Here is a short interview with Chomsky on linguistics, critics, teaching, politics and cycling. Note the discussion of Wolfe (and Everett implicitly). He makes the main point cleanly: the Piraha findings (if they
are findings) are irrelevant to any evaluation of his claims about the structure of FL/UG. Correct and, rhetorically, the right way to make the point.
I find it a lot better than some of the ones I've seen, but the way he puts what I'll call the 'One Human Language Idea' would still raise a lot of hackles, because it's really a hypothesis, or perhaps even just a conjecture, presented as more established than it really is, or at least has been shown to be, in the eyes of many.
ReplyDeleteI'll try another way of putting it. The basic principles that same Evidence implies same range of Grammars (I think we should put variation into the picture from the start, it's not 1965 anymore, and we don't need an ideal speaker-hearer). Rejecting it makes no sense. Things get interesting when the E is different, to a greater or lesser degree.
Perhaps the most interesting case today is when the E is slightly different, including in some ways that only involve rare or complex data that it would plausibly be difficult to learn a difference in G's from. Then the project is to find some more obvious differences in E from which the subtle ones might follow, given a theory of UG. The difference in behavior between 'quirky subjects' in German vs Icelandic might be a paradigm case, since the complement subject deletion/omission sentences that are generally taken to provide the best evidence that the supposed quirky subjects in Icelandic really are subjects are reasonably rare (I had to read five novels to finally find one, back in the day, and it appeared in a translation from English). But there are some more obvious indications, such as the overt subject position between a fronted adverb and an auxiliary, where only uncontroversial nominative subjects and the putative quirky ones go.
Another kind of case is where the E from the two languages is just different, so the project is to try to find the simplest possible
difference between G's that will produce it. This implies looking for more E in both languages, so if stuff is being found out, we can say that the method seems to be working, without appearing to be too dogmatic about our current ideas of what the G's are like (things might have gone better if, in Aspects, Chomsky had used the term 'Universal meta Grammar' (UmG), since that was what he was really producing there). It is assumed here that we want a common notation for writing grammars, which is often challenged, but still in ways that at bottom make no sense, because describing every language from absolute scratch without importing any information from previous work on any of the others is impossible, and nobody does it. You can try to be careful about what you're importing, and avoid the stupidest errors, but you're going to be importing ideas no matter what, so it's better to declare it. Another assumption here is that you can write formal grammars at all, but I think that is also vindicated, given a sensibly limited idea of what they are actually going to be able to do.
The last case is where we have relatively little E about some language, and the G seems kind of amorphous, to the extent that we can figure it out, but there is some other language that we know much more about that has some rather vague similarities, so we can then try to see what happens when we try to apply ideas that seem to work from the better known one to the lesser known ones. Legate's analysis of Warlpiri would be a good example of that, with various continential European languages a la Rizzi as the importation source. As in the other cases, if the process doesn't result in more information about the lesser known language, it isn't working.
There are other modes, but I think the above three are the commonest. The proposal here is to claim relatively little interest in whatever the grand nature of the G's is supposed to be, and more in the nature of the apparent differences between the G's for different languages, which I think is much more tangible subject in our present state of knowledge.