2). Social psychology itself is defined as "the study of an individual's behaviour in his or her social context" (p. 2). Contrast this to Parker's (1989b) book on social psychology, in which the uncritical acceptance of the individual subject as psychology's proper object of research is itself the main topic of discussion.

Where traditional social psychology produces information on the attitudes, motivations and intentions of individual subjects, what (apart from a critique of traditional social psychology) does discourse analysis produce? Potter and Wetherell's (1987) initial answer - interpretive repertoires - may at least in part explain why discourse analysis has aroused so much interest in such a short time. An interpretive repertoire is "basically a lexicon of terms and metaphors drawn upon to characterize and evaluate actions and events" (Potter & Wetherell, 1987, p. 138). Interpretive repertoires are typically "organized around specific metaphors and figures of speech" (p. 149). A repertoire relating to the term community in radio, television and newspaper reports on, as well as eyewitness accounts of, a 'race riot' in Britain is presented in Table 5.2.

What made Potter, Wetherell and Reicher's work special is that, despite the fact that they drew their ideological inspiration from structuralist and poststructuralist sources (as reflected among other things in the fact that their object of interest is the language being used, not the characteristics of the language users), they collect, analyse and present data in a format which is intelligible to traditional positivist researchers. Unlike 'hard-core' deconstructionists such as Derrida, they therefore (at least minimally) wish to create the impression that their 'findings' are of the same order as that produced by hard-nosed empiricists in the social sciences.

Table 5.2 The community repertoire (from Potter & Reicher, 1987, p. 32)


Paradigmatic alternatives____________________________________________________________________

Local residents or local residents with specific social organisation

Black community or White community or Mixed community

Currently exists or Existed in past or May exist in future


Sample predicates Metaphors (where relevant)






Close-knit Spatial



Mature Organic



Acts Agency




The idea of constructing such repertoires is not new. Black (1962) in his classic work on metaphors, for instance calls for the identification of 'archetypies' in bodies of speech or writing. By archetype he means:

a systematic repertoire of ideas by means of which a given thinker describes, by analogical extension, some domain to which those ideas do not immediately and literally apply. Thus, a detailed account of a particular archetype would require a list of key words and expressions, with statements of their interconnections and their paradigmatic meanings in the field from which they were originally drawn (p. 241).

Although the exact nature of the discourse analytic product is somewhat in contention, it is certainly more tangible than merely a philosophical critique of modernity. Rather than an interpretive repertoire, Parker (1989a) wishes to expose discourses. A discourse is "a system of statements which constructs an object. This fictive object will then be reproduced in the various texts written or spoken within the domain of discourses" (p. 62). Gilbert and Mulkey (1984), on the other hand, produce interpretive devices, such as the TWOD (Truth Will Out Device) often used in scientific writing (and of which numerous examples were cited in Chapter 4). Wetherell and Potter (1992), in some of their later work no longer speak of interpretive repertoires, but of 'maps', which (much like repertoires) refer to the organisation of discursive "practices, arguments and representations" (p. 1).

There are also relatively specific descriptions of how to arrive at these repertoires, maps, discourses or devices. Potter and Wetherell (1987) offer ten steps to discourse analysis; Parker (1989b) explains how to deconstruct a text in three steps; and again (Parker, 1990a; 1992) how to analyse discourses in twenty steps. Although many of these steps are described in rather nebulous terms, this is clearly a far cry from the Foucaultian whale. Potter & Wetherell (1987) expressly acknowledge that one of the weaknesses of earlier approaches, such as ethnomethodology is that "the reader of the ethnographic report of this kind is dependent on the researcher's description both for what they know about the data and for their evaluation of the researcher's conclusions" (p. 30), while Gilbert and Mulkey (1984) emphasise the importance of providing "closely documented descriptions" (p. 14) of whatever features are identified in texts.

However, even as they have moved away from the 'salt flats of abstraction' towards the 'beautiful specifics of culture', discourse analysts seem to be gripped by a fear that they will reinstate a regime of truth as oppressive as the subjectifying empiricism they are trying to subvert. Thus to counteract any negative side-effect of his 'three steps', Parker (1989b) warns that "it might be tempting to think of deconstruction merely as a method." (p. 58) and, again, that his 'twenty steps' do not "constitute a method" (Parker, 1992, p. 5). Even as he generates more detailed methodological specifications, Parker (1992) protests more and more vehemently that he is not advocating a method, e.g. "discourse analysis is not, or should not be, a 'method' to be wheeled on and applied to any and every topic" (p. 122). To use Parker's own methodology, this is clearly a case of a putative 'discourse analysis discourse', meeting criterion 6 (steps 11 and 12) for official certification - i.e. employing disingenuous reflexive strategies of the 'don't get me wrong' variety.

Of Potter and Wetherell's ten steps, Parker (1989b) says that they "can be taken with a pinch of salt, but ... can be used as part of the presentational rhetoric to get through institutional barriers" (p. 160). Potter and Wetherell for their part (Potter, Wetherell, Gill & Edwards, 1990), accuse Parker of having an overly reified vision of discourses, describing his position rather graphically as "endorsing something akin to the geology of plate tectonics - great plates (discourses) on the earth's crust circulate and clash together; some plates grind violently together; others slip quietly over top of one another; volcanoes burst through while massive forces work unseen below" (p. 209). Perhaps for the same reason they reject the reifying obsession with sampling which traditional empiricist research supposedly suffers from.

In the post-structuralist world the sin of reification is of course a grievous one, and even Foucault (1980) at times feared that he might succumb:

And after all, is it not perhaps the case that these fragments of genealogies are no sooner brought to light, that the particular elements of the knowledge that one seeks to disinter are no sooner accredited and put into circulation, than they run the risk of re-codification, re-colonisation? (p. 86)

It can however be argued that no degree of looseness in one's methodology can protect one against this sin, as the looseness itself soon enough becomes a reified article of faith. Burman (1991) argues that although discourse analysis as currently practised has helped to draw psychology's attention to how language "produces and constrains meaning, where meaning does not, or not only, reside within individuals' heads" (p. 327), it does not hold the monopoly on 'progressive' research. Although the application of a particular methodology may be radical and politicising, the method itself may be as open as any other to being used in a falsely value-free way.

Indifference to traditional research concerns such as sampling issues therefore does not guarantee an escape from reification and is open to the usual criticisms of subjectivity and lack of replicability. That this is not a trivial issue is demonstrated by two book-length publications produced in the discourse analytic spirit.

Salomon Rettig's (1990) Discursive Social Psychology of Evidence illustrates just how seriously things can go wrong when Potter and Wetherell's (1987) instruction to underplay sampling issues is followed. Rettig's book consists of 40 pages of theory, followed by 160 pages of transcription and (very minimal) analysis of half a dozen or so conversations. In case one mistook this for any form of serious research, Rettig (1990) is quick to point out that:

There is no claim to universality, nor to scientific rigor. There is, however, a claim to the authenticity of the material. I hope that the reader will enjoy the material as much as those of us who participated in its production, for it has been a very satisfying human enterprise, indeed (p. vi-vii).

As one may expect from contexts in which authenticity and enjoyment are privileged over rigour, there are numerous factual errors of the most elementary kind, such as a claim (p. 83) that three participants in a conversation about the sex of a particular participant in another (transcribed) conversation all agreed that she was a woman, when the transcription indicates that one thought she was a man and another was uncertain.

The second example, Labov and Fanshel's (1977) classic analysis of a therapeutic encounter, although published long before discourse analysis came on the scene, and not prone to the same sorts of infelicities as Rettig's (1990) work, also helps to illustrate the importance of attending to sampling issues. Just the opposite of Rettig's, theirs is a 361 page analysis of 15 minutes (8 pages transcribed) of a therapy session. Their central concern is to explicate "the sequencing rules [which] operate between abstract speech actions" (p. 350) and in order to do this they have to analyse exchanges in microscopic detail. In part Labov and Fanshel limit themselves to 15 minutes from a single conversation for purely practical reasons (15 minutes each from 100 conversations would presumably require an impossible 36 100 pages of analysis)(10), but additionally they rely on an assumption that the kinds of sequencing rules they will discover operate in other conversations as well. This assumption simply does not hold for the discourses or interpretive repertoires with which discourse analysis wishes to work, and larger (and more diverse) samples are in fact better, as Potter and Reicher (1987) and Potter and Wetherell (1987) argue in drawing their texts from a range of sources such as broadcast media, Hansard, newspapers and interviews. Parker (1992) also emphasises (in step 11) the importance of finding and describing a discourse as it occurs in more than one kind of text. This does not mean that these analysts necessarily see individuals as the source of language or wish to relate linguistic features to the characteristics of individual subjects.

To gain respectability within the current status quo, which may be necessary if it is to have a real impact on psychological research, discourse analysts therefore have to pay some attention to traditional signs of scientific rigour, such as using representative sampling strategies, replicable analyses, and succinct modes of reportage. This is exactly the kind of call that has long been made by others involved in 'qualitative' data analysis, such as Miles and Huberman (1984) and Kirk and Miller (1986) who are adamant that qualitative research "does not imply a commitment to innumeracy" (Kirk & Miller, 1986, p. 10).

Despite its genuflections to methodological rigour, discourse analysis has thus far certainly succeeded in avoiding becoming a mere methodology. A survey of the discourse analytic studies published in Burman and Parker (1993) reveals a reasonable degree of heterogeneity. Although the data source for these studies were almost invariably interview transcripts(11) (Gill, 1993; Macnaghten, 1993; Marks, 1993; Marshall & Raabe, 1993; Moir, 1993; Stenner, 1993; Widdicombe, 1993), the number of people interviewed varied from unspecified (Widdicombe, 1993) to ten or fewer (Gill, 1993; Marshall & Raabe, 1993; Stenner, 1993) to 40 (Moir, 1993), while the type of person interviewed included students (Moir, 1993), people classed as conservatives or liberals on a psychometric measure (Marshall & Raabe, 1993), disc-jockeys (Gill, 1993), members of the 'gothic' subculture (Widdicombe, 1993), a married couple (Stenner, 1993), and helping professionals (Marks, 1993).

The aim of these analyses seem to be twofold: 1) To identify the kinds of socially conditioned repertoires or discourses used by and reproduced in the interviews; and 2) to show how these are used, together with other conversational and textual techniques or gambits, to achieve purposes such as coherence or the silencing of less powerful participants. In Widdicombe's (1993) formulation, "the object of analysis is to explicate the culturally available resources and tacit reasoning procedures which seem to inform what is said, and to identify the nature of the interactional tasks thereby addressed" (p. 97). However, what counts as 'culturally available resources' and 'interactional tasks' vary widely. Some of the 'culturally available resources' identified in these studies are sexist constructions of women's capacities and men's willingness to listen to women (Gill, 1993); the different ways in which nature can be depicted (Macnaghten, 1993); the 'subject positions' available to men and women (Stenner, 1993); and the therapeutic and reflective obligations of professionals versus the 'needs' of clients (Marks, 1993). The 'interactional tasks' that these are used to address include justifying the small number of female DJs (Gill, 1993); justifying or opposing a new land-fill site (Macnaghten, 1993); making the adoption of 'gothic' style appear as an authentic individual choice (Widdicombe, 1993); making the choice of a particular career appear authentic (Moir, 1993) and deflecting talk from interprofessional conflict (Marks, 1993).

Most commonly, deductions are warranted by means of short illustrative extracts from interview transcripts (e.g. Gill, 1993; Macnaghten, 1993; Moir, 1993; Stenner, 1993), and sometimes by the device of presenting a close reading of a single longer extract (Widdicombe, 1993). Apart from discourse theory, eclectic reference is made to a variety of other approaches, such as Holland's occupational types (Moir, 1993), feminism (Gill, 1993), sociological and social psychological work on youth subcultures (Widdicombe, 1993) and action research (Marks, 1993).

It should not however be thought that discourse analysis is nothing more than a particular ideological orientation with no specific methodology. Two aspects distinguishing discourse analysis as a methodology stand out. The first, which has already been alluded to, is that, to a far greater extent than the philosophical movements from which it draws its inspiration, it relies on traditional empiricist distinctions between theory, method and data. Discourse analysts, despite all they may say against positivism, feel compelled to back up their claims about how people use language and language uses people with the kind of evidence recognisable to empiricists as 'data'. The second distinguishing feature is that in discourse analysis a conscious attempt is made to work 'from the bottom up', that is, to derive theoretical insights from data rather than to impose theoretical systems on the data. Although most pronounced in 'Grounded Theory' (Glaser & Strauss, 1967), this is a common theme in most forms of qualitative research. Thus Gilbert and Mulkey (1984) call for discourse analysts to stay 'close to their data': "Instead of applying an abstract, preconceived language to our data in order to show how discourse arises from and reproduces complex social structures, we ... begin with an examination of those terms and interpretative features which seem to arise naturally in the course of participants' own discourse..." (p. 16)

This principle is sometimes presented in discourse analysis as entailing having to stay near the surface of the data. Potter and Wetherell (1987), for example, call for researchers to 'range over' rather than penetrate texts, saying that "we do not intend to use the discourse as a pathway to entities or phenomena lying 'beyond' the text" (p. 49). It can be argued that analysts such as Potter and Wetherell do indeed wish to discover entities 'beyond' the text, and that these entities are merely of a different sort from that typically discovered by for example psychologically oriented content analysis - i.e., trans-personal discourses rather than individual attributes. Although the surface-depth distinction may therefore be suspect, the principle is clear - discourse analysis is in the first place concerned with understanding how language itself works, rather than treating it as a window onto some other reality. In Gilbert and Mulkey's (1984) phrase, accounts are treated as "topic instead of resource" (p. 13) - i.e., it is discourse which is of interest, rather than the individual actors through whom the discourse speaks.

As currently constituted discourse analysis is a qualitative approach. This is so for both historical and methodological reasons. Historically, quantitative positivist research has dominated the social sciences, and discourse analysis is therefore 'naturally' allied with the various qualitative approaches which have been formulated in opposition to this hegemony. Methodologically, quantitative research has been perceived as imposing preconceived categories on data as, for example, in quantitative content analysts counting the number of words showing 'negative affect' in a text. This is clearly incompatible with the 'bottom-up' discourse analytic approach discussed above. However, imposing versus discovering categories in data is clearly a matter of degree, and it is possible to imagine quantitative approaches that approach the discourse analytic ideal of letting the data speak for itself. The advantage of incorporating quantitative techniques into the discourse analytic repertoire would be that this would help discourse analysis escape from the unproductive quantitative-qualitative dichotomy into which it has been historically interpellated.

One possible source of quantitative techniques which can be applied to language is the discipline of linguistics. Currently available highly structured, automated and computational approaches to linguistic analysis are therefore reviewed as a possible tool for discourse analysis in Chapter 6; and in the following chapters new techniques developed from these are applied and evaluated. The dissertation concludes, from a methodological point of view, with an assessment of the extent to which the adoption of such techniques leaves discourse analysis vulnerable to reification and "recuperation(12) by positivist research" (Burman, 1991, p. 334).

1. 1 All references to Whorf are from his collected works edited by Carrol (1956).

2. 2 Webster was much possessed by death / and saw the skull beneath the skin. - T.S. Eliot, Whispers of immortality

3. 3 Since Foucault's original work, there has moreover been a proliferation of texts equally concerned with the history of discourses of insanity, and of insane discourses, but presented in a more conservative idiom and making use of the usual convention of close reference to sources of evidence (cf., Ingram, 1991; Porter, 1987a; Porter, 1987b; Scull, 1991b; Turner, 1990).

4. 4 Including a few now outmoded psychiatric terms such as functional, organic and nervous disease.

5. 5 Van Dijk (1987b) also attempted the impossible task of sketching the shift to language without reference to continental linguistics and philosophy: "Structural and generative grammars in the 1960s and 1970s have been busy developing formal systems of analysis, in which language users and social contexts were nearly fully ignored. Pragmatics introduced the notion of speech act, and thereby came a step closer to the study of social interaction, but its approach remained fairly philosophical and abstract. Textlinguistics and more generally discourse analysis (including conversation analysis) broke the rigid sentence boundary of current grammars, and focused on the more natural units of language use and communication, viz. text and talk" (p. 15).

6. 6 Published posthumously in 1975.

7. There is, however, considerable controversy about the role of text among theorists who could broadly be described as 'postmodern'. While some give priority to language and text as conventionally understood, others merely wish to interpret social practices and institutions as if they were texts, while yet others consider the postmodern emphasis on textuality a distraction from researching the concrete practices and effects of power.

8. Who took it from Hassan (1985).

9. 9 The extent to which such critiques talk past the structuralist/post-structuralist enterprise is illustrated by Foucault's (1980) remark about truth as itself a construction: "We are subjected to the production of the truth through power and we cannot exercise power except through the production of truth ... we must speak the truth; we are constrained or condemned to confess or to discover the truth. Power never ceases its interrogation, its inquisition, its registration of truth: it institutionalises, professionalises and rewards its pursuit. In the last analysis we must produce truth as we must produce wealth." (p. 93)

10. 10 One suspects that practical considerations are also in part behind Potter and Wetherell's (1987) argument, as is evidenced by their observation that the ratio of recorded to transcribed time is easily 1:10. Some suggestions for overcoming this difficulty are presented in the next chapter.

11. 11 Showing perhaps the extent to which these authors, despite their professed allegiance to post-structuralism, are still subject to phonocentric biases. One exception is Macnaghten (1993) who also uses official documents.

12. A Marxist term, meaning to render opposition harmless by recruiting it back into the political mainstream.

To contents page

To Psychiatric Activism Page