My title paraphrases Oscar Wilde's memorable description of an English gentleman hunting a fox. For most  thnographers, translating local terms as though they had stable meanings is intellectually indigestible, yet we cannot dispense with it – any more than we could survive without reifying the categories of ordinary social life. Our descriptions, designed to illuminate how and for what the words are used, deprecatingly suggest that the language of the other is so different, so exotic, that we cannot really translate at all. We are all, in this sense, Whorfian extremists, if only by affectation.

But in fact this position, viewed pragmatically rather than referentially, is not an affectation at all. It is, rather, the simple fact that, as John Leavitt (1996) has observed, even those of us who believe that psychological inner states are neither attributable to whole populations nor even safely identifiable in individuals write our best ethnographic vignettes as though we could do both those things. The act of translating terms-in-context is a useful fiction because it suggests that we can identify the meanings that social actors intend. We all engage in this fiction, without which ethnographic description would be impossible.

And so the central paradox emerges: the plausibility of our accounts depends on a device that is itself predicated on an imaginative act of empathy with informants. We write as though we deduced those intuitions from regularly occurring actions and contexts, just as we do in our own everyday lives. This is a commonsense approach – but it fails to ask whose common sense is being invoked. We translate by declaring the terms untranslatable; we make translation the key metaphor of our reporting; and yet we know that translation, like comparison in Evans-Pritchard's famous adage, is ultimately impossible. (Greek shepherds and peasants say much the same thing about knowing other people's intentions, and similarly keep guessing at them.) But what else can we do?

Here I shall argue that the difficulty disappears if we treat ethnographic translation and literary translation as two different, if closely related, enterprises. This has to do with the difference between ethnography and fiction at the most basic level. Following Gregory Bateson (1958: 1) and Michael Jackson (personal communication, cited in Herzfeld 1997b: 24), I subscribe to the view that the major difference between these two representational genres concerns their management of psychological explicitness: a novelist usually backgrounds all the formal cultural principles that the ethnographer would want to spell out, but does describe innermost thoughts (see Herzfeld 1997b: 23). In an ethnography, by contrast, everything is at the level of collective representation because even highly individualistic acts are usually mentioned for the light they shed on communal values or on the scope of deviation. The very choice of marginal communities – and sometimes of marginalized viewpoints within them (Steedly 1993: 31) – is often a strategic and methodological device with which to explore the very forces that decide what is marginal or central.