The Turn to Discourse in Social Psychology
Wood, Linda A., Kroger, Rolf O., Canadian Psychology
We discuss the emerging turn to discursive social psychology as an alternative to experimental social psychology. We note that the barriers to change are rooted in the history of the discipline, in the failure of researchers to recognize the distinction between movements and actions and in their reluctance to switch from positivist to post-positivist criteria. We outline the tenets of discursive psychology and of its associated method, discourse analysis. Illustrations of discourse analysis are drawn primarily from a recent study of date rape. Throughout, we emphasize the centrality of discourse in social life and the definition of the social being as Homo loquens.
Nous discutons du passage a la psychologie sociale discursive comme solution de rechange a la psychologie sociale experimentale. Nous constatons que les obstacles au changement trouvent leur source dans l'histoire de la discipline, dans l'echec des chercheurs a reconnaitre la distinction entre les mouvements et les actions, et dans leur resistance a passer des criteres positivistes aux criteres post-positivistes. Nous soulignons les principes de la psychologie discursive et de l'analyse du discours, methode qui y est associee. Les exemples d'analyse du discours sont principalement tires d'une etude recente sur le viol commis par des connaissances. Dans tout l'article, nous mettons l'accent sur l'importance du discours dans la vie sociale et sur les definitions de l'etre social qu'est l'Homo loquens.
A decisive turn in the development of social psychology occurred when social psychology adopted the experimental method under the influence of Kurt Lewin. Lewin (1951) had been inspired by the success of field theory in physics and the possibilities he saw in applying field-theoretic principles of force, tension, constraint and context to the study of social-psychological issues. His experiments in leadership style (autocratic, democratic, laissez-faire) became classics in the new experimental social psychology. Festinger, Lewin's most prominent pupil, along with a phalanx of researchers who subsequently became the standard-bearers of experimental social psychology, adopted Lewin's program, first at MIT and later at Michigan and Minnesota. Festinger took an intuitive and creative insight into social processes (derived from his observation of the circulation of rumours about the consequences of earthquakes in India; Festinger, 1957) into the laboratory to spawn two decades of research on cognitive dissonance. Research on cognitive dissonance became the touchstone of experimental social psychology. Interest in the topic has declined, but the commitment to the experimental method remains seemingly unshaken.
This commitment has not gone unchallenged. Experimental social psychology came to experience a crisis of confidence on methodological grounds. The crisis literature focused first on ethical questions (Baumrind, 1964). Should an entire discipline rely on lying (technically, deception) as a necessary methodological tool? This was a difficult question in the politically volatile and optimistic 1960s, and is still so today. It involves more lasting questions concerning, for example, the opacity of results obtained via the strategy of deception (Kroger & Wood, 1980). The crisis literature then turned to methodological concerns. Orne (1962) raised the spectre of demand characteristics: the tendency of social psychological experiments to be peculiar social situations in which subjects respond to the social demands of the experimental situation and not just to the independent variables selected by the experimenter. Control over the subject's responses, the raison d'etre of the experimental method, seemed impaired. Rosenthal (1966) added the notion that in the fragile environment of the social-psychological experiment, the characteristics and expectancies of the experimenter might contribute to the results of the experiment as significantly as the classically defined independent variable.
Clearly, there was something amiss in the realm of experimental social psychology. Festinger himself departed from social psychology in 1964, still a relatively young man, because "I...needed an injection of intellectual stimulation from new sources to continue to be productive" (Festinger, 1980, p. 248). The crisis of confidence generated a voluminous literature which also gradually went into decline. It never did answer its own questions in a definitive way. Again with hindsight, it is now apparent that it could not do so because it was also hopelessly mired, as was experimental social psychology itself, in positivism as the philosophy of science. The research on demand characteristics and experimenter effects was thus severely constrained. Fundamental change in research practice could only come from fundamental change in theoretical orientation.(f.1)
Onto this scene burst the publication of Harre and Secord's (1972) The explanation of social behaviour. It was widely cited but, curiously, also widely disregarded. It proved an enigma to seasoned experimental social psychologists whose training had been in an entirely different tradition. They were steeped in the assumptions of positivism and they seemed unfamiliar with Johann Gottfried Herder, Wundt's Volkerpsychologie, Vygotsky's and G.H. Mead's elaboration of the Wundtian ideas, with the resonances that these ideas found in the work of their contemporaries in adjacent disciplines (e.g., Goffman and Garfinkel). And the very size of experimental social psychology in the post-War period permitted a kind of parochial inwardness that militated against looking at the work in neighbouring disciplines which was seen as less rigorous than the results of experimental studies.
The difficulties were compounded because Harre and Secord's argument was grounded not in the familiar language of positivism, but in the Wittgensteinian philosophy of language. That philosophy was put to brilliant use by John Austin (1962) in his classic How to do things with words. The Wittgensteinian stance proved to be, and seemingly continues to be, a formidable barrier to change and improved understanding. But it also cleared the way outside the bastions of experimental social psychology for a profound shift that resulted in an explosion of creative work by scholars and researchers not committed to the old verities. They created the turn to discourse.
We cannot enlarge here on the Wundtian legacy (see Kroger & Scheibe, 1990; Kroger & Wood, 1992a). Whatever the ultimate judgment on the Wundtian legacy may be, it is clear that Wundt, as early as 1863 in his Vorlesungen uber die Menschen-und Thierseele (Lectures on human and animal psychology), identified a fundamental problem of psychology, a problem that has delayed the turn to discourse and that continues to challenge the discipline. That is, he recognized the dual nature of psychology: rooted in biology on the one hand, in culture on the other. He anticipated the vital distinction between res naturam and res artem articulated in recent times by Stuart Hampshire (1978). The distinction is between the natural world and the cultural world. The cultural world is constructed by human beings via language to suit their species-specific projects. The distinction includes the point that different methods and modes of explanation are required to deal with the problems raised by the two different realms. To try to assimilate the cultural to the natural, to subsume res artem under res naturam -- that is, to engage in the reductionist project -- has been shown to be unproductive. The two realms require two different ontologies, not just the Newtonian space/time paradigm that is appropriate to the natural world (Harre & Gillett, 1994; Wood & Kroger, 1998). One message is that the method must suit the problem. Method cannot be dictated by some ideological position (e.g., the positivist view that the experiment is the acme of methodological sophistication for all areas of psychology, that the independent-dependent variable format is the sine qua non of psychological investigation (Bickhard, 1992; Winston, 1990; Koch, 1959). The questions of social psychology seem to call for more subtle and varied approaches.
We must leave it to future historians of psychology to sort out the reasons why the experimental social psychologists ignored the wider historical background of social psychology, embodied in the Wundtian legacy, and remained reluctant to take up the Wittgensteinian message regarding the pragmatic functions of language -- the action component of language -- while remaining fixed on the purely referential, descriptive functions of language. This focus encouraged a kind of literalism about the use of language. It entailed the faith that, among other matters, the self-reports of subjects could be taken more or less at face value and the instructions of experimenters could be taken to have literal import (Kroger, 1988; Kroger & Wood, 1980). The turn to discourse calls these basic assumptions into question (Edwards & Potter, 1992; Potter & Wetherell, 1987).
Our point is not that language has been neglected but how it has been treated. There is no question that some social psychologists have been concerned with questions of language, albeit in a restricted sense. What springs readily to mind is the work of Roger Brown on language acquisition, of Lambert and his colleagues at McGill on attitudes toward language in the context of the debate about bilingualism in Canada, and of Rommetveit on structure and meaning, to cite only a few prominent examples. There is also the extensive literature on the "semantics of social structure" (Brown, 1965), on the forms of address. In fact, we contributed to that literature in a modest way (e.g., Kroger & Wood, 1992b). But, as we have shown in an empirical survey of textbooks and handbooks in social psychology (Kroger & Wood, 1992a), that literature never did become part of the mainstream of social psychology. For example, it was not until the third edition of the authoritative, multi-volume Handbook of Social Psychology that there appeared a separate chapter on language (Clark, 1985), and that chapter concentrated on cognitive rather than social issues. However, what is more important than the consignment of language to the periphery was its treatment in the positivist tradition as just another independent or dependent variable.
There is a further barrier to understanding the turn to discourse. It is the conflation of movement and action into the concept of behaviour. It is critical to distinguish between observable physical movements and the interpretation of those movements (Harre & Secord, 1972). We can talk, for example, about raising the right arm and curling the fingers (movement) or about giving the Communist salute at a Communist rally (action). These are not simply two different descriptions; the latter is an interpretation of the former. Actions in this sense are the meanings we give to movements.(f.2)
The question arises: what is the subject matter of social psychology? It appears that social psychologists take the topics or phenomena of the discipline to be matters such as conformity, obedience, prejudice, helping, etc., as reflected in the chapter headings of standard textbooks. The impression given is one of certainty about what social psychologists study. But if one digs a little deeper there is yet again the Jamesian buzzing bloom of conceptual confusion. There is little evidence that the distinctions between physical movements and their meaning, between res naturam and res artem, have penetrated mainstream research in social psychology.
It is clear that social psychologists are not interested in studying physical movements or mere utterances (e.g., the movement of the lever in the Milgram experiment; the utterance "Line A is longer than line B" in the Asch experiment); rather, the concern is the meaning of those movements, the actions thought to be constituted by these movements (e.g., "obedience," "conformity"). The problem is that movements have multiple meanings, depending on the situation, the context, and how speaker and hearer are positioned vis-a-vis each other (cf. Kroger & Wood, 1980). This multiplexity is difficult to capture in the traditional experiment in which causality is thought to be embodied in the unidirectionality of the independent variable selected by the experimenter, and in which that variable is thought to function no matter what the wider social context of the experiment might be. That was one of the unresolved questions raised by the crisis of confidence. Austin (1962) was careful in spelling out the conditions that must be met for an utterance (or other movement) to have performative force as a particular speech act (or, more generally, as an action in the social realm). For example, "I dub thee Sir Lancelot" does not constitute the act of knighting unless uttered by the monarch in a public ceremony with Lancelot (and not someone else) kneeling before her.(f.3)
When social psychologists conflate movements and actions into the summary concept of behaviour, they arbitrarily privilege and reify their own interpretations: that the movement is obedience or conformity or whatever. But the construction of movements as particular sorts of phenomena, the giving of meaning to those movements, is not straightforward by any means. In any event, it is these meanings that are the stuff of science, not mere "behaviour" or raw data points or undigested "facts." And for social scientists the meanings are not only contingent (on the situation and the persons involved, including the interpreter), they are in addition not neutral and they do have consequences. The interpretation of a particular movement as "conformity" (versus, for example, "cooperation" or "solidarity") is not only disputable scientifically, but it can carry negative connotations and can affect the person whose movements are so "described" (unlike the proverbial rock). Whether "I can help you to get an A" said by a professor to a student is called sexual harassment or good teaching may be strongly contested and is highly consequential for both parties. The matter is especially problematic when the movements of interest are utterances (interviews, questionnaire responses, judgments of lines). Such talk, or occasioned language use, needs to be treated systematically and conceptually. It cannot simply be taken literally and treated as transparently equivalent to some particular action.
We cannot manage this task unless we consider how talk is used -- that is, unless we treat it as discourse, not merely as a conduit for messages or as a reflection of some presumed inner entity. The failure to see talk as action in its own right is an issue not only with respect to the actions of participants, but also with respect to the actions of researchers. If we do not understand how movements are formulated as actions in and through discourse, we fall into the trap of thinking that we are offering objective descriptions of movements, and fail to see that we ourselves are using language to impose our own interpretations, for example, that the pushing of the lever is obedience.(f.4)
Let us offer one concrete example of this point. It is tempting to say that if we are analysing a conversation between two persons from different generations, we are studying "intergenerational communication" (N. Coupland, personal communication, May 14, 1994). But we do not actually "have" intergenerational communication; there is no such "thing" in the sense of some physical object or movement. All we have is talk between two people of particular ages; whether it is reasonable to frame it as intergenerational depends on how the people involved treat it, whether they see their ages as "different" and take this into account in their conversation, whether they make a point, via talk, of reaching across the "generation gap." Dedicated professors would likely reject the intergenerational interpretation of their interactions with students in favour of something like "teacher-student talk about English literature."
In sum, the turn to discourse has been delayed by, among other matters, the conflation of movements and actions and by the failure to recognize that talk is action. And it has been delayed by the failure to observe that what we do incessantly is to talk and so manage (or mismanage) our relations with our fellow social beings. Any faculty meeting gives ample opportunity to observe this process.
The Discursive Turn
What is the turn to discourse? We begin, conceptually, with the tenets of discursive psychology, the basic set of assumptions underlying the turn to discourse (Edwards & Potter, 1992; Harre & Gillett, 1994). We then turn, on the methodological side, to the analysis of discourse as action (Wood & Kroger, 1998).
The major assumption of discursive psychology is that the phenomena of interest in social and psychological research are constituted in and through discourse (Edwards & Potter, 1992; Potter & Wetherell, 1987). As Sampson put it, "Discourse theorists maintain that talk is constitutive of the realities within which we live, rather than expressive of an earlier, discourse-independent reality" (1993, p. 1221). This is not an easy point. Discursive psychology involves a major shift from the conventional view of language as a tool for description and as a medium for communication to a view of language as social practice, as a way of doing things. We must free ourselves from the common-sense conviction that talk is just talk, that the real action is elsewhere. The world runs on talk (and, of course, on writing).
It is true that people as physical beings get killed -- by means beyond the realm of language -- but after the fact we, as social beings, still have to decide whether those killings are to be seen as Murder One, Terrorism, Manslaughter, Accident or Suicide. The answers, if they come, come through talk, through press conferences, through newspaper stories, through technical reports, through negotiations between stakeholders. They will not come through the so-called raw facts which are always subject to interpretation. The social-psychological meaning and the impact of incidents like those surrounding airline-crashes are constructed in talk and so it is on the talk that we must concentrate to apprehend "reality," both physical and social. To downgrade talk ("it's just talk, when are they going to do something about it") is to engage in a stereotypic, common-sense denial of the performative force of language. It is to deny the revolutionary insights into the multiform nature of language first broached by Wittgenstein and subsequently elaborated by the philosophers of mind (e.g., Austin, 1962) and anchored empirically in the burgeoning literature on the analysis of discourse (e.g., Edwards & Potter, 1992).
We require at least three reversals in the practice of research. First, we must abandon the division between talk and action and emphasize talk as action, the performative force of language in use (Austin, 1962). Second, we must move to the notion that talk is the event of interest, that it is the talk that constructs reality, for example, the reality of the TWA crash, the outcome of the US Tennis Open (which, at crucial junctures, is determined not by the raw physical movements of the ball but by the calls of the chair referee and line judges). At the most general level, the topic of social psychology is discourse because it is in and through discourse that the specific topics of interest (e.g., attribution, social comparison) are constituted. The earlier research on language and social psychology has not redressed the neglect of discourse in social psychology. Language use in that orientation was usually viewed simply as another specialized topic within larger topics (e.g., matched-guise studies of prejudice, studies in attribution, attitude change). That view fails to recognize that prejudice, for example, is constituted in and through language use. The problem here is that language use is treated as a social practice rather than as the social practice.(f.5)
The turn to discourse simultaneously requires that we abandon the notion that talk is merely indicative of presumed deeper internal entities or of external events. It is probably more straightforward to adopt the notion that mind is constituted discursively (Harre & Gillett, 1994). Mind is the software, not to be confused with the brain, the hardware that is necessary but not sufficient to tell us what we are about.
Lastly, we must abandon the efforts entrenched in traditional psychology to eliminate variability through techniques of data reduction. These efforts are a holdover from the positivist search for generality that does violence to the peculiar epistemological requirements of social psychology (Boulding, 1980). Instead, we must exploit variability as a matter of interest in itself, as a reflection of the necessarily variegated nature of social life which we ignore at the peril of oversimplification (Potter & Wetherell, 1987; Wood & Kroger, 1995, pp. 84-85). These are major tasks and major shifts that require a redefinition of the narrow image of science long carried by psychologists (see Kroger, 1991). The redefinition requires attention especially to the distinction between res artem and res naturam, a distinction that cannot be overemphasized for our present purposes.
The call of discursive psychology is for increased conceptual and methodological rigour, not for a decline into a mushy, relativistic, touchy-feely methodology.(f.6) The new methods must be suitable to penetrate beyond the common-sense appearances of social interaction. We can no longer take at face value, if we ever could, the self-reports of participants in traditional experiments and interview and questionnaire studies that are given a semblance of scientific respectability through the intricacies of contemporary, computer-assisted statistical analysis and other scientistic devices (Harre, Clarke & De Carlo, 1985, ch. 1). We must go back to the data, to the initial utterance, to the performative force of these initial actions. That is where our attention must be, not on the esoteric numerical transformations that are so prominent in the traditional literature. This is not to deny the power of numbers where they serve unambiguously as descriptive and analytic tools. But the adoration of numbers, of quantification for its own sake, is less than useful.(f.7)
Why discourse analysis?
Discourse analysis should be the method of choice for social psychology because it provides a methodology tied to a theory of human conduct appropriate to the epistemological requirements of the discipline. That is, the use of discourse analysis involves an explicit theory of method that is grounded in an explicit theory of social psychology, namely discursive psychology. This grounding is similar to what we find in physics where the use of a thermometer, for example, involves an explicit theory of how the thermometer works in terms of the properties of mercury, and where the interpretation of the results of measurements of temperature takes into account that theory of method.
There is an additional reason for using discourse analysis. It promotes flexibility. We have used the analogy of the microscope to sort out the array of different varieties of approaches to discourse analysis that have appeared in the literature (Wood & Kroger, 1998). We employ that analogy here to show that like the use of the microscope in biology, discourse analysis permits numerous possible ways of assessing the material at hand. In biology, one can use a variety of different levels of magnification, different sorts of stains to detect the presence of particular substances, different focuses, and so on. Similarly, discourse analysis can be seen to consist of an extensive set of devices or strategies for examining discourse in a variety of ways. We can vary the level of magnification and can look at large chunks or fine details; we can focus on the discourse as a whole or on specific discursive devices (e.g., the use of certain forms of address; e.g. Kroger, 1982), just as one might use a particular stain. We can examine discourse in terms of a particular theoretical concern (e.g., facework), just as one might employ a particular lens (e.g., ultraviolet) to highlight certain features on a slide. We can change our focus to place certain features of talk in the foreground or background of analysis.
We might think of the techniques and tools of discourse analysis as prostheses, as extensions of the unaided senses, in the same way that a microscope enables us to see what cannot be seen with the naked eye (see Boulding, 1980). The natural sciences, in answering questions arising from res naturam, largely use physical prostheses: various kinds of apparatus. The human sciences, including social psychology, in answering questions arising from res artem, largely use conceptual prostheses, that is, theoretical concepts and distinctions that highlight pertinent features and direct our attention to patterns in the flux of data. An example is the distinction between empiricist and contingent repertoires,(f.8) which helps us to identify the ways in which scientists simultaneously defend their own work and criticize that of others (Potter & Wetherell, 1987).
In this section, we offer some concrete examples of the analysis of discourse to demonstrate some of the features of doing discourse analysis. We first distinguish discourse analysis from other forms of qualitative analysis, and from content analysis which uses qualitative data but subjects them to quantitative analysis in the positivist manner. It is arguable whether content analysis represents an advance, either conceptually or methodologically, over the most common methods in the social sciences.
First, discourse-analytic techniques are grounded in the principles of a general social-psychological theory, namely, discursive psychology, as we have pointed out. Second, there is the requirement that analysis of the discourse must precede coding because the potential code (e.g., "interruption") for a particular item (word, phrase, etc.) depends upon the precise working of that item in its particular context (by which is meant both various features of the setting and the sequence of talk in which the item occurs) (e.g., Schegloff, 1993). For example, an utterance cannot be identified as an interruption simply because it "looks" like one (e.g., there is overlapping speech); the identification must be justified by other relevant features of the discourse in which the utterance occurs (e.g., it is treated as an interruption by the participants). Rather than coding or sorting discourse into categories and then combining these categories in a progressively more abstract synthesis (as, e.g., in Glaser & Strauss's  grounded theory), discourse analysts focus on taking the discourse apart in multiple and microscopic ways to see what it consists of and how it is put together to accomplish different actions. It is here that the microscope metaphor comes into play. Third, discourse analysis attempts to elucidate the social functions and consequences of discourse. These activities draw upon the identification of patterns in the discourse (variability and consistency in structure and content) and require that hypothesized functions be supported by evidence in the discourse itself (cf. Potter & Wetherell, 1987). Analysis begins with the details of discourse and must be thoroughly grounded in discourse. As we said, an interruption is not an interruption unless it is so treated in the discourse of the participants.(f.9)
Note that discourse analysis does not involve only the analysis of style or structure at the expense of content (or vice versa), but an analysis of both and of the ways in which they work together to achieve particular functions. As Fairclough (1992) argues, "one cannot properly analyse content without simultaneously analysing form, because contents are always necessarily realized in forms, and different contents entail different forms, and vice versa. In brief, form is a part of content" (p. 194). Further, content itself is a problematic category in that it can refer to lexical items (e.g., "I'm sorry"), function (e.g., "apology") and so on at different analytical levels. Discourse analysts would treat this category as they do others, that is, not as a label for a set of specific items, but as a resource of speakers who work actively to construct various categories and to use them for a variety of purposes (e.g., description, evaluation).
Discourse analysis and everyday understanding. We are not arguing that ordinary speakers require the technical apparatus of discourse analysis to forward their projects. Clearly, if ordinary understanding could not be regularly achieved in everyday social interaction by ordinary speakers, social life as we know it would not be possible. Serious disruptions would occur if the phrase, "I'm sorry," were regularly taken only as a description of another's internal state and not also as an apology. What is going on is often obvious, at least at some level or in some sense that is adequate for ordinary interaction. But we do need special tools, the technical apparatus of discourse analysis, to enhance everyday understanding for purposes of systematic inquiry (e.g., into the ways in which description can serve to construct facts and make attributions [see Edwards & Potter, 1992]) and for technical applications (e.g., the analysis of black-box conversations after an airline crash).
More fundamentally, we require special tools to understand precisely why and how discourse works in the way that it does. The ability to understand discourse in the sense that we can recognize what is going on (not necessarily with awareness, but in the sense that we orient to it, i.e., respond and act accordingly) rarely enables us to identify precisely the particular features of content and structure that achieve a particular function, nor how they do so. The ability to ride a bicycle does not mean that we can describe precisely how we do it.
Language is inherently and necessarily ambiguous. Ambiguity affords possibilities for social life that would be absent if talk were always straightforward, directly to the point (Brown & Levinson, 1987). If all talk were in the unforgiving Gricean (Grice, 1975) mode,(f.10) it would destroy the delicate balance between competition and cooperation that is the universal and necessary hallmark of social life (Levi-Strauss, 1968). And it would severely challenge "face," the universal effort to maintain one's honour and reputation in the expressive social order where people live once matters of mere survival, the requirements of the practical social order, are satisfied (Harre, 1979). The indeterminacy or indexicality of ordinary language is not a weakness but a strength (Potter, 1996, p. 44). We now turn to some particular examples of the analytic process.
Interpretive Repertoires. The examples that we consider below are drawn from a previous analysis of interviews with women who had been raped by dates or acquaintances (Wood & Rennie, 1994).
(1) ... Hollywood rape stuff. Just clear cut, you know, stalked down a street, total stranger, dragged into an alley, raped, police and the whole stuff, or somebody break into your house and. Horrifying stories. But they all seemed like these women could never doubt that they were raped...so I thought, "Well, what a shitty experience, and terrifying, but they don't have that ambiguity, was I raped, was I not raped?" (Kim, 684-691)
In this excerpt, Kim describes rape in terms of the "Hollywood scenario." We can draw on the relatively broad analytic unit of the interpretive repertoire to understand the structure and function of this bit of discourse. Interpretive repertoires may be defined as
...building blocks speakers use for constructing versions of actions, cognitive processes and other phenomena. Any particular repertoire is constituted out of a restricted range of terms used in a specific stylistic and grammatical fashion. Commonly these terms are derived from one or more key metaphors and the presence of a repertoire will often be signalled by certain tropes or figures of speech (Wetherell & Potter, 1988, p. 172).
This definition is fairly typical. Repertoires are used selectively and flexibly and speakers may draw on one or more repertoire on any one occasion, depending on the function of the discourse (Wetherell & Potter, 1988). Repertoires are not simply a set of terms; the terms are organized and systematically related (see Potter, 1996, pp. 115-116). Nor are repertoires necessarily fixed in form or content; any particular repertoire could potentially appear in slightly different versions. Some repertoires, derived from formal or ritual sources, are likely to be canonical, relatively fixed. However, that does not preclude their creative use as a resource in everyday talk.
We must be careful to distinguish repertoires from other related concepts. Interpretive repertoires are sometimes referred to as cultural texts or discourses. Repertoires have some features in common with the notion of scripts (Schank & Abelson, 1977), except that repertoires are not necessarily ordered in time, like a script that tells people what to do in a restaurant, from entering to leaving it. In addition, the term script usually invokes a cognitive metaperspective that is at odds with the discursive approach (Edwards, 1994). In particular, repertoires, unlike some postulated cognitive entities, are not seen as having causal powers but are taken to be resources used by speakers for their purposes. Finally, a repertoire is not a theme in the sense of a common characteristic or distillate of a set of features of a particular discourse.
The analysis of repertoires is further distinguished by an emphasis on the function and consequences of the repertoire in the discourse. There are several general functions of repertoires; they are interpretive systems that can be used for formulating the nature of phenomena and that can be drawn on to characterize and evaluate actions, events and other phenomena (Potter & Wetherell, 1987, p. 138). Previous work has identified some specific uses of particular repertoires, for example, accounting for error in scientific work by drawing on repertoires embodying the empiricist version of the nature of science (see Potter & Wetherell, 1987, ch. 7).
We turn now to the repertoire of rape deployed in Excerpt 1. The Hollywood or "standard" rape repertoire includes not only the act of sexual intercourse without the consent of the victim. It also involves certain kinds of identifies or personas (an agent, and a patient, who are strangers to each other); a particular way of carrying out the act (brutal, violent); and a vocabulary of motives (the exercise of power and sexual satisfaction for the rapist, escaping or reducing injury for the victim). The repertoire also includes a particular sequence of events: the rapist's sudden appearance, the making of threats, the uttering of pleas by the victim, the sexual act and the disappearance of the rapist. This repertoire, then, is in the form of a narrative.
How does this repertoire function? In this excerpt and others, we see the woman attempting to match her own construction of events to that of the repertoire. A successful match would permit her to make sense of her experience and to account for her actions; if she can draw on the rape repertoire, she can justify to herself and to others her failure to resist, ascribe negative motives to the man, and so on. The problem in this particular case is that the match is unsuccessful -- the man was not a stranger, there was no real violence -- and the experience is ambiguous ("was I raped, was I not raped?"). Further evidence that the use of the Hollywood repertoire is not helpful is provided by the women's consideration of a different repertoire, the "date" repertoire.
(2) Um, I think the whole dating situation. We were on a date, it's hard to separate the rape from the date situation, it's not like it's a stranger in the dark alley, it's someone who you are on a date with and events progress.... (Kelly, 351-354)
(3) Yeah, I think it will [i.e., the rape will always influence her], for sure, I think I will always be suspicious because here was Dan, Mr. Golden Boy from next door far as most people were concerned, "Such a nice guy." It makes me realize it could be anybody. (Mary, 1300-1303)
The date repertoire also includes identities (acquaintances or friends; Golden/college boy; both parties are agents), shared motives (entertainment, companionship, sexual satisfaction, love), and a sequence of events that may include consensual sexual intercourse (and possibly subsequent regrets and an appraisal that the date "got out of hand"). But like the rape repertoire, the date repertoire is not helpful for making sense of the experience of the participants in this study -- there are too many disparate elements (e.g., the use of force, the general lack of consideration, the failure to respond to or even acknowledge the woman's refusals).
At this point, we would consider revising the hypothesized functions of these particular repertoires. The potential function of sense-making is not realized. Rather than drawing on the repertoires to name the experience, the women use them to identify what the experience was not. If the experience was not a rape, it must be a date, but this interpretation does not hold up when the experience is compared to the date repertoire. In this case, the participants provide evidence that the repertoires (resources) are tried out (by comparing their experience to the relevant repertoires), as hypothesized, but that they are found wanting (because the participants discard the interpretations suggested by the repertoires). There is also explicit evidence that the problem of interpreting the episode has not been solved.
(4) And you know, I saw the ad (for the study) two weeks ago, and I read it, and I'm going, "Well, let's go and see this person (interviewer). Let's find out whether it was an actual rape." (Ann, 500-506) [italics added]
The difficulty experienced by the women in using the repertoires is that to do so constructs their experience as deviant because the repertoires do not fit them. Note that this claim of difficulty does not entail a reference to some prior or hidden internal state; rather, the difficulty is in the text. That is, to entertain and discard various possibilities, to hesitate or avoid naming the experience are not "signs" of difficulty that exists elsewhere, but are themselves constitutive of the difficulty. The claim that formulating the nature of their experience is a critical issue for the women in the study is discussed in more detail in Wood and Rennie (1994).
There is additional discursive evidence for our claims about the repertoires and their functions. For example, "you know" in the first line of Excerpt 1 assumes that the interviewer is familiar with the Hollywood rape repertoire. Because the interviewer and the woman are strangers, such an assumption means that the woman is referring to a cultural repertoire, not to information exchanged in a previous encounter with the interviewer or to shared experience.
There is specific evidence that there are two different repertoires: 1) inconsistencies between the two are noticeable to both analysts and participants; 2) the repertoires tend to appear in separate passages; and 3) when the repertoires are used together, participants orient to the potential inconsistencies (Wetherell & Potter, 1988). A subtle example of the third sort of evidence can be seen in Excerpt 2. The claim that "it's hard to separate the rape from the date situation," together with other statements by that participant, rests on the assumption that they can be separated, and the use of "situation" indexes the date repertoire. A date is just a date but a "date situation" (as in the message over the police radio, "we've got a hostage situation here") invokes not a singular, concrete episode but a class of events, a repertoire. And the "you know" linked to the "Let's find out whether it was an actual rape" in Excerpt 4 assumes that the interviewer also recognizes the inconsistency between the two repertoires and the difficulties in interpretation that they pose. The evidence for all of these assertions is in the text. We can also draw on our own cultural knowledge of repertoires concerning sexual relations (a knowledge that is assumed by participants, as noted above). Such knowledge is on its own insufficient to warrant analyses of repertoires. But we can look to other work in which the two repertoires are well documented. For example, Coates, Bavelas and Gibson (1994) have shown in an analysis of Canadian court decisions that there are interpretive repertoires for stranger rape and consensual sex, but "virtually no accurate vocabulary or narrative structure for cases in which the assailant is not a stranger to the victim" (p. 189).
Coates et al. found that for cases that do not fit the stranger rape repertoire, judges assimilate the assault to the consensual sex or date repertoire. For example, they characterize the assault as a romantic story that portrays the offender in a positive way, that tends to avoid the attribution of agency, and that treats the two parties as one discursive unit (e.g., "the couple," "they"), and in general they employ a vocabulary that is suited to consensual sex (e.g., "invited," "fondled"). As Coates et al. point out, these anomalies are not "oddities in the judgements...but they are well integrated into the texts in which they occur" (p. 197). Why do the judges resort to the inappropriate repertoire while our participants reject it? There are several reasons: the judges are not describing their own experience; their accounts are produced for a very different audience; and their position requires them to produce a coherent account of events, particularly one that will justify the sentence that they impose.
The Coates et al. analysis suggests another reason why neither the rape nor date repertoires work for our participants. In the judges' accounts, the date repertoire includes elements of pleasure, eroticism and affection. These elements are generally missing from our participants' accounts; they are also the elements that are most at odds with the participants' experience. Grounding an interpretation in the discourse does not always mean identifying the presence of a particular item; an analytic strategy that is often highly useful is to identify what is missing.
The sort of analysis we have just described, that is, one that draws on other discourses as well as the particular discourse in hand, is a type of "intertextual" analysis (Fairclough, 1992). Put simply, this means that we are still looking closely at the discourse, endeavouring to ground our claims, but rather than doing so only through a consideration of the immediate context (that within the discourse at hand, e.g., the utterances adjacent to those being analyzed at the moment), we also consider the external context (i.e., other orders of discourse). Fairclough argues that "intertextual analysis crucially mediates the connection between language and social context" (p. 195); it is thus particularly useful for the analysis of those sorts of issues that are related to broad cultural practices. This does not mean that in the present case we are making claims about the general operation of the rape and date repertoires in North American culture. Such claims would require an examination of those repertoires in the context of an analysis of relations between men and women.
Speech Acts. Speech acts are smaller chunks of talk than repertoires and require a more fine-grained level of analysis. They are usually single utterances (sentences) that accomplish a specific function. For example, "I am sorry" (usually) does an apology; "Can you pass the salt?" is usually interpreted as a request (despite its literal meaning as a question about ability). As we noted above, one does not always require special methods to see what a speaker is doing with words, and the identification of speech acts in particular is a routine everyday accomplishment. However, systematic analysis requires special methods even in these cases. As well, speakers and hearers quite often disagree as to the precise status of an utterance, given the inherent ambiguities of language. Witness the common marital disagreement about whether the phrase "The dishes are dirty" is a description or a complaint.
Concepts such as social comparison and attribution that have been prominent in previous work in social psychology can be treated as speech acts, with some important provisos. For example, attributions are viewed by discourse analysts as social (discursive) actions rather than as cognitive processes (see Edwards & Potter, 1993). We can apply the idea of social comparison to Excerpt 1. Kim compares herself twice to other women in a therapy group with respect to the relative certainty with which they identify their experience: "...these women could never doubt that they were raped" and "...they don't have that ambiguity." What function does this comparison serve? The function can be identified in the other, embedded comparison that Kim makes (twice) in this excerpt: "Horrifying stories. But...could never doubt that they were raped" and "Well, what a shitty experience, and terrifying, but they don't have that ambiguity." The contrast (via the conjunction "but") between the other women's dreadful experience and their certainty about what happened to them can justify Kim's claim that her experience was also dreadful, even though it did not match theirs in violence, because she faces the problem of defining what it was that happened to her ("was I raped, was I not"). The comparisons also function to provide a basis for sympathy, for accounting for her behaviour during the event, for her reactions to it, and for her participation in the study.
(5) The whole thing I was thinking the whole time was it's my fault because I shouldn't have been, my parents always told me you don't go out with a boy alone and don't go and do this or do that. (Barb, 70-74)
The speech act is self-blaming. Such attributions were not uncommon in the data. We can gain an even better appreciation of the workings of attributions by drawing on units that are smaller than the speech act, that is, on grammatical features.(f.11)
Grammatical features. Grammar figures in discursive psychology in two ways. We can speak about the grammar of social life in that social actions have syntax and semantics. In addition, we can speak about the function of grammar in discourse, that is, the way in which psychological and social-psychological matters are grammatically encoded. The grammatically encoded is the bedrock of social life. We have space to discuss only two examples.(f.12)
(6) I went through a big period of, you know, I shouldn't have worn that top, you know, I shouldn't have let him kiss me, I shouldn't have this, you know, this, I shouldn't have done that. (Mary, 111-115)
(7) I shouldn't have gone into that house that night. I shouldn't have, you know, 2:30 in the morning, I should have known the guy was drunk. (Ann, 370-373)
In Excerpts 5, 6 and 7, we see the modal "should," which works to create speech acts of obligation or propriety. Participants are thus saying not only that they could have acted or thought differently but also that there are standards of conduct and knowledge which they violated. In so doing, they are implicitly doing attributions of self-blame for the rape.
(8) Sometimes I would think that I was kind of stupid doing what I did, like not trying to get away more, or stuff like that. (Leslie, 290-292)
(9) If I hadn't crawled out my window, if I hadn't been so friendly to him, if I had worked harder in getting away, you know, in a sense, yeah I did. (Leslie, 612-615) (when asked if she blames herself)
Leslie also blames herself, but the blame is for her stupidity in acting as she did, not for moral errors. The analysis reveals the nature and extent of the self-blame, not just whether participants attribute blame to themselves. And these attributions vary both within and between participants. For example, the women speaking in Excerpts 5, 6 and 7 use modals of obligation, while Leslie in Excerpts 8 and 9 entertains alternatives (e.g., the conditional "if").
Modals are characteristic