Virtually all of the best work on ethnic identifications in highland Southeast Asia (and here I include Southwest China) has emphasized the tremendous volatility of identities over time and the multiplicity of identities held at any one time by a groups and individuals. There seems very little that one can say about identifications that will "stay put" for very long unless it one is talking about the comparatively durable and static identity labels devised by colonial and native states in the modern era. But, on the ground, identifications are shifting, along with cultivation and place of residence.

The classic statement of this paradox is, of course, to be found in Edmund Leach's justly famous book Political Systems of Highland Burma. Not only does he document the shifts in identity but he suggests a model of oscillation and political choice that drives those shifts. The shifts in identity are, on Leach's view, essentially political. The successful Kachin chief, with a large following and a resource base (e.g. position on the trade routes, gems, etc) imitates the political structure of a petty Shan kingdom. It is the only model of "kingdomness" available to him and it, in turn, derives from the Chinese, Burmese, and Hindu models at greater remove which the Shans have emulated. Such Kachin kingdoms, Leach explains, tend to be self-liquidating because of the way in which they institute a kind of clientelistic serfdom that violates the lineage exchange system of the (gumlao) Kachin. The broader point that Leach makes is that identifications (e.g. ethnicity) ought to be seen as a kind of "positionality" vis-à-vis state structures. It expresses a political attitude toward authority and "statelike systems of political coordination".

Here, I want to combine this perspective of Leach with an account of the multiplicity of identities available to highland actors at any one time. Leach, in my view, has given an entirely satisfactory account for many of the changes in identifications over time but not a satisfactory account of the multiplicity of identifications available to actors at any one time.

Let's begin with the observation of most ethnographers of highland Southeast Asia. Most peoples, villages, and individuals have more than one possible identity depending on the situation. Their options are underwritten by the fact that they are likely to speak two, three, or more languages, have a series of customs that are found in more than one cultural zone, dress differently in different contexts, and move geographically between one hillside and another, between countryside and market town, between encounters with other cultivators and with officials. A part of this multiplicity of identities is built into the very fabric of culture; there are few if any "traits" (linguistic or cultural) that are definitive markers of durable identifications. Virtually any trait that one might propose as definitive will yield a host if indeterminate, "in-between" cases that are difficult if not impossible to classify. Of course, it is frequently in the interest of actors to create and occupy such "indeterminate ethnic niches" as it serves their purposes to be able to claim (or to move between) more than one identity. Here is where the border of ethnic (identity) classification, like any border, is an opportunity as well as an obstacle. There is no border that does not produce opportunities for profit (financial and social) by transgressing them. The trader may find it advantageous to be, say, a Pa-O when collecting and buying forest produce and a Burman or Shan when she goes to market. Here my point is a simple one. The "in-betweenness" ethnographers observe is, at least in part, a social creation: a result of the very social boundary itself and the fact that it generates opportunities for those who can straddle that boundary.

So far, this is pretty simple. It is not original to say that identifications are in part "strategic" or that official identifications create, willy-nilly their own new structures of opportunity which local actors are quick to exploit. Peter Sahlins has worked this out in subtle detail.

What I want to add is actually something that might be called a statistical observation. Let's imagine, that local actors have two or three or four different shades of identity which they can deploy strategically, depending on the situation. We cannot say, then, that they "have" an identity in any simple sense of that word. They have many identities which are contingent and, given the right circumstances, they are capable of inventing new identities on top of the ones they have today. But they are not polymorphous perverse! Their identities are largely structured by the contingent contexts in which they act. As hill padi farmers in their hamlet, they may be the Karenni from a small watershed, in the nearby town they may be Karenni pure and simple, in regional politics they may be Karen, displaced fifty miles they may be clients of a Shan petty chief, and, if they've been to school, they may even have a "Burman" identity to deploy. Each of these identities is triggered by an action context.

Here is then the point. What the naive observer may record as a shift in identities is frequently nothing more and nothing less that a shift in the relative statistical frequency of different action contexts. If, say, the presence of the state and its officials becomes progressively more intense over time, the identity appropriate to interaction with the state will be observed more frequently; actors will deploy it more frequently. It would be a mistake for the ethnographer to record this as a change in identity! Rather it is merely a change in the statistical frequency of an action context relative to other action contexts. Seen this way, if the statistical frequency of this action context diminishes vis-à-vis other action contexts, so will the identity associated with that action context diminish proportionately. Of course, is there is a shift of action contexts that is large and durable, if one or more action contexts come to dominate the life-world of actors, then these identities may be seen to dominate. To take a grisly example, Jewish citizens of Germany who might have seen themselves as secular, part of the intelligentsia, middle-class, and thoroughly German, even Protestant!, would by the end of the 1930s seen their action context dominated, not to say exhausted, by Jewishness whether they liked it or not. One could say that they had no choice but to become Jews; one could also say that what had been a marginal identity for them had become the only one available. They had not changed but their action context had changed radically.