“Judging” grammaticality through software: MiniJudge. (Via Jean Crawford, Starr Linguist)
As a complete outsider to the minimalist program (and to those language sciences which focus on syntax), my perception has often been that judgements of grammaticality relied too heavily on introspection by native speakers. Proponents of these generative theories often talk of “instincts” or “intuitions” for those judgements that native speakers are able to make unconsciously and that non-native speakers have a hard time making. Maybe using software for those judgements would take the generative methodology out of introspective mode.
As a linguistic anthropologist, I just wish linguists and other language scientists could talk to each other.
I agree that native speaker instincts/intuitions can get a little bit out of hand, which is why I prefer to rely on the intuitions and introspection of a native speaker other than myself. The more consensus you have on grammaticality of a sentence, the more likely it is to be part of the actual grammar. There is also the problem of the “polite informant” and the “data fabricator”, which I should blog about sometime. It’s really an issue of field methods, which doesn’t always get taught (or get taught well when it is taught). Anyway, another topic for another entry, not a comment.
I like it, though.
The thing that gets me the most, I guess, is the part about distrust for corpora. A corpus needs not be seen as fully representative or even reliable but if you find examples in a text (oral or written) which relate to your analysis, you can more easily argue about grammaticality than if you try by yourself to assess whether or not something “can be said” in your native language.
Will continue following your blog. You seem to be on a roll!