At the Utrecht conference there was some discussion about the FoxP2 gene that I have been slow to report because I have been trying to get the story straight. I have had a couple of exchanges with Morten Christiansen about his presentation and with his help I think I understand what he was talking about. The basic argument is that genes shape the brain and the brain shapes the way language is learned, thus genes like FoxP2 contribute the way language is learned. The idea changes the role I’ve seen discussed for FoxP2.
Christiansen is a vocal advocate for the position that language has adapted to the brain, rather than the brain adapting to language. This blog has been generally appreciative of Christiansen’s position, although sometimes the quarrel over general-learning versus specific-language-learning can become so refined that it sounds like the medieval scholastics arguing over the proper terminology for transubstantiation. Yet the implications of the dispute are important for this blog. What sort of evolving did the brain do before language could appear? What sort of constraints on language were imposed by the way the brain shaped cultural evolution? Generative grammarians interested in the origins of language look for an explanation of how the universal grammar might have evolved, while Christiansen denies that a universal grammar ever did evolve. So more than scholastic niceties are at stake.
The basic premise of Christiansen’s argument that language was shaped by the brain is that one generation of speakers speak language L. The listeners modify L so that it is easier to speak and remember, giving us L1, which is modified by further speakers producing L2. For example, Christiansen writes me about work he has done with Florencia Reali showing, “that when a language is introduced with no word order, consistent word order emergences by way of cultural evolution of language.”
As an example of the way the brain affects language, Christiansen looked at sequential learning, which is the ability to organize discrete elements occurring in a sequence. Notice that phrase discrete elements. In language these case be phrases or clauses rather than words, so Christiansen is not challenging the generative grammarian’s most sacred proposition, that sentences are hierarchically structured rather than being strings of words placed one after another.
If language is learned rather than built into us, sequential learning becomes critical. The listener has to be able to discover the basis of the sequential organization. Christiansen did work early in this century (reported in 2001) in which he found that non-human primates are surprisingly bad at sequential learning of hierarchically ordered sequences.
Sufferers from strokes sometimes develop agrammatic aphasia, in which they are no longer able to speak syntactically. Christiansen has done work with these people and found that they have trouble with sequential learning.
There are some a priori reasons for thinking FoxP2 might be important in sequential learning. The cortical-striatal system in the brain is “implicated” in sequential learning and FoxP2 is important for the development of the cortical striatal system. So Christiansen invstigated. He studied 159 8th graders. (For readers, not familiar with American schools, 8th grade students are typically about 13 years old.) 100 were normal language learners; 59 of them had language impairment, although their non-verbal IQs were similar to those of the other group.
DNA tests looked at one part of the FoxP2 gene. A typical gene has a molecule symbolized by T; atypical genes can be symbolized by C. As each person’s genome gets a gene from each parent, the genotypes can be classed as TT, TC, or CC. The DNA tests found that:
- 70% of TT students had normal language, 30% were language impaired;
- 70% of CC students had impaired language; 30% were normal;
- 51% of TC students had normal language; 49% were language impaired.
This data is pretty clear. Language impairment does not come only from an abornormality in FoxP2, but much of it does.
Christiansen also conducted a sequential learning study with the students and found that the CC students had a much tougher time learning the task than the TT or CT. Thus, the FoxP2 gene that is associated with language learning problems is also associated with difficulties in a non-verbal, sequential learning task. This is the kind of finding expected by people (like Christiansen) who do not believe that speech origins depended on the evolution of language-specific qualities. It is one more blow to those who believe that there is a universal grammar that rests on language specific abilities.