Does that mean there is a discourse analytic answer to any psychological question?
No. One of the mistakes that people sometimes make when they are starting discourse work is to treat discourse analysis as a method that can simply be plugged in to a predefined question: ‘I am interested in the factors that cause people to smoke, should I use an observational study, an experimental simulation, or discourse analysis?’ What this misses is that, first, discourse analysis is not just a method but is a whole perspective on social life and its research, and, second, that all methods involve a range of theoretical assumptions.
Traditional psychology has often been concerned with factors and outcomes, and these ideas are thoroughly enmeshed in thinking about experimentation and questionnaire design. The logic of discourse analysis is a rhetorical and normative one. Any rhetorical move can have a rhetorical counter move — categorization, say, can be countered by particularization (Billig, 1985). Effectiveness is not guaranteed as it should be a with a causal process; for a norm is not a mechanical template. Norms are oriented to, but they are also regularly deviated from, although such deviations may themselves be marked by a range of accounts or sanctions. One of the skills involved in discourse analytic work is in formulating questions that are theoretically coherent and analytically manageable. Simply importing a question cold from a traditional psychological framework is asking for trouble.
Another important difference in the formulation of questions is that traditional psychologists have become devoted to hypothetico deductivism, where quality research is seen to hang on a well formed question or precisely specified hypothesis. Without endorsing a naive, assumption free inductivism, discourse researchers have often found it productive to collect and explore a set of materials — interview transcripts or natural records of some kind — without being hampered by the need to start from a specific hypothesis. Indeed, the devotion to a fully formulated prior hypothesis has probably been one of the reasons why psychologists have been so reluctant to study records of natural interaction such as everyday conversations between familiars or interactions in the workplace.
What are interpretative repertoires?
I noted above that discourse analysis is concerned both with the organization of texts and talk in practices, and with the discursive resources that those practices draw on. The notion of interpretative repertoires is intended primarily to help in specifying and analysing interpretative resources. Interpretative repertories are systematically related sets of terms, often used with stylistic and grammatical coherence, and often organized around one or more central metaphors. They are historically developed and make up an important part of the common sense of a culture; although some may be specific to certain institutional domains.
The idea of an interpretative repertoire is intended to accommodate to the twin considerations that there are resources available with an off-the-shelf character that can be used in a range of different settings to do particular tasks, and that these resources have a more bespoke flexibility which allows them to be selectively drawn on and reworked according to the setting. It is the attempt to accommodate to this flexible, local use that distinguishes interpretative repertoires from the more Foucaultian notion of discourses (Parker, 1992; Potter, et al., 1990). Participants will often draw on a number of different repertoires, flitting between them as they construct the sense of a particular phenomenon, or as they perform different actions. Billig (1992) refers to this as the kaleidoscope of common sense.
The classic research using this notion is Gilbert and Mulkay’s (1984) study of scientists discourse, which records the way scientists use one interpretative repertoire in their formal writing for justifying facts, and another in their informal talk when accounting for why competing scientists were in error. More recently, the notion has been developed in a number of studies with a more social psychological focus (e.g. Marshall and Raabe, 1993; Potter and Reicher, 1987; Wetherell et al, 1987; Wetherell and Potter, 1992). The overall analytic goal in these studies is the identification of repertories and the explication of the practices they are part of. For methodological discussions relating to repertoire analysis see in particular Coyle (1995), Potter and Wetherell (1987, ch. 7, forthcoming) and Wetherell and Potter (1992, ch. 4) and Wooffitt (1992a).
Although the notion of interpretative repertories has proved analytically fruitful, it has some limitations. For example, it is much more difficult to make clear and consistent judgements about the boundaries of particular repertories outside of constrained institutional settings such as science discourse. Another problem is that the generality of the repertoire notion may obscure local interactional business done by particular discourse forms (see Potter, forthcoming; Wooffitt, 1992b).
Is repertoire analysis the main task of discourse analysis?
No. Although it has been one important development, it is increasingly supplemented or replaced by studies of the way specific actions are accomplished, or of the devices and procedures through which factual versions are constructed. These are studies asking the following sorts of questions. How is a blaming achieved? How is a particular version of the world made to seem solid and unproblematic? How are social categories constructed and managed in practice? Such questions require an understanding of what Billig (1987) calls the witcraft of rhetoric: the detailed, contextually sensitive manner in which versions are constructed and arguments deployed as well as an appreciation of the conversational organizations in which such procedures are embedded. Indeed, it is here that the study of discourse shades both into the study of rhetoric and work on conversation analysis.
There is less written on methodological aspects of this style of discourse work. However, Potter and Wetherell (1994) describe the methodological decisions and analytic practices in one study of this kind, trying to show how specific conclusions were arrived at. Wooffitt (1992a,b) also provides a clear an helpful introduction to such analysis.
How are discourse and conversation analysis similar and different?
Conversation analysis has developed from pioneering work of Sacks, Schegloff and Jefferson to provide an elaborate and systematic account of talk-in-interaction. It is sometimes crudely stereotyped by psychologists as the study of the rules of turn taking in trivial conversations. However, the point of conversation analysis is to explicate the fundamental sense that interaction has for its participants. It is a fast growing and notably cumulative field which has highlighted major deficiencies in the speech act approaches that psychologists often look to for an account of language practices (Levinson, 1983; Schegloff, 1988) as well as providing striking analyses of topics as diverse as intersubjectivity (Schegloff, 1992), public speaking (Atkinson, 1984) and courtroom interaction (Drew, 1992).
Conversation analysis is relevant to discourse analysis in two ways. First, it provides a powerful and general understanding of interaction that has the potential to illuminate a wide range of research questions. After all, much human interaction is performed through conversation and to understand many of the more psychological and social phenomena that discourse analysts are interested in it is necessary to understand how they emerge out of the general pragmatics of conversation. This involves being mindful of basic features such as turn organization, pairing of actions, normative ranking of alternative turns, as well as considering the findings of many studies showing the delicate way in which actions are embedded in sequences of discourse. A basic practical understanding of CA is a prerequisite for producing high class DA work. The best and most accessible current introduction is undoubtedly Nofsinger (1991), although Heritage (1984) has an excellent chapter situating CA in its ethnomethodological context (see also Heritage, 1988).
There are specific areas where discourse and conversation analysis come together. Like discourse analysts, conversation analysts have paid a lot of attention to the way versions are constructed and actions performed. One of the points argued by Sacks (1992) is that interaction is not merely organized in its general forms, but is also organized in its particulars. Any level of detail in talk — hesitations, repairs, pauses — can be crucial for a piece of interaction; indeed, much of the business of interaction may be happening in the details. Workers in this tradition have also developed a sophisticated critique of cognitivist approaches to interaction, and at the same time attempted to identify grounds for making inferences about cognitive entities (Coulter, 1989; Pomerantz, 1990/91). CA and DA are starting to develop complementary alternatives to cognitivist theory.
Making conversational interaction a topic of study is also important for methodological reasons. CA highlights a symmetry between the position of the participant and analyst in a conversation. In conversations speaker provide their own ongoing interpretation of what is going on. In most cases, a turn of talk is based upon and displays some sort of analysis of the sense of the previous turn of talk. A turn may be responded to as a question, a criticism, an invitation and so on — and in responding to it in this way the speaker displays their understanding. And if the displayed understanding is faulty various repair mechanisms can come into play in the next turn to sort things out. What this provides, then, is a way for the analyst to use the participants’ own, situated, analyses to help check the adequacy of their own analysts’ claims. The point is not that the analyst is forced to take such displayed understanding at face value, nor that interaction is always well oiled and explicit; nevertheless, it is one important resource for understanding interaction.
This ability to use participants’ understandings to build up the analysts’ account distinguishes this work from other types of constructionist research which have focused on texts or documents, or considered talk abstracted from its conversational context. While discourse analysts have often worked with interview material, conversation analysts have worked almost exclusively with natural occurring records of interaction collected with tape recorders or video and transcribed to a high degree of detail.
There are a number of discussions focused on CA methods. Atkinson and Heritage (1984b), Heritage (forthcoming), Wooffitt (1990) and Wootton (1989) cover a range of practical issues, Psathas (1990, 1995) attempts to characterise the analytic mentality that is involved in this work, and Drew (in press) introduces CA in a way that is designed specifically for psychologists.