It was very nice that the change of address notification for our subscription to the world's finest news magazine worked so well: the August 27th issue of The Economist arrived in Santa Cruz and was available for reading on the flight to Boston (Barbara and I have moved to Massachusetts for the academic year 2005-2006, hence the extensive clearing and packing chores that account for my near-total absence from Language Log over the past several weeks), and the September 3rd issue arrived at the new address in Cambridge to make us feel we were truly home. After some preliminary set-up chores in our new apartment, I settled down to read the usually excellent technology column — but I was horrified. The Economist reports on the recent sequencing of the Pan troglodytes (chimpanzee) genome, a topic I turned to with some interest, and suddenly (bottom of page 70 to top of page 71) I read this:
And then there is FOXP2. This gene is known to be involved in language (people who lack a functional version of it cannot learn to speak).
Gaaaahhh! Not true. An absolute howler. Affected people do learn to speak. They are just very bad at it, and for mostly articulatory reasons — problems with fine-scale movements of the tongue. The instinct for language is there to a large extent, but execution is very clumsy. There has been some sensational press about the FOXP2 gene and the discovery of its location and effects over the past few years, but it has been long on sensation and short on factual accuracy. FOXP2 does not constitute a special gene controlling just linguistic capabilities. It is not a "grammar gene", though some have irresponsibly called it that.
Here's a passage from Alec MacAndrew's authoritative survey of the issue on The Evolution Pages. He starts by referring to "the KE family", the family whose inherited tendency to certain speech and other defects led to the discovery of the gene:
The KE family were brought to the attention of the scientific community in about 1990. Over three generations of this family, about half the family members suffer from a number of problems, the most obvious of which is severe difficulty in speaking, to such an extent that the speech of the affected people is largely unintelligible, and they are taught signs as a supplement to speech as children. It is a complicated condition including elements of impairment in speech articulation and other linguistic skills, and broader intellectual and physical problems. From the outset it seemed quite likely, from the pattern of inheritance, that the disorder is associated with a mutation in a single autosomal-dominant gene. It is rather surprising that such a diffuse condition should be linked to a single genetic defect, but it turned out to be so for reasons that we shall see later.
From the beginning, there has been a range of views in the professional scientific community with regard to whether the gene in question is a `language' or a `grammar' specific gene. Those disagreements continue in a somewhat abated form today.
The Disorder is not grammar or speech-specific
In 1995, Vargha-Khadem et al. published a paper investigating the phenotype of the disorder and showing quite clearly that it is not grammar or speech specific (1). They tested affected and unaffected family members and concluded that the disorder had the following characteristics: defects in processing words according to grammatical rules; understanding of more complex sentence structure such as sentences with embedded relative clauses; inability to form intelligible speech; defects in the ability to move the mouth and face not associated with speaking (relative immobility of the lower face and mouth, particularly the upper lip); and significantly reduced IQ in the affected compared with the unaffected in both the verbal and the non-verbal domain.
This last finding, about IQ, has been swept under the carpet by some commentators (including Nicholas Wade in the New York Times (2)), who claim that since the ranges of IQ of the affected and unaffected overlap and furthermore that since some of the affected achieve scores above the population mean for non-verbal IQ, then the disorder does not include a general intellectual challenge and is therefore language specific. But, note that the mean of the affected non-verbal IQ is 86 (range 71 - 111) versus a mean IQ for the unaffected of 104 (range 84 - 119). Not only is the mean significantly different between affected and unaffected family members but three of the affected had non-verbal IQ scores below 85, which is the normal lower limit for classifying speech defects as `specific language impairment' (that is, any disorder that affects speech only and is not caused by more general cognitive problems). Note, however, that the KE disorder cannot be explained solely by a general cognitive deficiency, because it is present in individuals whose non-verbal IQ is close to or a little above the population average and because it is accompanied by deficiencies in motor control of the face and mouth.
So the gene does have some effect on ability to articulate speech, but has other correlates as well, including significant intelligence deficit. What's really interesting, though, comes later in the piece. MacAndrews says:
The key point, that all the popular reports missed, is that FOXP2 is a transcription factor — in other words it has the potential to affect the expression of an unknown, but potentially large number of other genes. No wonder the syndrome presents in such a diffuse way. We know now that a FOXP2 homologue is strongly expressed in the development of the mouse brain. So not only does it potentially affect many other genes, but it is known to be important in the development of the brain (by being strongly expressed in the brain of the mouse embryo). I expect that breaking FOXP2 in mice would result in some compromises to brain structure and function - an experiment that someone is sure to do.
But there's more. That was dated 2003. This summer (while Barbara and I were engrossed in house-cleaning and book-packing) there was a new development. MacAndrews adds this note to his post:
27th June 2005: That experiment has just been reported. (Shu et al, Altered ultrasonic vocalization in mice with a disruption in the FOXP2 gene (7)). They report that: 'Disruption of both copies of the Foxp2 gene caused severe motor impairment, premature death, and an absence of ultrasonic vocalizations that are elicited when pups are removed from their mothers. Disruption of a single copy of the gene led to modest developmental delay but a significant alteration in ultrasonic vocalization in response to such separation. Learning and memory appear normal in the heterozygous animals. Cerebellar abnormalities were observed in mice with disruptions in Foxp2, with Purkinje cells particularly affected. Our findings support a role for Foxp2 in cerebellar development and in a developmental process that subsumes social communication functions in diverse organisms.'
This is exactly as I predicted above, two years ago.
FOXP2 is not the mythical grammar gene, no matter what you may have read about it in popular accounts, especially those favoring the view that language is a genetically inherited attribute of our species. FOXP2 relates to motor routines of various sorts including some of those that are involved in speaking. But its effects are wide-ranging. One of the things the Vargha-Khadem paper reveals is that affected members of the KE family cannot even wipe their upper and lower lips with the tongue tip and put their tongue back in their mouth, so profound is the disruption of their tongue-movement motor routines. Licking your lips isn't grammar. As MacAndrews goes on to say in conclusion:
We should beware of popular reports of scientific discoveries: almost all the popular reports of FOXP2 claimed that it was the gene for language or even more ludicrously the gene for grammar - the truth is more complicated and far more interesting than that. There are many popular reports of scientific discoveries which are equally sensationalised.
No-one should imagine that the development of language relied exclusively on a single mutation in FOXP2. They are many other changes that enable speech. Not least of these are profound anatomical changes that make the human supralarygeal pathway entirely different from any other mammal. The larynx has descended so that it provides a resonant column for speech (but, as an unfortunate side-effect, predisposes humans to choking on food). Also, the nasal cavity can be closed thus preventing vowels from being nasalised and thus increasing their comprehensibility. These changes cannot have happened over such a short period as 100,000 years. Furthermore the genetic basis for language will be found to involve many more genes that influence both cognitive and motor skills.
Human mind needs human cognition and human cognition relies on human speech. We cannot envisage humanness without the ability to think abstractly, but abstract thought requires language. This finding confirms that the molecular basis for the origin of human speech and, indeed, the human mind, is critical. Ultimately, we will find great insight from further unravelling the evolutionary roots of human speech - in contrast to Noam Chomsky's lack of interest in this subject.
Steven Pinker's view about FOXP2 is that the fixed human-specific mutations in the gene might enable fine oro-facial movements and so trigger the development of language.
My personal view is that the breaking of FOXP2 in the KE family is more likely to have caused a cognitive deficiency during development in those affected rather than a purely physical deficiency in oro-facial motor skills, and that these motor deficiencies are a secondary phenomenon, perhaps caused by lack of use.
It will not be easy to unravel the pathways by which language evolved in humans. If we are to have any hope of doing so, we will need close collaboration between linguists and biologists, who have, until recently, been rather suspicious of one another.
Here's to such collaboration.
Posted by Geoffrey K. Pullum at September 5, 2005 11:12 AM