Who are the best subjects for judgment tasks intended to test grammatical hypotheses? Michael Devitt ([2006a], [2006b]) argues, on the basis of a hypothesis concerning the psychology of such judgments, that linguists themselves are. We present empirical evidence suggesting that the relevant divide is not between linguists and non-linguists, but between subjects with and without minimally sufficient task-specific knowledge. In particular, we show that subjects with at least some minimal exposure to or knowledge of such tasks tend to perform consistently with one another—greater knowledge of linguistics makes no further difference—while at the same time exhibiting markedly greater in-group consistency than those who have no previous exposure to or knowledge of such tasks and their goals.
Background and Clarification
The Context of Devitt's Claim and His Psychological Model