Moving forward by doing analysis

Given that I have been working for several years on advancing an approach to studying membership categorization in a sequentially sensitive manner, I am broadly appreciative of Stokoe’s contributions, both in the article to which I am responding and in her previous work. In the discussion that follows, I offer some comments on a range of points Stokoe raises with respect to work on membership categories based on interactional materials. I then attempt to locate my own recent work with respect to the matters Stokoe raises. Finally, I conclude by arguing that the crucial feature of ‘moving forward with membership categorization analysis’ should be a commitment to doing analysis (without regard to the way in which the analytic approach itself is categorized), and that Stokoe makes an invaluable contribution in this regard.

As a first point in response to Stokoe, I would argue that she somewhat underplays the degree to which recent conversation analytic work has attended to categorial matters. Stokoe acknowledges that ‘a focus on membership categorization is currently enjoying something of a renaissance within ‘‘the discipline’’ of CA itself’ (p. 278). However, she goes on to frame her current contribution as a challenge to arguments such as Raymond and Heritage’s assertion that ‘establishing the mechanisms by which a specific identity is made relevant and consequential in any particular episode of interaction has remained . . . elusive (Raymond and Heritage, 2006: 677)’ (p. 279). While I agree with Stokoe that work in this area has historically been rare, and that this is related to assumptions about the ‘capturability’ of categorial phenomena, a number of studies have made contributions in this regard. In fact, examples can be found in some of the studies Stokoe cites in describing the sort of argument she is challenging. For example, Hansen (2005) demonstrates how speakers use ethnic categories as practical resources in dealing with interactional contingencies arising at particular sequential positions in the course of a contentious discussion. In addition, Raymond and Heritage (2006) provide an analysis of the ways in which speakers’ orientations to the relevance of their membership in a particular category (in this case, ‘grandparent’) are recurrently displayed through the deployment of practices for managing epistemic authority and subordination at different sequential positions in assessment sequences. My own work (e.g. Whitehead, 2009, 2011a, 2011b; Whitehead and Lerner, 2009) has been similarly engaged in examining circumstances in which the use of membership categorization devices (MCDs) is a consequential feature of unfolding sequences of action, as has that of a number of other authors, including Hopper and LeBaron (1998), Kitzinger (e.g. 2000, 2005a, 2005b), Lerner et al. (forthcoming), Schegloff (e.g. 2007) and Stokoe herself (e.g. 2009, 2010). Although this latest contribution by Stokoe more explicitly describes a systematic programme and set of resources for producing this type of research, these previous contributions provide additional demonstrations of some ways in which such work might proceed.

A second point concerns Stokoe’s discussion (see p. 279) of how issues of ‘capturability’ have been used as a basis for not systematically examining categorial phenomena, and in particular for not examining categorial phenomena concurrently with sequential aspects of talk. This discussion appears to implicitly invoke a distinction between ‘naturally occurring’ talk-in-interaction and talk that is ‘got up’ for research purposes. That is, the supposed lack of systematic ‘capturability’ of categorial phenomena that Stokoe describes has been used as an argument for the advantages of (researcher-generated) interviews over naturally occurring interactions, as a way of ensuring categorial content that is relevant to the researcher’s agenda ¹ (also see Stokoe, 2009). Thus, although she does not explicitly say so, Stokoe appears to be making ‘a case for the systematic analysis of membership categories and related phenomena . . . show[ing] how to track categorial concerns in the same way that CA has pursued sequential practices’ (p. 279) by drawing in particular on the types of naturally occurring materials (be they talk or texts) that conversation analysts have traditionally preferred to work with. ²

While the merits of distinguishing between ‘natural’ and ‘contrived’ data have been discussed at length elsewhere – see, for example Speer (2002) and responses by Lynch (2002), Potter (2002), and ten Have (2002) – and I do not wish to recapitulate this discussion, I would argue that attending to categorial and sequential phenomena concurrently does not necessarily require the use of ‘naturally occurring’ data. Research interviews, focus groups and the like can be ‘naturalized’ (see Potter, 2002; Speer, 2002) by analysing them as situated interactions in their own right, thereby opening up possibilities for doing the kind of analysis Stokoe describes regardless of whether the data is viewed as ‘natural’ or ‘contrived’, or whether one even buys into such a distinction. An example along these lines can be found in Grancea’s (2010) examination of the ways in which ethnic categorizations are deployed and resisted at particular points in the course of complaint sequences produced during research focus groups.

I would like to be clear, however, that I am not disputing the benefits that can be gained by examining how categorial phenomena are realized in interactions in which particular categories are not pre-specified as discussion topics in accordance with a particular research agenda (see Whitehead, 2011a, 2011b, for discussion and demonstrations of some of these benefits; also see Whitehead and Lerner, 2009). Instead, I am arguing that the possibility of also producing valuable findings by examining sequential and categorial phenomena concurrently in researcher-generated data should not be foreclosed prior to an empirically based assessment of their utility (or lack thereof) for examining particular phenomena.

A third point I wish to raise concerns Stokoe’s claim (p. 282) that:

Schegloff (2005) himself appears to rely on analysts’ categories in his study of a teenage girl’s ‘whining’; that is, the girl, Virginia, can be seen to be whining because it is a ‘childish’ attenuation of ‘crying’ and she is a ‘child’ (p. 469). Although Virginia is categorized as a child by her mother and brother (which she resists!), Schegloff’s specific claims about her ‘whining’ rest on his categorization of her.

In my view, this claim mischaracterizes Schegloff (2005) in two ways. First, Schegloff’s characterization of Virginia’s activity as ‘whining’ is based not solely on her incumbency in the category ‘child’, as Stokoe seems to suggest – instead, it is also based on a number of features of its production, including its prosody, its sequential placement, the action it appears to implement, and the common features it shares with candidate instances of a similar practice. Second, as Stokoe notes, Virginia is categorized as a child by her mother and brother, thereby demonstrating that, for these participants, Virginia can be relevantly categorized as such. Schegloff’s categorization of her as a child is thus based on evidence internal to the data (both how she acts and how she is treated and formulated), rather than constituting an imposition of ‘his [analyst’s] categorization of her’. There still remains the matter of Virginia having resisted being categorized by her mother and brother as a child, which raises the question of whether analysts can justifiably use as evidence categorizations that are both produced and resisted by participants of an interaction. Setting aside the importance of such contestations as an object of analysis in their own right, I would argue that, at the very least, using a resisted categorization internal to the data as evidence for an analytic claim does not constitute the type of analyst-imposed categorization Stokoe attributes to Schegloff.

However, the broader issue that Stokoe raises in this regard, namely ‘how far can one claim the relevance of categorial phenomena that are not formulated explicitly and unambiguously by speakers?’ (p. 282), is a crucial one. While Stokoe’s points regarding the importance of the deniability of categorial practices as a participants’ resource provide a solution that is grounded in how these practices matter for participants rather than analysts, there remains the analyst’s problem of determining whether, in particularly inexplicit cases, participants are consequentially oriented to a particular practice or utterance as (even potentially) category-relevant. This can lead to the temptation, as Stokoe notes, ‘to try to unpack what is apparently unsaid by members and produce an analysis of their subtle categorization work’ (p. 282). This is where the conversation analytic approach of working with collections of a candidate phenomenon can be useful. While it may not always be possible to settle questions arising from highly inexplicit practices based on a single case, it may be possible to marshal as evidence other cases that share important features in common with the questionable case, while being somewhat more explicit with respect to participants’ orientations to the categorial feature(s) of interest. Thus, although collections do not guarantee a definitive solution to difficulties of this sort, they do constitute an additional and potentially valuable resource in the analyst’s tool kit in instances in which the analytic aim is to explicate how practices are employed in a general sense (as opposed to explicating a particular case). It is also important to recognize, however, that it may not always be possible to make strong claims. Some claims will have to be weak if the data do not provide sufficient evidence for strong claims, but not being able to make a strong claim is no reason to make no claim at all.

Notwithstanding the importance of the issues I have taken up thus far, the main topic and contributions of Stokoe’s article relate to the debates and divergences between CA and MCA, and to the possibility of ‘moving forward with MCA’. My engagement with Stokoe’s (and others’) discussion of these matters has provoked me to attempt to characterize how my own work is located with respect to them. Two recently published articles may provide instructive examples in this respect. I embarked on the first of these studies (Whitehead, 2009) with thoroughly categorial concerns: I was interested in the social organization of race; I sought out a data source (‘race training’ workshops) in which I knew I would find people talking about race and, moreover, being accountable for doing so in ‘acceptable’ ways; and I approached the analysis of my materials by beginning with those places in which racial categories were demonstrably consequential for the unfolding interaction. The result was an analysis of a set of practices – practices that did not seem to be setting specific at the time, and turned out empirically to be quite general – through which speakers displayed an orientation to their conduct as being shaped by the contingencies of being (treatable as) a member of a particular (racial) category in the course of producing descriptions of the actions of members of other (racial) categories. However, in pursuing these concerns and producing these analyses, I could not avoid attending to the actions to which these practices were contributing, and the positioning of these actions within broader sequences and contexts of action. This study thus began and ended with a set of categorial concerns, but I was compelled (by the empirical details of the materials on which it was based) to incorporate a sensitivity to sequential matters because the surfacing of category-related matters was situated in – entangled in – those details.

The second, somewhat contrasting, example is reported in Whitehead and Lerner (2009). Again, this research was initiated with an explicit interest in categorial matters. We trawled through hundreds of hours of existing data searching for explicit references to racial categories, in the hopes of developing a collection that would tell us something about the social organization of race. As we engaged with the resulting collection, however, it became apparent that these references to racial categories were recurrently emerging in specific sequential positions, and a central focus of the resulting analysis was on how these particular sequential positions provided systematic ‘homes’ for the surfacing of particular racial categories (especially those that would ordinarily remain implicitly taken-for-granted or ‘invisible’). This study thus ended up primarily examining how sequences may be constitutive features of the realization of categorial forms of organization, even though it was initiated from a firmly categorial set of concerns.

What these two examples suggest to me is that, while a pre-existing commitment to studying membership categories may shape the way a data set is constituted, the details of the data can (and, I would argue, should) be a primary determinant of the character of the final product – whether it be focused more squarely on categorial or sequential matters, or (more likely, based on the recurrently intimate relationship between them ‘in the wild’) on a combination of the two. As a result, although I have been trained primarily in CA, under the direction of people who would most likely be viewed as representatives of a CA perspective, I am not overly concerned about whether my contributions are characterized as exemplifying a CA or MCA approach, or both. I am concerned, however, with whether my work is faithful to the empirical details of the data on which it is based, and whether it makes a contribution to the understanding of social organization broadly, and social categories and talk-in-interaction in particular. I would thus argue that, regardless of where our research interests come from or how the data that we examine is selected, the crucial issue is whether, once we have the data in front of us, we are bound by its integrity (Schegloff, 2005). Indeed, it is not clear to me how a distinction between the so-called CA- and MCA-based approaches to membership categorization can be sustained if all analysts examining categorial matters are (as Stokoe suggests) committed to the view that categories are ‘relevant for the doing of some activity’ (Sacks, 1992, vol. I: 597), and are (accordingly) bound by the integrity of the interactional occasions on which categories become relevant.

In light of the above discussion, I agree with Schegloff’s (1999: 580) suggestion that:

The danger in exchanges like this is that the contributors and readers get drawn further and further into secondary discussions about the work, and further and further away from doing the work – whatever the work they choose to do is. Indeed, the ultimate danger is that this becomes the work they choose to do.

In this regard, there is much to appreciate in Stokoe’s latest contribution, which goes beyond producing or debating programmatic statements, drawing together a set of previously disparate or under-explicated resources that can be applied to doing analysis of categorial features of interactional data. Moreover, she has provided an extensive set of empirical demonstrations of the utility of these resources, thereby showing how categorial phenomena can be realized in ways that are systematically related to a range of sequential and action environments. In doing so, she has offered a promising way forward for a holistic approach to the analysis of categorial phenomena, making another substantial contribution to what strikes me as being an exciting time to be involved in this type of work.