« Antidote to Anti-Racist Math | Gene Expression Front Page | Dawkins reconsiders Out of Africa? »
February 02, 2005

The children of Universal Grammar

Nick Wade just wrote an article titled, A New Language Arises, and Scientists Watch It Evolve, in The New York Times. Wade normally does the bioscience beat from what I recall, so it is no surprise he spins this into a Nature vs. Nurture story:


Linguists have long disputed whether language is transmitted just through culture, as part of the brain's general learning ability or is internally generated with the help of genetically specified neural circuits that prescribe the elements of grammar

Over the past few years I have read several books that touch on the language & biology issue. This thorough review of the evolutionary origins of language mentions the famous ban by the French Academy on mooting this topic. It is no surprise that humans are verbose about their verbosity I suppose.

I would actually be curious if there are any linguists who believe that language emerges just out of our general intelligence capacity mediated by culture. I know that Wade glosses over the genuine texture in this field (lingua-cognitive science?). There are many flavors of "innate" conceptions of language capacity/aptitude, and even disagreemants among "Universal Grammarians." Last year I posted an exchange between Chomsky et. al. and Pinker et. al. over the issue of how UG emerged. While Pinker takes a adaptationist viewpoint, not surprising given his thorough Darwinian outlook, Chomsky's "recursion only" spandrelian thesis seems parsimonious to the point of deux ex machina (my cards on the table, I tend to favor adaptation as a null hyopthesis for complex phenotypes which I suspect are metabolically expensive with fitness implications which might swing both ways across a wide arc).

These debates are fresh in my mind because recently I read The Symbolic Species by Terrence Deacon. PLOS summarizes Deacon's views:


...many of the language universals reflect semiotic constraints inherent in the requirements for producing symbolic reference rather than innate predispositions. Thus, neither evolved innate predispositions nor culturally evolved and transmitted regularities can be considered as the ultimate source of language universals.

Believe it or not, this is a rather transparent and plain spoken rendering of Deacon's sometimes opaque jargonese. The short of it is that Deacon advances a view of innateness which rejects UG. The chapter where Deacon takes on Chomsky & co. and UG he offers his counter-proposal, that UG is an artifact of the reality that language coevolved with the human brain , the ones we see being babbled around us tend to reflect the learning biases of children (the Baldwin Effect looms large in Deacon's narrative). That is, languages were shaped and "adapted" to the architecture of the human mind.

How is this different from UG? Frankly sometimes I wonder. Deacon basically forwards the proposition that languages were selected and constrained, some would say "canalized," toward certain universal structures (which UG theorists posited emerge out of UG), but these structures are only byproducts of cognitive biases and limitations of the human symbolic mind. UG is really just a reification of tendencies of language that emerge out of pecularities of our symbolic mind. On the level of axons, dendrites and synapses, I don't see that there would be a great difference between many of Pinker's views and Deacon's hypothesis when it comes to UG, but I might be wrong. On the other hand, Deacon promotes both neural connectionism and attempts to diffuse points in favor of mental modularity in the context of language, which seems to me a pretty straight on attack on Pinker's model (Deacon's book was published in 1997, so he makes copious references to The Language Instinct). In The Blank Slate Pinker attempts to throttle neural connectionism (which I suspect many of his readers were confused by as connectionist vs. non-connectionist debates were not that salient to them) and implicitly promotes massive modularity.

But, one point Deacon makes forcefully, and which I think is likely on the right track, is that UG has tended to encourage "hopeful monsterism" among many theorists who utilize it as a given in other fields. Robin Dunbar in Grooming, Gossip, and the Evolution of Language and Richard Klein in The Dawn of Human Culture both appeal to hopeful monsters, alluding to what seems to be a saltational "rewiring" of the human mind that led to the "Great Leap Forward." Steven Mithen in Prehistory of the Mind tiptoes up toward hopeful monsterism as well, contending that the Great Leap Forward occurred because the human brain made a jump to transdomain cognitive fluidity, in particular, through the use of analogy. This is all great, but the problem is that the answer to the question seems rather like pointing to the aether or concocting phlogistan, it is something of a serendipitous blackbox which we must wonder over, but not consider overmuch how the miracle is actually constructed, as we know it must have been. Now, I must admit, there are researchers who are exploring the possibility that Mithen is on to something with his thesis of transdomain cognitive fluidity, and I think that is the right direction to go, but Deacon's book is a dense read that is worth it just so that one can expiate the sins of devouring more readable tracts which end with the conclusion of a hopeful monster at the end of the tunnel. There is something vanilla plausible in Deacon's gradualist non-modular thesis of language evolution which makes one more suspicious of modularist saltational sexiness.

So, back to the article I pointed to above, I think Pinker is probably thinking of people like Deacon when he contends that the enormous leaps and bounds in linguistic evolution that the Bedouin seem to illustrate, as well as the recent Nicaraguan story, strengthens the hands of those who would want to minimize the importance of long term cultural evolution and selection in shaping the character of a language. But who knows? Because cognitive scientists are often not mathematically precise, a "get out of jail card" always exists via careful and weasily clarification of exactly the extent of cultural evolution needed, or in other contexts what exactly is meant by innate.

In the interview with William Hamilton below he dismisses the importance of math in his work. Frankly, I think that that is probably self-effacing bullshit. Terms like "modularity" leave a great deal of wiggle room (what is massive modularity as opposed to !massive modularity?), while the neo-saltationists have a tendency to present their hypotheses in a way that leaves a great deal of slack for audience interpretation so that they aren't tagged with the "explain everything with nothing" label. I wouldn't mind if cognitive scientists just made up some numbers for how much of language capacity is modular and how much is general intelligence, some percentages would give a rough sense of proportion to calibrate terms like "massive" and "moderate" and "not modular" (from what I can see, no hardcore massive modularist is a pure modularist, while I doubt even the most hardcore neural plasticist would deny some modularity). Theorists like Dunbar and Klein should also wander into the genetics departments and egg them on to look for more selective sweeps, since the "rewiring" mutation must have been felicitous indeed and been a great fitness boost to those who had it.

In any case, I talk, but I don't know why. I guess I'm somewhere between moderate and massive modularist, not quite a pure adaptationist, but sure as hell not a spandrelist, while I think people should go poking around the synapses more before spending all this time philosophizing about "deep structure" in grammar. I'll leave it to you to figure out what I really mean.

Related: On the mind & brain, FOXP2, Days of Ice & Awe, The Dusk of Human Culture and Grooming => Language ~ Larger Social Groups.

Posted by razib at 03:27 AM