Become a Fan

« Most underrated philosopher of the week: Kevin Meeker | Main | Impossible demands for "proof" in the Giffords assassination attempt »

10 January 2011


TrackBack URL for this entry:

Listed below are links to weblogs that reference Some interesting data on poverty of stimulus claims:


Feed You can follow this conversation by subscribing to the comment feed for this post.

Ken Aizawa

Hi, Jon,

It's been a long time (10+ years) since I've read or thought about this stuff, but it's not clear to me that there is a lot here that would disturb the kind of theory Fodor or Chomsky hold(Pinker, I'm not sure about.) Fodor and Chomsky, if I recall correctly, proposed that large parts of syntax are innate, but that this innate language component is distinct from other intellectual capacities. (Think of Fodor's modules versus central systems.) So, for F&C, it might take very little data to pick out the parameters of one's local language so that the difference between 13,000,000 and 48,000,000 utterances does not really matter. Children might get the right syntactic parameter settings on 5,000,000 utterances. Notice that the claim Hart and Risley appear to be making is not that affluent children have a better command of the syntax of their dialect than do AFDC children. It is more like "they don't speak like the professors at the local university," whatever that means. Or, the H&R claim is implicitly that AFDC children have lower "attainment" than do affluent children. But, "attainment" isn't supposed to be innate; only certain syntactic parameter options.

Or, put the point another way. The additional sentence exposures of the affluent children might not help them learn syntax of their local language any better, but it might have a dramatic influence on central systems. And notice that the additional sentence exposures of affluent children are not just exposures to utterances, but are interactions (such as the handling of objects).

Eric Schliesser

I think Ken is right about this. (It's also been a decade for me.) All that Chomsky requires is that there is exposure to some 'normal' environment in a critical period.

Jon Cogburn

Ken, that all seems right to me, but I do think there's still an open question that the data set would allow us to examine.

Consider what's going on right now with the Piraha (and this was the straw that broke the camels back with Pinker). I mean it would be cool if someone went over the data concerning language exposure and four year old output with an eye to the syntax of the four year old's dialects. It should not be thought of as a priori impossible that the kids with less than one fourth the vocabulary also lack some of the syntactic resources involving for example modals or recursion (it is the analogous move with respect to the Piraha that drove Pinker nuts).

But you are still right that one could still hold onto the idea that natural language syntax is somehow innate universal grammar plus tweaked parameters by just saying thirteen million isn't enough. And one could I guess say the same thing about the Piraha at the end of the day, though maybe not. Everett spent something like a year trying to teach them to count to ten or add one and one. They did want the skill because they knew they were being ripped off in trade, but he could not teach them. His explanation for why he couldn't really fascinatingly combined culture and syntax in a way that does contradict Chomsky and Fodor. (Full Disclosure- I think the Chomskyan story is false for almost purely syntactic reasons, check out how much more coverage and much better constrained are contemporary non-transformational syntactic frameworks that posit more lexical structure such as HPSG).


Hi, Jon,

One of the arguments I remember for thinking that not much linguistic exposure is necessary for learning grammar went like this. There were apparently cases where children could be corrected over and over about some grammatical constructions and they simply would not get it. The idea, then, was that children set parameters in a particular order. If you give them data before it's time, then it's not much use. At the right time, the data sets the parameters pretty quickly. Alongside this were put observations that children in bilingual communities, French and some form of sign language as I recall, proceed through the "same stages" in synch. So, if they produce two-word sentences in French, then they produce two-word sentences in their sign language. So, the idea is again, when "the time is right" children set a parameter on little data. So, little data that they are able to keep their development in the distinct languages in stride. If learning to set a parameter were to take a lot of data, then one might expect the children to be farther advanced in one language over another. But, they are not.

Now, all that is what I gleaned from classes at Maryland, I guess, 10+ years ago. But, I do think that there is a good deal of work done on the dialects of children. I know that I recall a course with Stephen Crain in which he was exploring the abilities of children. The hypothesis he had, I think, was that children always speak a dialect that matches some parameter setting of a language spoken by an adult.

As for HPSG, I believe that part of the rationale for the move to the GB framework was supposed to be that, while the Extended Standard Theory, or Revised Extended Standard Theory, or whatever, had good data coverage, it was not sufficiently "principled". So, even by the High Church account, GB was an abandonment of coverage for "principle". They wanted to cover the data "in the right way". Now maybe HPSG has better coverage and is better constrained, but my exposure to HPSG suggested that what one could build into the lexical structure was pretty wild. It looked to me like there was a vast amount of structure there.

But, this is all by way of pretty handwavy recollection. And, alas, I have not kept up with the Piraha stuff, which does appear to be kind of surprising. No embedded recursion? I hadn't heard that. Only two color terms? I thought that was a part of some work by Berlin and Kay maybe 40 years ago or more.


The "no embedding" stuff reminded me of what seems to have been a kind of 1990's parade case for linguistic nativism, namely, the apparent spontaneous emergence of Nicaraguan sign language in a school for orphan hearing impaired children. This stands in contrast to Chomsky's comment in the Wikipedia story that one can have the capacity for, say, clause embedding, without manifesting it. Chomsky seems right on that score, but, if one accepts this, it does raise the issue of what does, and does not, get "spontaneously" manifest. Is pro-drop spontaneously manifest, but not embedding? If so, why?

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Your comment could not be posted. Error type:
Your comment has been saved. Comments are moderated and will not appear until approved by the author. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.


Post a comment

Comments are moderated, and will not appear until the author has approved them.