Linguists are still debating the force of "stimulus poverty arguments." This may never end, but a new rejoinder from Robert Berwick et al. (Chomsky is 4th author) appeared in Cognitive Science (2011 issue 7). In this article, the authors argue informally, but relatively correctly it seems to me, that a host of recent efforts to model the learning of language with virtually no prior assumptions about its structure in fact fail to account for the interesting structural constraints found in real languages.
They point out that recent proposals in Bayesian learning (Perfors et al. 2011) and learning of substitutable languages (recent papers by Clark and Eyraud) remain tied to string-based models of sentences, while the interesting syntactic constraints that are observed have to be represented using syntactic structure of some kind. This leaves the question unanswered, where do the structural constraints come from, and could they be learned? I think that some of my own recent work has addressed the issue of learning structures rather than strings, but then we have to learn from (at least partially) structured data, which presents its own plausibility problems.
An interesting point was brought up by my colleague Chris Golston, who is a linguist very interested in biology and evolution. He reminded me that stimulus poverty and learnability arguments have limited force in reality, because simply proving that a thing does not have to be innate, does not show it is not innate. A simple example comes to us from birdsong; it seems clear to everyone involved that birdsong is not that complicated and could theoretically be entirely learned, yet it is also widely held that birds harbor a considerable amount of innate propensity to sing a certain way. Not being a birdsong expert at all, it is possible I have misread the consensus in that field, but that's how I would summarize it for now. It is, I believe, quite likely that human language has this same general property---perhaps more of it is innate than really "needs to be" from a theoretical perspective.
Subscribe to:
Post Comments (Atom)
The second paragraph seems backwards -- The Stimulus Poverty arguments are arguments *for* innateness not *against* it. I think they are (potentially) very strong -- they are arguments that something has to be innate, and therefore is innate.
ReplyDeleteOn the songbird issue, yes, some of it is innate, since if you raise some species in isolation they start singing, which again is an excellent argument for innateness.
Not clear what the corresponding argument is for language, if you abandon the poverty of the stimulus? I mean when children are deprived of language, they don't, contra Herodotus, start speaking Hebrew.
Well I was a bit unclear; the Berwick et al. paper was singling out Perfors et al. and Clark & Eyraud (that's you, right Alex?) as examples in which authors tried to show that language could in theory be learned without anything innate. At least I thought that's why they brought these in. Then Berwick et al. pointed out that the kinds of languages and constraints on syntax that could really be learned by the cited techniques do not really address the reality of natural language, and so do not negate the old stimulus poverty arguments in favor of innate principles.
ReplyDeleteYes, I was being a bit obtuse ..
DeleteI found that paper hard to follow for a number of reasons:
It's not clear what the POS argument is for in this version. Previously, it was meant to be in support of a rich innate language faculty, but Berwick et al. no longer seem to believe in that, so it's not obvious what they want to use the argument for. The conclusion they draw is that there is just structure-dependent movement, but as you certainly know, in Stabler's formulation this is equivalent to MCFGs, which is kind of uncontroversial (I think that characterisation is about right) and doesn't explain the learnability problem that it is meant to. If English formed polar questions by fronting the first auxiliary then that language would be in the class of MCFGs, so their proposed solution doesn't follow from that argument.
Not so sure about the technical issue here, but the general issue is one resulting from a paradigm shift.
ReplyDeleteIt's true that the old POS argument was used to support the rich innate UG that was needed for Principles and Parameters theory.
But nowadays we have the Minimalist Program, which proposes a minimalist approach to UG as well.
This much I agree with, so I am no longer certain what POS arguments are supposed to be accomplishing anymore.
The only people left to argue with are total anti-nativists, and I have my doubts that a POS argument is the right way to deal with that disagreement.