Re: The Starling's Song 2: The Revenge (was Re: Cognitive Linguistics, "The Language Instinct", and High-Functioning Autistics)

On Thu, 11 May 2006 12:05:48 -0700, David J. Peterson <dedalvs@...> wrote:

>Regarding our recent discussion of cognition and animals, I
>found this article stating that starlings' songs show recursive
>center-embedding on the UCSD homepage:
>
>http://ucsdnews.ucsd.edu/newsrel/soc/gentner_starling06.asp
>
>I'd heard someone talking about this guy giving a colloquium
>a few weeks back, but they must not have explained it well,
>because I sure didn't get *this* out of their description.
Well, they don't _show_ recursive center embedding; the claim is that they
can recognize a language that shows recursive center embedding.  If that's
not a misleading thing to call it: the language is a^n b^n = {0, ab, aabb,
aaabbb, ...} (a and b being rattles and warbles, resp.) which is produced by
a context-free grammar
  S ::=  a S b  |  empty

But one can explain this result by attributing much more restricted
capabilities to the starlings than processing arbitrary center embedding.
The one I think is most plausible is that they're just counting or timing
the strings of as and bs and comparing the lengths.

Language Log has a substantial discussion:
  http://itre.cis.upenn.edu/~myl/languagelog/archives/003101.html

Alex