I've just read a review article about the Foxp2 gene and its role in the evolution of language and am happy to have been brought up to date. The big assertion in this paper is that Foxp2 is critical for learning vocalizations, maybe not so critical for other aspects of speech and language. (Wolfgang Enard, "Foxp2 and the role of cortical-basal ganglia circuits in speech and language evolution.") The paper is most notable for its presentation of specific facts that support the picture of Foxp2's role:
- [the human genome has undergone] 17.5 million single nucleotide changes and … 2.5 million insertions and deletions … since we split from the lineage leading to chimpanzees some 6 million years ago: That totals 20 million nucleotide mutations, which averages to 3 1/3 per year, or perhaps about 40 per generation. Most of these changes probably produce no important changes in the organism, but the numbers are provocative. First, 40 per generation is pretty small in a generation. I bet there are more changes than that in the ordinary transmission of genes between parent and child. But most of those changes are not tolerated, not even as drift. You can think of the genome as a mountain that suffers from constant erosion, but because of selection the mountain doesn't go away. Negative selection keeps the mass of functioning genes in place, still functioning. Other changes are tolerated because they do no harm. Much of the genome is subject to a random drift that doesn't alter survival rates. Finally there is positive selection, a change that is favored over the old form. I don't suppose we yet have a count of positively selected changes, but one day we will be able to say that roughly X% of the changes since our last common ancestor with the chimpanzee were positively selected.
- The human capacity for vocal learning, that is to imitate complex vocalizations… is absent or very limited in primates and mammals, [so] the proficiency of all humans to learn vocalizations is very probably a phenotype that required genetic changes: that's putting it mildly, but there is reason for caution. Humans do many unique things, like solve calculus equations, that are probably cultural and require only very generalized skills. But vocal learning is different. It is universal to the species and mastered at an early age, too early for any formal schooling. So very likely some of those 20 million nucleotide mutations support vocal learning.
A particularly conservative gene is the Foxp2
- Foxp2 ranks among the 5% most conserved proteins: the fact that this gene is a conservative one is well known, but it is impressive to see the conservatism expressed this way.
- Of the 708 aligned amino acids [in Foxp2 in] mouse and human differ at 3 positions. Remarkably, 2 of these changes occurred in the short timescale of human evolution after the human lineage split from the lineage leading to chimpanzees: Two of 708 works out to 3/10% so the change is still very small. There is, however, a kind of multiplier in this case because Foxp2 is a "transcription factor," meaning it works through intermediaries, effecting the operation of other genes.
- This rate [of change] is also unusual when compared to the evolution of 175 transcription factors in primates that even have been preselected for some evidence of positive selection. Only five of them have a faster evolution in humans when compared to the other primates and these are not associated with brain function, but with spermatogenesis and the immune system
- The two amino acid changes happened before the split from Neanderthals and Denisovans 270,000 – 440,000 years ago. The independent findings in Neanderthals and Denisovans make it also very unlikely that contamination with human DNA or gene flow from or two Neanderthals happened: When the gene was first discovered it was given a very recent data for its evolution, but that date has now been discredited. Denisovans, by the way, are a Homo group that split off from Neanderthals after the Neanderthals split from us.
The gene has effects throughout the body, but the critical one in human evolution appears to alter the brain.
- Mice [with non-functioning versions of Foxp2] show severe developmental delays [and] die within four weeks after birth: this suggests why the gene is so highly conserved. Mutations that make the gene non-functional are lethal mutations, thrown out of the genome immediately.
- Reducing [Foxp2] protein levels affects motor skill learning in mice, vocal imitation in birds, and speech and language development in humans: so even if the mutation doesn't end function, simply reducing function reduces an individual's competitive advantages.
- Humanized Foxp2 [in mice] does not have a detectable effect in many organ systems in which Foxp2 is expressed but seems to effect especially the brain
- Evidence … strongly suggests that … reducing the dosage of Foxp2 impairs speech and language development: it goes on to say that people without two Human Foxp2 genes resemble "patients who have acquired aphasia from stroke." It looks like the gene is still highly conserved in the species, since reduced function imposes a serious challenge. Apparently the story of the gene's evolution is: last common ancestor of chimpanzees, bonobos, and us gave all lineages the same Foxp2; at some point the human gene altered, details of the process remain unknown; the human lineage conserved the new gene so that the last common ancestor of us, Neanderthals and Denisovans gave all lineages the same gene.
The gene is useful for motor learning and with a slight alteration can be used for vocal learning
- The effects of humanized Foxp2 seems to be specific for CBG circuits … [which] build loops with all cortical areas [and] play a role in a large number of processes related to the learning and control of motor and cognitive processes
- Within [the bird brain area] that is specialized for vocal learning Foxp2 expression is increased … during the weeks of vocal learning and decreased two hours after singing: this detail underscores the modern view of the genome. The gene was originally thought of as a determiner of traits, like blue eyes and curved earlobes. It turns out to play a much more active role in the body's metabolism. It is expressed during relevant activites.
- These circuits seem not generally required for adult language processing, but might be more relevant during language acquisition.
That last point makes me wonder. Could I be given a few doses of human Foxp2 juice and learn to speak Turkish without an accent? Less melodramatically, does the end of the critical learning period arise from some reduced Foxp2 activity?