A few days ago, Razib pointed to a story on ScienceDaily: "Many English Speakers Cannot Understand Basic Grammar". The underlying research is by cognitive linguist Ewa Dabrowska, who put a bunch of people through picture-sentence matching and discovered that they many do really badly with simple passive voice constructions. The story presents the study as news because it refutes the assumption, attributed to Chomsky, that grammar derives from innate cognitive abilities that do not vary among normal people.
I think it's very interesting, but I was waiting for Language Log to comment on the story. Now Mark Liberman has given a short account of it, based on a look at the preprint of the study. With a generally positive reaction, he concludes:
[A]lmost half a century after the work of Peter Wason (see here and here), I don't think anyone should find it shocking that significant numbers of people find it difficult to "understand" some fairly elementary sentences. I don't mean to say that there's nothing new here, just that Dabrowska seems to me to overstate the "consensus" about the distribution of linguistic (and in particular semantic) abilities of certain sorts.
The extreme version of Chomsky's position is obviously wrong from the standpoint of evolutionary biology. That's one of the reasons why Chomsky has consistently denied that grammar evolved under natural selection. But my reading of the field is that Liberman is correct -- most reasonable linguists don't subscribe to the extreme Chomskian view. For many years, people have been trying to investigate the acquisition of grammar from a developmental standpoint, and it's clear that some "rules" are learned very idiosyncratically and relatively late in childhood or adolescence. So the idea that these things don't vary has for a long time been known to be empirically false.
Still, I often see significant pushback against scholars who question the assumption of the grammatical unity of mankind. The comments section of the Liberman's post shows one way that these conversations develop -- picking away at the assumptions of the study, while claiming that the participants who showed a low ability to judge the grammar constructions were either not paying attention or just poor test takers. If we move to the position where variation is assumed to be the norm, I think that will be a step forward.
A question: If the passive voice is actually harder for a large number of people to comprehend, doesn't it follow that politicians and bureaucrats are unfairly discriminating against these people when they make routine use of the passive voice in speeches and official communications?