The Language Gene That Wasnt

KE family child. 'Looh ah'

Interviewer. 'And where do you live, Laura?' Child. Unintelligible. Interviewer. And how old are you?' Child. 'Foa. Foa wes.'

One of the most extraordinary scientific detective stories of modern times concerns the discovery of a gene for speech and language impediment in a large British family. It is a tale that is full of serendipity, riven with scientific controversy, and with important implications for human evolution.

Specific Language Impairment (SLI) tends to run in families but even by the standards of SLI the KE family are an exceptional example. Elizabeth Auger's special educational needs unit at a primary school in Brentford, west London, was inundated with them. At one point, in the late 1980s, she had seven children in her unit at the same time, all from this one extended family. Several family members from the older generations had been previously referred to her. They all suffered, to a greater or lesser extent, from a kind of palsied speech. It was as if the lower part of their faces was somewhat frozen, they simply couldn't form words properly. Several of them had halting speech, as if they were searching for words. Sentences were short and broken and vocabulary was limited. Their speech has been called telegrammatic. When you watch videos of the children struggling with even the most simple sentences you see a mouth that simply will not go where it needs to go to form fluent words. Consonants were a living nightmare. They were omitted, elided, or approximated, as in words like 'boon' for 'spoon', 'able' for 'table', and 'bu' for 'blue'. Polysyllabic words like 'parallelogram' would be completely beyond them. They would often neglect to inflect verbs, using only the present-tense stem. Even the more fluent, older, affected members of the family only managed some fluency of speech by gliding over the normal transitions that clearly mark individual syllables, and by ignoring distinctions between singular and plural, or appropriate verb tense endings, and by filling up the awkward gaps with shrugs, hand gestures, and grunts.

Auger decided it was high time this family was properly investigated and raised the matter with a medical geneticist, Michael Baraitser, of the Institute of Child Health in London. Baraitser discussed it with his head of department, Marcus Pembrey, and Elizabeth Auger was invited to a meeting at the Institute, along with two affected members of the family. Most linguists at the time were very hostile to the idea that genetics could be involved in language impairment, but to Baraitser and Pembrey, the occurrence of such similar symptoms across such a large family suggested the opposite. However, with such a complex phenomenon as language, it would be naive to imagine that those genetics would be a simple matter: it would be far more likely that many genes were involved. Nevertheless it was very exciting and their registrar, Jane Hurst, was assigned to visit the family, note their symptoms, and take blood samples for genetic testing.

Hurst, although a clinical geneticist and not an expert on the psychology of language, found no difficulty distinguishing between affected and unaffected members of the family. She used no tests for language ability whatsoever, she just listened to them talking. In all, throughout three generations, she counted 16 affected individuals. No unaffected parent had ever produced an affected child, and every affected parent had passed on the language disorder. Painstakingly, she drew the first family pedigree for the KEs and it quite clearly showed something that none of the researchers had expected. This particular language impairment did not require an orchestra of genes at all, it bore the tell-tale pattern of the effect of one single dominant gene, on one of the chromosomes other than the sex chromosomes.

Gene mutations are described as being either dominant or recessive. If a gene mutation is recessive we must receive two copies of it, one from our mother and one from our father, in order for it to lead to a trait or characteristic that shows up in us. Scientists call this the homozygous condition. However, if a gene mutation is dominant, its effect can never be masked even if we receive only one mutant copy from either parent. Each child therefore has a 50% chance of inheriting that gene mutation, and the trait that goes with it. This was the pattern noted in the KE family. Could this be the first evidence of genes for language? The BBC was onto it in a flash. Could they film the bloods being sampled? Hurst remembers that only one blood sample remained to be taken, that of the all-important grandmother, the oldest surviving affected member of the family, from whom all impairment flowed. Well into her first pregnancy, Hurst recalls the difficult car journey from her home in Brighton to west London in January 1989. 'I felt terrible with morning sickness. I thought, "Oh! God! I'm going to be sick on camera, taking her blood!"'

Meanwhile, over in Montreal, Canada, the psycho-linguist Myrna Gopnik was preparing to visit her son and his family in Oxford, England. Gopnik had recently founded a Cognitive Science Group at McGill University which met regularly and included paediatricians. One of them had introduced her to an interesting 14-year-old patient who had great difficulty constructing fluent narrative speech. He spoke in very simple sentences and failed to use proper tenses and plurals, where appropriate, for words. His father shared the same difficulties. Although a successful computer scientist, he had no simple everyday conversation and his speech was laboured and agrammatic. It reminded her of the speech of aphasics, whose language difficulties arise in adulthood because of brain traumas like strokes, especially if they affect Broca's Area, the main language processing centre of the brain. While she was talking to the father she noticed he had very poor inflection. He told her he had 'two computer' for instance: he couldn't make the plural. His wife told her she thought her husband's language was impaired and that it took great effort for him to construct simple sentences with inflected words in them. Even simple conversational sentences like 'We built the pool about five years ago' required great and laboured thought. It reminded Gopnik of how tired one gets after a day of stumbling through conversation abroad in a foreign language. These people behaved, she thought, as if they had no native language. None of it 'came naturally', it all had to be computed.

Some years earlier, Noam Chomsky had shaken the world of linguistics by pointing out something that Darwin had previously drawn attention to: that children pick up languages astonishingly easily, yet, of course, they do not automatically seem to know how to bake cakes. Chomsky had proposed that children learn language too fast to be absorbing the rules of grammar via exposure to speech in everyday life. Their brains must be pre-disposed to it. There must be a construct, he said, called Universal Grammar already programmed into young brains. Chomsky fell short of any accurate biological description of his language organ and never mentioned genes. However, Steven Pinker, formerly a colleague of Gopnik's at McGill, but then at the Massachusetts Institute of Technology, was expanding Chomsky's theory into a truly biological dimension. He suggested that language was constructed in the brain by a number of computational modules, each responsible for a particular component, like the appropriate inflection of words. It was likely, said Pinker, that each of these computational modules would be underpinned by its own discrete genetic foundations.

While in Oxford, Myrna Gopnik was invited to give a guest lecture at the Psychology Department. Wandering around the building with her daughter-in-law, her gaze alighted on a bulletin board which gave notice of interesting forthcoming events. The BBC's Antenna science television series, it noted, was transmitting a film that week about a family in west London who appeared to be carrying a gene for language disorder. This was the very film for which the heavily pregnant Jane Hurst had forced herself up from the south coast. A few days later the Gopnik family watched, fascinated. Could this be one of the genes Pinker had invoked?

Gopnik located the primary school, rang Elizabeth Auger, and obtained an invitation to visit and give an informal talk about language impairment, based on her observations of language-impaired individuals in Montreal. In the audience was the KE grandmother and several other family members. Much of what Gopnik had to say resonated with her audience and she was invited by Auger, and encouraged by the grandmother, Mrs. K, to study them. Gopnik recalls being told by Auger that, although she had mentioned the KE family to members of the Institute of Child Health, there had appeared to have been little interest, leaving the field open. This, although Jane Hurst had been taking bloods since 1987! Gopnik agreed to work with the KEs and returned to Montreal to prepare a battery of tests.

By now, the Institute of Child Health's own language expert, Faraneh Vargha-Khadem, was also on the case and had already begun testingfamilymembersinherlanguagelaboratory. WhywasGopnik invited on board at all? It may be that Mrs. K was deeply resentful of some of the comments made about her family's impairment. Conventional linguists were anxious to suggest an environmental route to the disorder and there had even been a peculiarly British social slur analysis which had suggested that, since the family were working class, mangled words and poor vocabulary were to be expected. For Mrs. K, having the family at the centre of important scientific investigations was an excellent way to regain lost dignity. Scientists? For once, you couldn't get enough of them.

The problem for the British scientists was not one of lack of interest but lack of cash. Marcus Pembrey, like many of his British contemporaries then, as now, was a pauper king. Although he had been fortunate enough to have stumbled over such a large family pedigree, and had taken advantage of that by collecting a large number of blood samples, there was no money in the kitty to find the genetic culprit. Pembrey decided to try to get Gopnik on board by inviting her to pool her results with those of Vargha-Khadem and the others. When Gopnik resisted, the Dean of his Institute questioned Gopnik's access to the family by suggesting to his opposite number at McGill that her intrusion would cause unnecessary stress on this vulnerable family and had not received ethical approval. His opposite number declined to intervene.

Gopnik waded in with a battery of tests originally designed for the study of aphasia. This has always been criticized because aphasia, as we have already seen, is strictly a language disorder caused by damage to the language processing centre of the brain in later life. The KE family clearly inherit their disorder and so it is present from birth. Furthermore, whatever the language problems, they were accompanied by severe difficulties in the production of words. Their disability appeared to involve speech and language. Nevertheless, Gopnik concluded that the core deficit in the KE family concerned an inability to change the tenses of words and the lack of a general rule for producing appropriate plurals. For instance, they might be able to distinguish between a picture of a solitary book, versus a pile of books, when asked to point to 'the book', a known word they might have filed in memory, but when shown a picture of an imaginary animal called a 'wug' they could not describe a group of these animals as 'wugs'. Altogether, she said, they lacked the underlying grammatical rules for changing word endings to indicate singular, plural, ortense. Theycouldthereforenotdoitautomatically, butonly if those words had been stored in memory. The defect, she stated, was not one of general cognition; some aspects of grammar were spared, and it was restricted to one feature of grammar. Astonishingly, she made no mention of the severe facial palsy of the KE family, a palsy others had long noted to be so severe as to prevent them articulating the sounds of speech. She had turned a blind eye to their inarticulacy.

Gopnik rushed into print, and published a letter to Nature in April 1990 titled 'Feature-blind grammar and dysphasia'. (Dysphasia is a mild form of aphasia.) The British scientists had implored her not to publish and warned her that she did not have the full story. When she went ahead anyway they accused her of simply 'gathering the low hanging fruit'. To this day, Gopnik asserts that, since the same grammatical faults turned up in members of the KE family when asked to write, the core deficit must be grammatical, and not one of articulation. Furthermore, she asserts, she never had any real problems understanding what they were saying, even on the telephone. For the next seven years she stood her ground, arguing strenuously that the articulation difficulties were a distraction, and implying that the London-based scientists, by stressing inarticulacy, were arguing that there were no underlying grammatical difficulties at all.

In the same year that Gopnik published in Nature, Steven Pinker published a major paper in Science, titled 'Rules of Language'. Pinker's battle was with those psychologists who continued to maintain that the human mind was a general-purpose learning mechanism—a homogeneous computer. He selected one aspect of language and provided evidence that the mind has a discrete module that operates a grammatical rule to form the tenses for regular verbs. This inflection rule turned 'walk' into 'walked' and 'turn' into 'turned', whereas the tenses of irregular verbs ('sit-sat', 'feel-felt', and 'tell-told', for instance) were generated from memory. He concluded:

Focusing on a single rule of grammar, we find evidence for a system that is modular ... more sophisticated than the kinds of rules that are explicitly taught, developing on a schedule not timed by environmental input, organised by principles that could not have been learned, possibly with a distinct neural substrate and genetic basis.

In her paper, Gopnik had referred to the evidence Marcus Pembrey's group had unearthed for a single gene responsible for the language difficulties experienced by the KE family. Pinker had posited a brain module responsible for a single feature of grammar, Gopnik had discovered it in a single large family pedigree, and it had a simple genetic basis. QED.

In his book The Language Instinct Pinker records the press fanfare in 1992 when Gopnik presented a paper at the annual meeting of the American Association for the Advancement of Science. 'Better grammar through genetics' ran one headline, while another quipped 'Poor grammar? It are all in the genes!'. Yet, hidden away in a much more obscure journal, in 1990, a paper by Jane Hurst, Michael Baraitser, and the staff at the special school in west London painted a very different picture of the KE family's impairment. Entitled 'An extended family with a dominantly inherited speech disorder' (notice the stress on speech rather than language), the paper reported that the family had serious communication difficulties due to a severe verbal dyspraxia, a lack of control over the facial muscles that allow the mouth to articulate the sounds of speech. There were also language difficulties: simple speech and poor comprehension. For instance, 'the girl is chased by the horse' became 'the girl is chasing the horse'. They concluded that the core deficit in the KEs involved both speech and language. So what was really going on in the KEs? Did they suffer from a language impairment, an articulation impairment, or both?

This acrimonious scientific disagreement took the next ten years to resolve. In fact it wasn't until 1995 that Faraneh Vargha-Khadem was even ready to publish her exhaustive description of the KE family trait. It suggested that Gopnik's account of what was afflicting the KE family was far from the whole story.

Although Vargha-Khadem agreed with Gopnik that the affected KEs were unable to inflect verbs with appropriate tense endings, she showed that this inability extended to irregular verbs and was not, as Gopnik had claimed, restricted to regular verbs. However, she also showed that there was an increased occurrence of over-regularizations in verb endings, like 'heared, finded, goed'. These 'tense mistakes', as any parent will tell you, are typical of very young children at the outset of language acquisition, suggesting that the KE family possessed at least some part of the grammatical rule for marking tense.

But comprehending language and constructing language in the brain is only half the problem. To actually produce the sounds of language involves a huge amount of fine motor control of the lips, mouth, tongue, and pharynx; the whole vocal tract. The tongue alone makes up to a hundred movements a second when you are talking. The mouth and lips have to coordinate as well. Vargha-Khadem found that the affected members of the KE family were impaired on this motor control, whether it involved attempts to string sentences together or even to make a string of facial movements like 'close your lips, then open your mouth, then stick out your tongue'. Whatever had gone wrong with the KE family involved language, speech, and mouth and facial neuromuscular control. It was a very broad or diffuse trait, or phenotype, and provided, she reported in a calculated swipe at Gopnik, 'no support for the existence of"grammar genes"'.

Were the KE family's problems with speech and language mirrored by abnormalities in the structure of their brains, and the way they worked? Vargha-Khadem and her colleagues used two imaging techniques to test this: PET (positron emission tomography) and MRI

(magnetic resonance interferometry). PET scans tell you which parts of the brain are being used on a particular task and how hard they are working by, in effect, measuring the rate at which glucose is being burned as fuel for energy. MRI measures the size, shape, and density of particular parts of the brain, giving an anatomical, as opposed to a metabolic, picture. Using these techniques, Faraneh Vargha-Khadem and her colleagues have shown that the brains of affected members of the KE family function very differently to those of normal people. There are clear abnormalities in parts of the brain involved in both motor actions and speech. Of particular interest was the caudate nucleus, a tail-like organ which is part of the basal ganglia, deep in the brain. The basal ganglia are important for fine, coordinated muscular control, exactly what is needed when our vocal tracts make bewil-deringly fast movements as we speak. The caudate contained less grey matter in affected family members than in normal subjects and, gratifyingly, its volume correlated with their performance on tests of coordination of mouth movements, and tests of nonsense word repetition which measure the brain's ability to process arbitrary words and re-assemble the phonemes involved so that they can be pushed back out again as speech. Here, memory cannot be involved because the word is both novel and nonsense.

The PET scans suggested the caudate was not only smaller but was having to work harder in the affected KE family members, especially when they had to repeat words. Another basal ganglia structure, the putamen, had significantly more grey matter than in controls. Other patients with combined pathology of these two areas are unable to properly pronounce words and syllables, and find it difficult to position the face, lips, and tongue to make speech sounds. Elsewhere, the cerebellum (a brain organ at the rear of the head, just above the brainstem), known to be involved in speech articulation, was anatomically abnormal, especially in regions of it that are involved in word generation. There was also abnormality in the planum temporale, a small brain region in the cerebral cortex which is slightly larger on the left than on the right, and is related to verbal ability.

Although there are clear grammatical abnormalities in the affected members of the KE family, there are only three tests for which the affected members of the family never overlap in their scores with non-affected members. These are the tests for complicated, sequential, facial movements ('first open your mouth wide, then close your lips tightly together, then make the "ah" sound'), and for word and non-word repetition. Thus higher regions of the brain were affected, most pointedly Broca's Area, which does the higher-order language processing, and the basal ganglia structures Broca's Area is connected to that control the complicated muscular orchestra we call speech. In this way both speech and language were impaired by the mutation of just one gene.

Vargha-Khadem and her colleagues had painstakingly built up a detailed picture of a complex condition that involved speech, language, and neuromuscular control. Their clear portrait of structural and functional brain abnormality involved many of the parts implicated in aspects of grammar and language and articulation. For the first time, in 1998, Myrna Gopnik changed her tune. In a remarkably conciliatory paper she acknowledged the existence and relevance of Hurst's original observations on the KE family dyspraxia, and Vargha-Khadem's vast corpus of neuroscience. She really had no choice. Vargha-Khadem had exhaustively described significant brain pathology for something Gopnik had chosen to ignore, the palsied speech, and shown it to be a central feature of the KE family disorder. Soon afterwards Gopnik slipped out of the picture and into retirement and simplistic talk of'grammar genes' went with her.

This left the London-based scientists still no nearer to finding the single gene ramifying through the KE family that was responsible for their grammatical defects, poor speech articulation, and lack of facial muscular control. What type of gene could be responsible for all this structural and functional abnormality? Faraneh Vargha-Khadem was getting impatient with Marcus Pembrey. She had spent five years pinning down the phenotype of the KE family disorder in excruciating detail. She was fearful that another scientific group would sweep in and grab the genetics.

Pembrey's money-juggling had finally run out of steam, and he no longer had the funds to move the hunt for the gene forward. A powerful ally was needed. They approached Tony Monaco at the Wellcome Trust Centre for Human Genetics, in Oxford. Monaco was eager to help. A new post-doctoral researcher, Simon Fisher, was put in charge of the investigation. Simon recalls a lazy summer holiday spread-eagled on a beach reading Steven Pinker's book The Language Instinct. He was fascinated by Pinker's account of the KE family, Gop-nik's work, and the evidence for the surprisingly simple genetics at the root of the disorder. 'I thought, "I'd give my eye-teeth to be able to research something as interesting as that", and when I turned up for work at the Wellcome—there it was—waiting for me!'

The search for the gene accelerated. Using a powerful new battery of genetic markers they narrowed the gene's location down to a section of the long arm of chromosome 7 containing some 6 million base-pairs of DNA. A great leap forward but still tantamount to a Chief of Police announcing 'We're closing in on the culprit, he's holed up in a house somewhere in Greater London!' In a constant reminder to the scientific community of the provenance of the gene they gave it the name SPCH 1. By consulting online databases of known DNA sequencestheyknewthisregion, code-named 7q3i, contained at least 70 genes. There was nothing else they could do but painstakingly screen the whole area looking for a tell-tale mutation. This long march down chromosome 7 wasthencutshortbyaspectacularpiece of luck as Jane Hurst once again entered the picture.

As a clinical geneticist, Hurst had become involved in the case of a pregnant woman who had undergone a routine amniocentesis to test for Down's Syndrome. When the genetics laboratory examined cells from her baby, withdrawn from the amniotic fluid, they discovered that a translocation had occurred on chromosome 7. Translocations occur between non-homologous chromosomes (chromosomes that are not one of a pair). Occasionally, bits of one chromosome can become detached and then re-attach themselves by mistake to another. In this way, the DNA on the detached piece of chromosome becomes translocated to somewhere else in the genome. In the case of this baby, CS, the exact point of the break was known. A piece of chromosome 7 had broken off and re-attached itself to chromosome 5 with a piece of chromosome 5 going in the opposite direction. At the time this was discovered the clinical genetics team had no idea whether or not this translocation would cause problems for CS. They kept a watching brief.

Although CS appeared to develop normally as a baby, it was noticed in 1998, when he was about two years old, that his language was delayed and that he had speech articulation difficulties remarkably similar to those of the KE family. He was referred straight back to Jane Hurst at about the time Simon Fisher, Faraneh Vargha-Khadem, and colleagues reported narrowing down the search for the KE family gene to the region of chromosome 7 at 7q3i. Hurst noticed that CS's translocation had occurred right inside the region the Wellcome team had discovered. She immediately picked up the phone to Fisher, 200 yards down the road, and told him: 'I've got the boy who is going to get you your gene.'

And so it proved. When they narrowed the search down to the exact point at which the chromosome breakage in CS had occurred they discovered that it had disrupted a gene for which they could find no previous record. They cross-matched with the KE family and found the same mutated gene in those whose speech and language were affected. The gene was given the name FOXP2. It was a new member of an ancient family of genes that have been steering embryonic development from fungi to mammals for millennia. Careful step-by-step analysis of the DNA sequence of this gene from the KE family, by graduate student Cecilia Lai, revealed one single point mutation, one single DNA nucleotide change—small, but enough to cause cognitive havoc. FOXP2 was certainly not a 'grammar gene' as such, but turned out to play a far more fundamental role in the chain of genetic events that lead to language.

Faraneh Vargha-Khadem has a neat party trick she likes to play on visitors who come to talk to her about her research, a trick that illustrates the very profound role of FOXP2. She plays them videotapes of the tortured speech of the KE family and then follows that up with a clip of an interview with a teenager called Alex. As Alex recounts a Lonely Planet-type backpacking holiday around Europe it gradually dawns on you that, while his powers of description are adequate and his use of English reasonable, his speech more resembles that of an

8 or 9 year old, than someone who is clearly at least 18. The reason? Alex had severe brain disease when he was very young and at age

9 he underwent a complete hemispherectomy: the entire left side of his brain had to be removed. He had never spoken in his life. After the operation he learned to speak even though his brain had to make use of structures in the right hemisphere to replace the lost areas normally used for speech and language in the left, and even though the critical period in brain development for language acquisition, the 'window of opportunity' during early childhood, had long been passed. On the one hand, the human brain is so awesomely plastic and adaptable that even radical surgery fails to de-rail speech and language, yet a tiny point mutation in a single gene, FOXP2, leaves affected individuals with dramatic impairments in speech articulation and language comprehension and production from which they never recover, despite decades of speech therapy.

The reason for the widespread effects of FOXP2 lies in the fact that it is a master-controller gene, one of a number of genes that produce transcription factors. These are proteins that bind to regions of the DNA of a number of downstream target genes to regulate them— turn them on or off. The mutation in the KE family occurred in a region of the gene that codes for a short but crucial section, only 80

to 100 amino-acids long, of one of these regulatory proteins. FOXP2 controls the activity of a genetic orchestra that, in some way, sculpts the neural pathways of the brain for the purpose of language and speech articulation. In an elegant piece of research, which finally nailed down the FOXP2 story, it has been shown that the sites in the brain where FOXP2 is normally active agree closely with the sites of brain pathology noted by Faraneh Vargha-Khadem and her team, using brain-scanners. Disruption of FOXP2 in these areas was associated with the brain pathology that caused the KE family disorder. This was the real QED!

Soon after the discovery of FOXP2, Tony Monaco, at the Wellcome Institute, had an important visitor. Svante Paabo is head of the Molecular Genetics Department at the Max Planck Institute for Evolutionary Anthropology in Leipzig. He was the first scientist to clone Neanderthal DNA, some 35,000 years old, from tissue retrieved from the long bones of a skeleton from a German cave. Paabo is a world leader in the attempt to single out the genes that make us human and one of his key strategies is to comb the world of psychiatric and psychological genetics for interesting mutations. Paabo had arrived on a fishing expedition. Monaco had FOXP2 to offer and they agreed that Paabo's laboratory should look to see if it had an interesting evolutionary tale to tell. Paabo entrusted the work to his research lieutenant Wolfgang Enard, who compared the amino-acid sequence of the FOXP2 protein in mouse, rhesus macaque, orang-utan, gorilla, chimpanzee, and human. He discovered that it had hardly changed at all over the 130 million years of evolution that separate the common ancestor of chimps and humans from the common ancestor that led to the mouse. Chimp, gorilla, and macaque sequences were identical and only one amino-acid change separated them from the mouse. However, a further two amino-acid substitutions have occurred over approximately 6 million years in the hominoid line that led to us. The pace of evolution had suddenly picked up. They looked at patterns of variation in the gene throughout a number of contemporary human groups from Africa, Europe, New Guinea, Asia, and South America, and their mathematical analysis suggested that these two amino-acid changes had been the target of natural selection. The two mutations had occurred and become fixed in human populations probably as recently as 200,000 years ago, the approximate date when modern humans, Homo sapiens sapiens, arrived on the scene.

The evolutionary tale of FOXP2 dovetails nicely with the evolution of our human lineage. But it gets us no nearer a precise description of exactly how FOXP2 is implicated in speech and language, grammar and articulation. The exact nature of the two amino-acid substitutions unfortunately tells us very little. Simon Fisher speculates that it may have caused a subtle change in the regulatory protein that decreased its ability to switch other genes on or off. Puzzlingly, one of the two 'unique' human lineage amino-acid substitutions also turns up in the great cats, and, with the exception of the Chronicles of Narnia, talking lions are thin on the ground. How the one surviving 'unique' amino-acid substitution can give rise to speech and language, something the great apes so conspicuously lack, remains, for the moment, a mystery. Since it is obviously unethical to excise the gene from humans and see what happens, Enard and Fisher are collaborating on a series of experiments with 'knock-in' mice where the human variant of the FOXP2 gene is incorporated into the mouse genome and they can see what changes, if any, occur. Precisely what changes might be expected is not clear. 'We are not expecting the mice to tell us,' jokes Enard.

In the meantime, some exciting new information on FOXP2 has arrived, thanks to studies of convergent evolution in birds and humans. (Convergent evolution occurs when evolution solves a problem the same way in two entirely unrelated species. For instance, the eye has arisen independently many times throughout the animal kingdom.) Bird-song has many similarities to human language: the basic 'vocabulary' of a bird-song is learned by imitating older family members, there is a critical period in development when young birds can acquire it, and it is plastic in that basic patterns of song are honed by each bird to personal taste. It turns out that the expression of FOXP2 in bird brains is particularly strong in those parts of the brain involved in song, areas which are comparable to the parts ofthe brain in which these two key genes are expressed in humans. The evidence has come from Stephanie White, in the United States, and Constance Scharff, in Germany, who have been working with zebra finches and canaries.

Zebra finches are fully mature by 90 days old, but the young male birds (only male zebra finches sing) have already acquired their model song by 40 days. There is a sensitive period for song acquisition between about 20 and 35 days during which the young finches start off, like human babies, by babbling, and proceed to more coherent song. They are tutored by their fathers, other older- generation males, and, occasionally, older siblings, and first achieve a copy, or rendition, of the tutor's song. However, this basic framework is then individualized by adding notes and making other changes.

Scharpf and her colleagues discovered that, while FOXP2 is expressed in a broad range of brain structures, its expression is particularly intense in a part of the basal ganglia called Area X, which is essential for vocal learning. This heightened activity begins when the chicks are between 35 and 50 days old, just as vocal learning occurs, and continues throughout life. In canaries, song type is seasonal. For parts of the year canary-song is very stereotyped, but at other times the song becomes unstable and plastic; its owner is experimenting. During these creative periods FOXP2 expression soars in Area X. The two research groups also recorded strong FOXP2 activity in a variety of structures that connect the basal ganglia (the location of Area X) with areas responsible for hearing and controlling the throat movements associated with bird-song.

Just as in the human work, these researchers have found that FOXP2 has a wide role to play in developing bird and human brains, but a specialist, more prominent, role in those parts of the brain that coordinate hearing and singing/talking. They conclude that these genes create a 'permissive environment' in parts of the brain within which vocal learning can evolve if, and once, other factors come into play.

Information from birds, together with Faraneh Vargha-Khadem's imaging studies on the affected members of the KE family, show us that the articulation of speech and the comprehension of speech use the same circuits in the brain. Language is not abstract, but grounded in and restrained by the physical process of articulation. Words are what our mouths can make them. As Vargha-Khadem puts it, language cannot be free-floating in the brain. As a comparison, she says, all humans have the inherent ability to walk, but it cannot be realized without the feet and legs to go with it, and, more precisely, the sensorimotor systems (those neural systems that control muscular responses to sensory information) in the brain that make them work. In the same way, language must have the appropriate sensorimotor system to map onto—the right mechanism in the brain with which it can fuse.

In a recent scientific paper, commenting on the role of FOXP2 in speech and language, the celebrated linguistic scientist Philip Lieberman makes the telling point that Broca's aphasia never occurs thanks to damage to Broca's Area, in the cortex, alone. It requires damage to the basal ganglia as well before the classic speech and language deficits show up. This failure to mention the role of these subcortical 'primitive' brain structures, Lieberman argues, goes right back to Broca himself in 1861, when he failed to point out that the key patient on whom his description of Broca's aphasia was based had extensive damage to basal ganglia structures as well as the eponymous cortical region. It was rather like praising the captain on the bridge for a successful ocean crossing and forgetting the sweaty endeavours in the engine room.

Broca's modular view of the 'seat of language' being situated in the higher-order 'advanced' cortex was subsequently followed by

Chomsky and Pinker. More modern brain research, says Lieberman, has shown that circuits looping out of the basal ganglia, into the cortex, and back again, are involved in the fine coordination of muscular actions that enable us to walk, dance, and contort our faces. They also allow us to make novel combinations of a finite repertoire of muscular actions of the vocal tract to form an infinite variety of words, and allow our brains to make an infinite number of novel sentences out of a finite number of words, using a finite number of syntactical rules. To imagine what he means, remind yourself of how many words fill the New Oxford Dictionary, all comprised of different mixtures of the twenty-six letters of the alphabet, or how thehugenumberofdifferentproteinsinthehumanbodyareall made by different combinations of only four nucleotide molecules in the DNA double helix.

Lieberman calls the basal ganglia a 'sequencing engine' deep in our brains, which can operate in all these contexts. Lieberman doesn't dispute that the content of thought, memory, and ideas are higherorder functions of the brain, but reminds us that it is these older brain structures that make speech, syntax, and creative thought possible. This is the 'permissive environment' evoked by those scientists studying bird-song, and we need to know how it is that subtle DNA sequence changes in human FOXP2 have led to crucial changes here to give us something apes so palpably lack, the power of speech and language.

Not surprisingly, several groups world-wide are now trying to discover the array of 'downstream' genes that seem to be under the control of FOXP2 in the brain. Although FOXP2 is not directly implicated in any other type of language disability, an international group of scientists, drawn mainly from the Wellcome Centre for Human Genetics in Oxford, has discovered one member of this 'downstream orchestra', a gene called CNTNAP2, whose activity is under direct control by FOXP2. CNTNAP2 is active in the developing cortex of the brain and variants of the gene have been associated with specific language impairment and language delays in children with autism. Tworecentpiecesofresearchbolsterthetwinideasof FOXP2 involvement in language as a form of social communication, and Lieber-man's idea that it is active in a 'sequencing engine' in the brain, responsible for rapid articulation of sounds.

In 2005, a group of scientists led by Joseph Buxbaum, of the Mount Sinai School of Medicine in New York, did the opposite of 'knocking in' the human FOXP2 gene to mice. They 'knocked out' murine FOXP2. When both copies of the gene were tampered with the mice were very severely affected, their movement was badly impaired, and they died prematurely. However, when only one copy of FOXP2 was disrupted there was only mild developmental delay in the cerebellum and they were left with a communication block. When lifted away from their mothers, the pups failed to make any ultrasonic distress calls. When Buxbaum's team analysed the vocalization patterns and the bandwidths of the vocalizations they concluded that much of the machinery in the vocal tract and brainstem associated with vocalizing was normal. The problem lay with subtle changes in the cerebellum. This makes complete sense because the formation of speech sounds is a motor activity and the cerebellum integrates this fine motor control. The cerebellum was one of the brain areas most affected in the KE family.

In late 2007, a group of Chinese scientists together with Stephen Rossiter, from Queen Mary College, London, reported on some fascinating research that connected mutations in the FOXP2 gene to echolocation in bats. Received wisdom, they said, was that FOXP2 was extremely conserved throughout the animal kingdom— there was no sequence variation—until you get to the comparison between humans and chimpanzees, where, as we know, there are two amino-acid substitutions in humans. This has led, they argue, to exclusive consideration of the role of FOXP2 in the context of the emergence of language—which humans have and chimps don't. However, when you compare echolocating with non-echolocating bats, they report, you find a range of sequence differences in FOXP2, suggesting that the correct context to explore the action of FOXP2 is sensorimotor coordination, not language per se—exactly the same conclusions reached by Varga-Khadem.

Echolocation is an extremely sophisticated business, being used for orientation and prey capture at high speeds. Many insects use very unpredictable, almost chaotic, flight patterns called protean behaviour, designed to lead predators a merry dance. To prevent themselves banging into things, and to fill their bellies, bats emit echolocation pulses at up to 200 sounds per second, decipher the bounce-back in milliseconds, and make necessary motor adjustments to their flight controls within the same tiny time-frame. Receiving and transmitting this ultrasonic barrage of sonar information requires an extraordinary and complex coordination of hearing, and nasal and orofacial movements, since the ultrasonic signals bats use for navigation are formed in the larynx. Bats, like song-birds and humans, also exhibit vocal learning. It makes language look like child's play.

When the scientists sequenced the entire FOXP2 gene in a range of echolocating and non-echolocating bats they found many different amino-acid substitutions in the resulting protein. These changes were not uniform along the gene but clustered in two of the exons— or coding sequences—exons 7 and 17. In several species of echolo-cating bats there was clear evidence for accelerated evolution. Leaf-nosed bats and all vesper or evening bats showed different mutations in exon 7; several species of echolocating bat shared amino-acid substitutions with whales—which also exhibit vocal learning; and in exon 17, amino-acid variation was considerable in all echolo-cating bats with up to eight different substitutions noted. The most divergent species was the slit-faced bat, Nycteris, which is commonly found throughout Malaysia and Indonesia. Interestingly, these bats emit very short multi-harmonic sonar calls and target their mainly insect prey, not by sonar bounce-back, but by eavesdropping on the tiny sounds they make. They conclude that echolocation in bats has involved the evolution of FOXP2 and that downstream cascades of genes under its regulatory control are involved in the development and maintenance of this very complex form of sensorimotor coordination. Humans, song-birds, and bats have at least one thing in common not shared with chimpanzees: complex forms of communication requiring fast and dexterous motor coordination of the muscles associated with sound (or ultrasound) production from the larynx. Examining FOXP2 variation more carefully across a broader range of species will help to decide whether its role is specific to vocal learning, articulation, or the complexity of the signals animals emit. This more circuitous route may lead us to a better understanding of thenatureoflanguageandwhywehumanscantalk, whereaschimps cannot.

+2 0

Post a comment