15 October 2009

A quote on fieldwork

I quote myself, from a recent discussion on fieldwork.

“Fieldwork is not science, it is a craft and an art. The analysis of language can be scientific, but the gathering of language data is no more scientific than botanists tramping through jungles, or geologists clambering along mountainsides.”

Linguists, like people in other social human sciences, sometimes fetishize Science. So they try so hard to be scientific, and to ensure that all their work is scientific. But some things shouldn’t be scientificized, particularly fieldwork. Some of the things a linguist does during fieldwork are certainly scientific, like the development of hypotheses and testing them via elicitations, and putting together a theory from the results. But there are many things in fieldwork that are more a craft, based on educated guesses, heuristics, and inexplicable hunches. A geologist looks at a ridge and just has a hunch that it is out of place for the area and deserves some digging and picking. A botanist looks across a valley and has a funny feeling that the trees on the other side look a little bit unusual. Those sorts of things have nothing to do with Science, they are semiconscious operations of minds which have been honed to catch subtle patterns in the universe, patterns that point to where hypotheses and theories could be developed. Even the best physicist or chemist starts out with gut feelings that lead towards the real discoveries. It is dishonest to deny these sorts of things, it’s just the way that people work.

I think that a great failure of the American Structuralist school in linguistics was to fetishize science, especially in the form of “discovery procedures”. A browse of Zellig Harris’s Methods in Structural Linguistics (1951) shows how the process of investigating a language is reduced to a pile of rules. But even Harris admitted that linguists didn’t really work that way:

In practice, linguists take unnumbered short cuts and intuitive or heuristic guesses, and keep many problems about a particular language before them at the same time...

Indeed, if linguists didn’t take those shortcuts or guesses, then it’s very likely that we would all find that languages only varied according to the possibilities inherent in the discovery procedures. Thus the only thing we would discover would be the discovery procedures themselves. That would be boring, as well as uninformative. So why should we go back to that sort of thing again? Why should we have long lists of tests to submit to our informants, laying the language out on the table with probes and scalpels at hand? Instead, like any good field geologist or biologist, tie those shoelaces and start tramping around in the language, looking for things that look odd or out of place. Listen carefully to the rumblings in your gut, because it doesn’t always mean that you’re hungry.

(I apologise for the stew of metaphors.)