the aim of linguistic research in the community must be to find out how people talk when they are not being systematically observed; yet we can only obtain this data by systematic observation.It was interesting to read this morning over on scienceblog.com about the parallel issues in 'field psychology' (here). The post describes how much even relatively mundane experiments suffer from these kinds of effects. What's striking to me is that in our own lab work, linguists really have the very same issues.
This calls to mind a kind of meta-paradox that has been bugging me for a while: At times the observer's paradox seems to get overvalued in sociolinguistic work – the real difference between truly 'unobserved' speech and speech being recorded in a careful, low-key way is typically pretty minor, as far as I can tell. But the kinds of lab effects talked about in the post seem really and sometimes grossly undervalued in much linguistic work we do in the lab. I'm regularly surprised at how little discussion is given to these problems in such publications and in talking with people doing that kind of work. It's often fairly easy to do some kind of control — testing free speech versus reading tasks, for instance. Ideally, I want to be able to compare tightly-controlled data with more sociolinguistic data.
The image, from here, is called "The Observer and the Observed", by Lewis Wickes Hine. It's part of a collection done from 1908-1924 for the National Child Labor Committee.