Is constructionist research necessarily qualitative?
Some researchers have treated the issue of quantification as a definitional one; it is part of the meaning of constructionist work in general and DA in particular that it is qualitative. However, I think this gives too much importance to the quantity/quality divide and risks an uncritical support of qualitative work and criticism of quantitative. I would rather see quantification as quite appropriate in certain situations depending on a range of analytic and theoretical considerations. Nevertheless, arguments for quantification in discourse work can easily miss the point that the research goal is often that of explicating what should count as an instance of something; what is a compliment, or a blaming, or an error account? That is not something decided by counting — is a prerequisite for counting.
The issue of how far quantification is appropriate in conversation and discourse analytic work has received some detailed attention recently. A useful source is a discussion between various conversation, communication and discourse research researchers which explores different perspectives on the role of counting (Wieder, 1993). Schegloff (1993) and Heritage (forthcoming) both provide clear arguments for being cautious when attempting to quantify because of a range of difficulties that arise when transforming discursive materials into numerical form. However, the issue of quantification is undoubtedly going to become more important with the growth of this research area.
In parallel this discussion of the role of quantification in discourse research there is an increasing interest in study quantification practices as a topic. Such work has looked at quantification in both technical and everyday situations. Some of the classic work in this area was done by Sacks (1992) on members’ measurement systems. He was particularly concerned to show that mundane practices of quantification are not a poor imitation of the seemingly more precise and accurate technical practices of statisticians and scientists, but have a subtle and sophisticated logic of their own. For example a request for an appointment at half past four is not merely a less precise way of requesting an appointment at four twenty eight. Rather, different sets of expectations about punctuality, the type of meeting it is, and what delay would count as late are called into play in each case.
Other work in ‘ethnostatistics’, as Gephart (1988) has dubbed the study of statistics as a cultural practice, has looked at the way health economists have performed cost-benefit analyses (Ashmore et al, 1989) the construction of accounts of the success of charity funded cancer treatment (Potter, et al, 1991) and the textual practices for representing the ‘subjects’ in social psychology experiments (Billig, 1994). This work eats away at the idea that there is a straightforward choice between doing quantitative and qualitative research, and that entirely different considerations are involved in each.
Is that it?
There are many other considerations about the nature of discourse research and constructionist work more generally that could be addressed here. However, coming to see these abstract considerations is quite different from learning to do analysis which is very much a craft skill like bike riding or chicken sexing. There is no substitute for learning by doing, and such learning is almost always better done collaboratively so that interpretations and ideas can be explored with co-workers. Such learning is time consuming, hard work and often frustrating. However, the goal is to develop an analytic mentality that is sensitive to the action orientation of talk and texts.