20% of college students not smarter than monkeys?

Another summary of the study with smart starlings, written by the brother of a Language Log contributor.

In the article it mentions an earlier paper on tamarins, which I haven't read but about which I have some doubts. At one point it says that the patterns the tamarins were tested for (which were much like the starling's recursive patterns) were also tested on college students, 80% of whom were able to detect the pattern. Maybe there's something about experimental methodology that I'm missing, but that number raises a big red flag for me.

Saying that 80% of college students recognize a certain pattern is fine. But this is supposed to be saying something about a very basic cognitive foundation of human language; I sure would be concerned if only 80% of college students were capable of understanding or using recursive embedding in their native language. So what does that mean? The remaining 20% of the human subjects couldn't figure out the recursively embedded pattern, but use recursive embedding in language just as fluently as the 80% who did.

Like I said, I haven't read the paper about the tamarins (yet). There could be a perfectly good reason for the numbers this guy's throwing around. But if 20% of college students can't do something, is it really significant that monkeys can't either? And then is it significant that Gentner's 9 starlings apparently can?

I just don't think that the ability to recognize arbitrary patterns of recursion has much of anything to do with the ability to understand a sentence like "the poison your mom ate killed her." Maybe I'd feel differently if any of the animal language studies provided really compelling evidence for the connection.

AAABBB versus ABABAB equals smart starlings?

I finally got around to acquiring and reading that article about starlings that everyone seems to be talking about. ["Recursive Syntactic Pattern Learning by Songbirds" in Nature, 27 April 2006.]

I basically agree with what everyone else is saying; it's an interesting experiment, and those birds sure are smart, but it says nothing about the nature of context-free grammar versus finite-state gramar and it especially says nothing about the human language faculty versus animal language.

In fact, I fail to see how this has anything to do with langauge at all. It's not even really about animal language, because aside from the fact that the experiments used samples of starling song motifs, there's no mention of what these birds actually do. And it's certainly not about human language; they don't even attempt to provide evidence that the patterns they're using correspond to the kind of recursive embedding we see in the syntax of natural languages.

Even if we assume Gentner et al are right in their claim that these starlings really are recognizing the recursive embedding of the motifs (instead of just detecting a switch from one motif to the other), there's nothing to show that starlings are capable of using recursion, let alone using it productively. Whether the ability to recognize patterns of this sort is uniquely human or not bears absolutely no implications for the question of whether the charactaristics of natural languages (and by extension the narrow language faculty) are uniquely human or not.