Susan A. Speer

`Natural' and `contrived' data: a sustainable distinction?

One of the most consistent and (for some) troubling findings of mainstream social scientific research, is that the context in which a research project takes place has a significant effect on the behaviour of respondents and the overall quality of the data (Banyard and Hunt, 2000; Schuman and Presser, 1996). Researchers working within both positivist and interpretative paradigms, for example, worry that their data collection techniques are not wholly neutral or objective instruments, and that there are numerous potential sources of ‘bias’ (Hammersley and Gomm, 1997) which may lead to invalid and erroneous results (‘researcher effects’, ‘reactivity’, ‘context effects’, ‘observer effects’, and so on). Because they represent a source of error, these effects are treated as problems which must be overcome by improving the research instrument or by taking ‘proper methodological precautions’ (p. 11). Books on interview methodology, for example, often advise researchers to minimize the ‘intrusive’ effects of a variety of interactional and contextual features. Some frequently recommended ‘techniques’ include taking time to build up a rapport with respondents, allowing for acclimatization periods (such that the researcher and his or her tools can ‘blend’ with the setting), dressing in a way that is not intimidating, learning about techniques to broach sensitive topics, being non-directive, and so on (see Judd et al., 1991). Thus, the primary aim is to eliminate extraneous, researchinduced ‘contaminants’ and uncover some more ‘pristine’ reality. One of the distinguishing features of discursive and conversation analytic (CA) approaches, by contrast, is their emphasis on the action orientation of talk, and the local, or ‘endogenous’ production of context (Potter and Wetherell, 1987; Schegloff, 1997a; Hutchby and Wooffitt, 1998). Since these approaches treat respondents, not as passive containers of knowledge, but as active participants within the research process who construct, rather than report on reality, ‘bias’ is regarded as both unavoidable and pervasive. Research contexts are thoroughly social, interactional occasions, and it is for this reason that, as Holstein D E B AT E 511

Reconsidering the Concept of Hegemonic Masculinity: Discursive Psychology, Conversation Analysis and Participants’ Orientations

This article provides a critical review of Wetherell and Edley’s (1999) discursive reformulation of the concept of ‘hegemonic masculinity’. While I retain some familiar features from Wetherell and Edley’s approach, I develop a discursive perspective that is located more firmly in the technical, conversation analytic tradition - as outlined in the recent exchange between Schegloff (1997, 1998) and Wetherell (1998). In particular, I argue that previous research is based on the assumption that we need to venture further than the limits of the text to explain why participants say what they do, and go beyond participants’ orientations to be able to say anything politically effective. Using data from two semi- structured interviews with men in their early 20s, I explore how participants construct masculinity and situate themselves (and others) in relation to those constructions. This involves an analysis that is more attentive to participant orientations and gendered category membership than that used in the analysis of masculinity so far. The article concludes with a discussion of the implications of this approach for feminist psychology.

What can conversation analysis contribute to feminist methodology? Putting reflexivity into practice:

This article uses conversation analysis to explore an issue central to the design and delivery of feminist research: the relationship between researcher and researched, and specifically, the impact of the former on the latter. A guiding principle of much feminist research is that it should be respondent-centred, allowing participants to set the agenda and define what is important in their own terms. Though not advocated as an explicitly feminist method, one technique deemed to be ideally suited to this end is the use of prompts as stimulus materials. In this article, I revisit data from my own research in which picture prompts were used to derive gender talk. Rather than treat prompts as facilitators of talk in which the respondents set the priorities, I demonstrate how the activity of showing a prompt itself requires work on the part of the moderator. I argue that even where the researcher tries to minimize her impact on the data collection process — as in this case through the use of prompts — that she is still influential and the data is thereby always an interactional product. Although many feminists acknowledge this, and advocate the importance of a reflexive orientation to our data collection practices, I suggest that most feminists do not, as yet, possess the analytic skills to do this reflexivity well. I consider the implications of this analysis for the way feminists and other researchers derive and analyse gender talk, and conceive of the relationship between the researcher and those researched.

Gatekeeping gender: some features of the use of hypothetical questions in the psychiatric assessment of transsexual patients

Psychiatrists, like other medical professionals with a diagnosing or prescribing role, control access to a range of forms of treatment, medication and service that their patient, or their patient’s carer, may want access to. In this article, we explore psychiatrist-patient interactions in the distinctive institutional site of a UK NHS Gender Identity Clinic, where the psychiatrist’s gatekeeping role is renowned. We focus on some interactional features of the psychiatrist’s gatekeeping role as it gets played out and oriented to in a specific class of question that they ask their patients. This class of questions involves the psychiatrist putting to the patient a possible future hypothetical scenario where the patient’s treatment is withdrawn. We show how these hypothetical questions function in the psychiatric assessment of transsexuals, and how the psychiatrist’s institutional, gatekeeping role is made manifest both in the design of the hypothetical question and in the response that is elicited from the patient. We end by considering the extent to which hypothetical questions may be deemed a ‘useful’ or a ‘successful’ strategy in the psychiatric assessment of transsexual patients.

Sexist talk: Gender categories, participants’ orientations and irony

This paper uses a discursive social psychological approach to develop and extend what we know about the constitution of sexist talk. Using data from a variety of sources where the topic under discussion is gender and leisure, I examine actual instances of sexism in action. Specifically, I examine the ways in which participants use arguments about possible injury to justify women’s non–participation in certain sport and leisure activities. I identify three resources that participants use to sustain sexist accounts, and bolster their arguments against attack. A fourth resource problematises just what counts as sexist talk, and provides evidence for the ways in which participants themselves can exploit sexist arguments in an ironic fashion to expose and challenge sexist assumptions. I end by considering the implications of this approach for future work on sexist talk, and discursive work on the relationship between gender and language more generally.

Feminism and Conversation Analysis: An Oxymoron?

Ever since the ‘turn to language’, feminist psychologists have been interested in pursuing social constructionist, post-structuralist and discourse analytic ideas in their analyses. However, there has been a corresponding reluctance to adopt a more technical, conversation analytic (CA) approach to feminist issues. This is primarily because of misgivings about CA’s political efficaciousness, practical utility and compatibility with feminist principles. One of the main objections, for example, has been that conversation analysis – as a ‘micro’ approach to interaction that concentrates exclusively on the participant’s perspective – cannot account for the ways in which ‘wider, macro power structures’ exert a determining effect on action. Neglecting this ‘top-down’ feature of society and context, some feminists argue, leads to apolitical and reductionist forms of analysis. Moreover, CA offers no way of commenting on patriarchal power or the oppressive constraints on women’s lives. As Kitzinger and Frith point out, ‘conversation analysis is often viewed as nit picking, obsessively concerned with details, and as unable to see beyond the “micro” level of the 0.2-second pause’ (1999: 311). Elsewhere, Frith highlights the limitations of an approach which relies on participants explicitly attending to the topic under investigation. It is questionable, says Frith, whether all the dimensions relevant to a piece of interaction – such as participants’ ‘shared whiteness’ – will be ‘interactionally displayed’ (1998: 535). Indeed, criticisms such as these have led many feminist psychologists to suggest that we need to go beyond participants’ practices to be able to say anything politically effective (see Wetherell, 1998). The assumption that we need to subsidize ‘micro’ analyses with features extrinsic

The Interactional Organization of the Gender Attribution Process

This article illustrates some interactional properties of the ‘gender attribution process’ - that is, the methodical procedures through which members come to identify others as male or female (Kessler and McKenna, 1978). Drawing on data from interviews and group discussions where respondents were asked to comment on a series of pictures containing gendered images, I explore instances where members have trouble identifying the gender of the person in the picture. I analyse the procedures through which they manage that trouble, and collectively assign a coherent gender identity to the person in the image - thus re-establishing what Garfinkel (1967) has termed the ‘natural attitude’ toward ‘normally sexed persons’. Although, in most instances, an ‘uncertain’ or ‘incorrect’ gender attribution is treated as an accountable phenomenon which requires identity work on the part of the speaker, in some cases, members’ displayed ‘doubt’ or failure to attribute, itself becomes an interactional resource, used as a relatively ‘safe’ form of negative identity attribution.

The Interactional Organization of Self-Praise: Epistemics, Preference Organization, and Implications for Identity Research.

This article contributes to a social psychological understanding of identity by identifying some features of the interactional organization of self-praise. Early conversation analytic work on the epistemics of self-assessment and constraints against self-praise has shown that praising oneself is an interactionally delicate matter that may leave one vulnerable to “unfavorable character assessment” or accusations of bragging (Pomerantz 1978:89). Drawing on data examples from a range of settings, this article develops Pomerantz’s work and examines the role of reported third-party compliments (e.g., “she . . . said ‘you look really lovely’”) in objectifying self-praise. Analyzing instances in which speakers initiate repair on their self-descriptions in favor of reported third-party compliments, I provide evidence of practices suggesting a norm against direct self-praise and an interactional preference for embedding positive self-descriptions within a third-party attribution. I consider the implications of these analyses for a social psychological understanding of identity and its measurement.

Participants Orientations, Ideology, and the Ontological Status of Hegemonic Masculinity: A Rejoinder to Nigel Edley

Nigel Edley’s response to my article addresses a number of vigorously debated issues. Commenting on the distinction between critical discourse analysis (CDA) and conversation analysis (CA), Edley claims that there is ‘even less of a difference between our approaches than Speer seems to imagine’ (2001: 136). Interestingly, he argues that my work actually helps advance his (and presumably Wetherell’s) CDA argument that there are ‘some significant problems attending a conversation analytic approach to studying men and masculinity’ (2001: 136). I am delighted that Edley chooses to illustrate his argument with (albeit strategic)1 reference to CA notions, and by conducting further analysis of my data (and without resorting to what purportedly lies beyond, behind or beneath it). This is, in itself, entirely consistent with and testament to the validity of a CA-aligned approach. However, his resultant claims are based on a number of misinterpretations and misreadings of my approach and the DP/CA field, eight of which I address here. These centre broadly on and around the relationship between participants’ orientations, ideology and the ontological status of hegemonic masculinity. First, Edley asserts that my ‘main argument’ is that ‘an adequate discursive psychology need not, and perhaps should not’ go beyond the limits of the text to explain why participants say what they do (2001: 136; my emphasis). However, I would not want to legislate about analysis in this way. To do so would be to reinforce precisely the form of intellectual imperialism that many discourse and conversation analysts (myself included) seek to counter. Second, Edley misrepresents my approach, and the perspective of others who adopt a broadly DP/CA position. For example, he directs his critique at CA, and not at CA-aligned DP – which is the approach I use in my article. The purpose of this article was to set out some of the advantages of using a more CA-aligned DP (associated particularly with the recent work of Derek Edwards2 and Jonathan Potter). This work shares some features in common with Wetherell’s discursive

Transcending the `natural'/`contrived' distinction: a rejoinder to ten Have, Lynch and Potter

related perspectives: conversation analysis (CA) (ten Have), discursive psychology (DP) (Potter) and ethnomethodology (Lynch). It is no easy task to respond in a way which remains faithful to the subtleties of each position. However, there are some commonalities across the responses as well as some areas of disagreement. Broadly, while each author acknowledges the virtues of discussing the problems that I have claimed are associated with the ‘natural’/‘contrived’ distinction, they differ in terms of the strategies each employs to overcome or transcend those problems. In this brief rejoinder, I will attempt to tease apart some of the issues their arguments raise, with a view to finding a productive way forward. Paul ten Have agrees with my overall argument ‘that the distinction between “natural” and “contrived” data should not be absolutized’ (this issue: 527). Expanding on the subtleties of his position, he notes that ‘whether one prefers relatively “natural” or more “experimental” data should depend on what kinds of phenomena one wants to study and how one intends to study them’ (this issue: 527). He says: