- On anthropology, education, culture, and more … - http://varenne.tc.columbia.edu/blgs/hhv -

On following indexes as ethnographic methodology

Ethnography, like most (all?) scientific methods, must initially proceed on the postulate that there is, over there, some “it” to write about.  All critiques of ethnography have succeeded in demonstrating that, for human phenomena at least, this postulate cannot stand.  Anthropologists, as Geertz put it, do not study villages, they study “in” villages (1973: 22 [1]).  The new question that has not been answered: what do they do when they arrive in a village, if they are not going to study “it”?  Geertz suggests they might study “colonial domination” but does not quite explain what that might be. I suspect Geertz would say this is an ideal-type (Weber 1949: 89-95 [2]). Parsons might say it is a “formal category.” In either case, the anthropologist is just as much as a loss as when Malinowski or Boas told her to record “everything.”

I venture to say that most anthropologists of the past half-century have, uncomfortably, proceeded “as if” there were some there there, and I have often proceeded in such a matter—or let students proceed as if they would find objects to write about.

We have to find a clear way of stating what one is to do in the absence of a postulated ‘it’.  Many have argued for what they call “multi-sited” ethnographies which, I think, it intended to account for ethnographic activities when the ethnographer moves from one setting to the next in an attempt to … do what?  I have not been convinced.  First, there is the danger that one is led back to the initial problem: what is a “site” that there can be several?  Second, there is the matter of the selection of the sites.  Does one make this selection on the sense that there is a population of sites from which one select a sample?  How else might one proceed?

Working with students this past academic year, planning various research projects, and continuing to think about webs/networks, polities, etc., has re-opened this question with some urgency.  At this moment, I am exploring the following in an expansion of Garfinkel as compounded by Latour.  The fundamental methodological principle is: trust the people to tell you what makes them make differences in their lives and that of others.  The people, too, are trying to figure ‘it’ out, and, in the process ‘constitute’ it.  All we have to do is follow them.  This, of course, is not easy since the constraints on their methods (getting their work acknowledged as relevant to the task in such a way that the task is accomplished) are not those under which we operate (getting work acknowledged as ‘social science research’).  So:

  1. start with a salient phenomenon in some population (cohort). That is, start with a local (national) topic of conversation among the population. The more contentious this conversation, the better for our purposes.
    1. NCLB, indigeneity, autism would be such phenomena (to mention ongoing work by Jill Koyama, Jeff Schiffer, Juliette de Wolfe). These are salient in the United States or Canada. They generate a lot of talk. And that talk is easy to find in many setting.
      1. Do not attempt to define, say, “autism” or “indigeneity.” The participants, in their talk might make it look as if it is an ‘it’. You can remain agnostic while accepting that the practices in which the participants engage are very real and produce concrete consequences.
      2. Do not attempt to define the setting either. Again, the people will tell you its boundaries and reach through their own practices.
    2. Postulating a web means that one can start anywhere convenient.
  2. The danger is to take this starting point as THE core point. To prevent this, it may be best to start with a setting “obviously” peripheral
  3. listen carefully for indexical sequences (e.g. “We are doing this against our better judgement because they make us do it.”)
    1. these sequences are going to be included within larger conversations and will include many indexes to the current conversation and cohort, as well as to other conversations and cohorts.
    2. again, the more contentious the conversation, the more likely it is that linked matters will be indexed.
    3. the indexical sequences need not be verbal though one can start with the verbal as it may be easiest to access
  4. follow the indexes to the next convenient setting/moment
  5. repeat until exhausted (or a year has passed—though the temporality of such a search needs also to be addressed).

More to come.