[Language] certainly is not a true instinct, for every language has to be learnt. It differs, however, widely from all ordinary arts, for man has an instinctive tendency to speak, as we see in the babble of our young children; whilst no child has an instinctive tendency to brew, bake or write.

EVOLUTION’S RAW MATERIAL is the gene pool of a species and the mutations that arise at random in those genes. This formidable constraint means that an organ or faculty cannot be created out of nothing; it can only be shaped, by gradual stages, out of some existing structure, and each of those intermediate stages must confer advantage in its own right.

One reason why human language is so deeply puzzling to biologists is that it seems to defy this rule. It is a vibrant, fully developed faculty in people, but is not possessed, even in rudimentary form, by any other species. It seems to have popped up into the recent human line from nowhere.

The origins of language would perhaps seem somewhat less mysterious if our archaic cousins, the Neanderthals and Homo erectus, had survived to tell what kind of communication skills they commanded. But these branches of the hominid tree have been docked, leaving only one survivor.

Primatologists have therefore looked for the roots of the language faculty in social primates such as apes and monkeys. These species do indeed possess many of the neural systems that are needed in support of language. They can make a wide range of elaborate sounds. They have acute senses of hearing with which to perceive and analyze the sounds made by members of their own species. As for thought, there is no doubt that the social primates are capable of quite elaborate cognitive processes, such as those required in keeping tally of who one’s relations are, who owes one favors, and where one stands in the social hierarchy.

But despite possessing much of the neural equipment for speech, monkeys and apes simply lack the ability to translate their thought into anything resembling human language.

Several primate species have communication systems of considerable sophistication. Gelada baboons have 22 different kinds of call, and gorillas have been recorded using some 30 different gestures.³⁴ One of the best studied animal communication systems is the repertoire of alarm calls uttered by the vervet monkeys of East Africa. Vervets lead a perilous existence, at constant risk from eagles, leopards and snakes, and they possess a distinctive warning call for each. When researchers record one of these calls and play it back to other vervets, the monkeys reliably scan the skies in response to the eagle call, look down at the ground at the snake call, and leap into bushes at the leopard call.

In an interesting link with human language, the basic mechanisms of the vervet’s calls seem to be innate but are refined by learning. Baby vervets will give the eagle call in response to almost anything airborne, including falling leaves, but by the time they are adults the call has become focused on eagles, particularly the martial eagle, while nonpredatory birds like vultures are ignored.³⁵

It is tempting to suppose that vervets have therefore developed a word for eagle, but that is not really the case. A vervet cannot combine two of its cries to state that, in its opinion, “Eagles are more dangerous than leopards!” Its calls can be used only as one-note alarms to warn that “An eagle is coming, take cover!” or “Leap—it’s a leopard!”

Besides appearing to lack precise words for things, animals also lack the ability for syntax. Though capuchin monkeys seem to obey an ordering rule in their calls (for example, call A is made before calls B and C but never after them), the meaning of such ordered calls, if any, is not yet clear to researchers.³⁶ Strenuous efforts have been made to teach language to chimpanzees. The first attempts focused on training the chimps to make humanlike sounds. Then, when the unsuitability of their vocal apparatus was accepted, they were taught to communicate in sign language. Chimps can learn a number of signs—about 125, according to their trainers, more like 25, according to skeptics—but there is no consistent evidence that they use the order of the signs to confer meaning, as is the essence of human language. Typical utterances of Nim Chimpsky, a chimp trained by Herbert Terrace of Columbia University, were “Me banana you banana you me give,” and “Give orange me give eat orange me eat orange give me eat orange give me you.” “The chimp’s abilities at anything one would want to call grammar were next to nil,” concludes Steven Pinker of Harvard University.³⁷

Still, evolution’s design principle is continuity, so there must have been some neurological structure in the mammalian brain that was adapted to generate the combinatorial systems of vocabulary and grammar, just as the mammalian ear and voice box were adjusted to analyze human voices and generate human speech sounds. In an unusual alliance, the animal communication experts Marc Hauser and Tecumseh Fitch recently joined with the linguist Noam Chomsky to propose that the human capacity for syntax might have evolved out of an animal brain module designed for some other purpose, such as navigation.³⁸ Their argument is that the essential feature of language is recursion, the ability to embed one phrase inside another in an indefinitely long chain. Recursion may also be a feature of faculties like navigation that require an animal to remember how to get from A to D, with an excursion to B and C if the way is blocked. If the genes that specify the brain’s navigation module were accidentally duplicated, the spare set would be free to evolve and perhaps acquire the function of encoding thought into language.

The Nature of Language

Many people think that thought would be impossible without language, and that the two are pretty much the same. Others equate language with speech. In the view of linguists, neither proposition is true. Animals may have quite rich thought processes—chimpanzees certainly know the position of all the individuals in their hierarchy and who must be recruited in a conspiracy—but are unable to put their thoughts into words. And speech is just one modality for language, which can also be written, or conveyed as signs, as in American Sign Language. Linguists regard this and other signing systems as proper languages with the same properties as spoken languages, including a fully developed syntax, or set of grammatical rules.

In the view of linguists, language is neither thought nor speech but rather a system for translating thought into a physical output, usually speech or writing. The brain behaves as if it were performing this translation process with a pair of combinatorial systems, one of which generates vocabulary, the other syntax.

The combinatorial system for vocabulary is a remarkable solution to a difficult problem. Many animal species communicate with a set of calls, each of which has a specific meaning. If the same principle were followed in human language, the calls after a certain number would start to merge into one other and become very hard to distinguish. But natural selection has somehow hit on a way of generating infinite variety, by basing the vocabulary system on a very small set of individually discrete sounds. The sounds can be joined in a limitless number of combinations, and any of these compound sounds can be arbitrarily associated with a meaning to form words. The system is called combinatorial because it is based on combining different elements to generate words.

The combinatorial system for syntax is tricky to describe because it seems to perform several different though related tasks. The original fix on it came from Noam Chomsky’s insight that there must be a universal system in the human brain for allowing children to learn the grammatical rules of whatever language they hear spoken around them. Languages have many different rules of grammar, but all seem to be variations on the same model. Chomsky called this learning machinery Universal Grammar, but the phrase is also used to refer to the basic design underlying all grammars.

This proposal still attracts objections from researchers who believe the mind is a general purpose learning system, a blank slate with no prepro gramming or genetically based circuitry dedicated to particular behaviors such as the faculty for language. It’s certainly true that human behavior seems to be under conscious control to a far greater extent than is that of other animals. But equally it is clear that many behaviors in animals are genetically guided. In some animals, like the laboratory roundworm, biologists have already learned how to alter certain genes and induce a different behavior. It’s reasonable to assume that there is a genetic basis for much human behavior, particularly such basic but highly complex faculties as learning a language or recognizing faces.

In the case of language, the combinatorial systems for vocabulary and syntax are so sophisticated that it seems unlikely an infant could quickly learn them from scratch. It would seem far more efficient for evolution to embed the general ability for learning language in the brain’s neural circuitry. As Darwin observed, the ability to learn the spoken language seems instinctual, but the ability to write is not, which is why it must be learned so laboriously in school. In support of the view that the basic elements of language are innate, a human gene that seems fairly specific to language has recently been identified, as discussed below.

The fact that children around the world learn languages so easily, and at the same stage of development, points strongly to the unfolding of a genetic program as the children reach a certain age. Chomsky asserted that Universal Grammar was innate, and indeed the language-learning machinery seems to be one of the many developmental programs that are wired into the genes and scheduled to unfold at a given time.

But Chomsky and other theoretical linguists have been less interested in the question of what evolutionary stimulus might have prompted the evolution of language. No full length article about the evolution of language appeared in the Linguistic Society of America’s journal Language until 2000. “Why linguists have tacitly accepted just such a self-denying ordinance should be a topic of some interest to sociologists of science,” writes Derek Bickerton of the University of Hawaii, one of the few linguists to have explored the origin of language.³⁹

Several leading linguists blame Chomsky for the neglect. His proposed system of Universal Grammar was such a complicated mechanism that his critics argued there was no way it could have evolved, since it would have been useless until the full structure was in place. This was a misguided criticism since evolution explains very well how enormously complex organs such as the eye or ear have evolved. Nonetheless, Chomsky, rather than debate the point, discouraged any discussion of evolution, several leading linguists now say. “Opponents of UG argue that there couldn’t be such a thing as UG, because there is no evolutionary route to arrive at it,” writes Ray Jackendoff. “Chomsky, in reply, has tended to deny the value of evolutionary argumentation.” ⁴⁰

“To the extent that Chomsky has been willing to speculate on language origins at all, his remarks have only served to discourage interest in the topic among theoretical linguists. He has adamantly opposed, for example, the idea that the principles of UG arose by virtue of their utility in fostering the survival and reproductive possibilities of those individuals possessing them,” writes Frederick Newmeyer, a linguist at the University of Washington, Seat tle.⁴¹ These two critics are not without standing; Newmeyer was president of the Linguistic Society of America in 2002, Jackendoff the following year.

Chomsky denies that he ever discouraged people from studying the evolution of language and says that his views have been misinterpreted. “I have never expressed the slightest objection to work on the evolution of language,” he says. He outlined his views briefly in lectures 25 years ago but left the subject hanging, he said, because not enough was understood. He still believes that it is easy to make up all sorts of situations to explain the evolution of language but hard to determine which ones, if any, make sense.⁴²

For outsiders looking in, it’s hard to understand why linguists such as Newmeyer and Jackendoff would blame Chomsky for the entire profession’s neglect of evolution, given that his colleagues, as independent academics, were presumably capable of thinking for themselves. However, Chomsky did have a significant impact on what others thought, says Steven Pinker, in part because of his intellectual stature and in part because of an aggressive style of debate that polarized the whole field.

“Why should one man’s opinion count for so much?” Pinker asks. “The fact is that Chomsky has had, and continues to have, an outsize influence in linguistics. He has rabid devotees, who hang on his every footnote, and sworn enemies, who say black whenever he says white. This doesn’t leave much space for linguists who accept some of his ideas (language as a mental, combinatorial, complex, partly innate system) but not others (the baroque and ever-changing technical details of his theory of grammar, his hostility to evolution or any other explanation of language in terms of its function).”⁴³

Like other social scientists, linguists have not made a habit of looking to evolution for explanations, even though it is the bedrock theory of biology. Pinker was one of the first linguists to do so. With Paul Bloom, he wrote an influential article with a self-declared “incredibly boring” goal. Its purpose was to explain to linguists that, contrary to the views of Chomsky and the science historian Stephen Jay Gould, “human language, like other specialized biological systems, evolved by natural selection.”⁴⁴

Pidgins, Creoles and Sign Languages

One of the Chomskyans’ problems with evolution of language, that language is too complex to have halfway steps, has been addressed by Derek Bickerton. Bickerton became interested in the subject through his study of a fascinating language phenomenon, the development of creoles from pidgins. Pidgins are languages of limited vocabulary and minimal grammar, usually invented by two populations who have no language in common. The children of pidgin-speakers do something very interesting: they spontaneously develop the pidgin into a fully fledged language with proper grammatical rules. These developed pidgin languages are called creoles.

It seemed to Bickerton, as he studied Hawaiian creoles, that their development offered an insight into the evolution of human language. The first language was pidgin-like, he suggested, consisting mostly of vocabulary, and syntax was grafted on later. Several possible remnants of this proto-language still survive. If children are not exposed to language in early childhood, when their Universal Grammar machine is switched on and primed to learn, they may never learn any language properly. This happens very rarely, in the case of feral children allegedly brought up by animals, or when pathological parents imprison their children in the house and refuse to speak to them. Genie, a 13-year-old California girl, was found in 1970 wandering the streets with her mother. The two had escaped from a house where Genie had been penned in a bedroom from the age of 18 months and denied conversation. After her rescue, intense efforts were made to teach her to talk, but she never acquired fully grammatical language. Her utterances were stuck at the level of sentences like “Want milk,” or “Applesauce buy store.”⁴⁵

Even this primitive form of language could have been extremely useful to an early human society. Other possible echoes of the inferred proto-language can be heard in syntax-free utterances such as “Ouch!” or the more interesting “Shh!,” which requires a listener.⁴⁶

Recently linguists have developed a new window into the innate basis of syntax through a remarkable discovery—the detection of two new languages in the act of coming to birth. Both are sign languages, developed spontaneously by deaf communities whose members were not taught the standard sign languages of their country. One is Nicaraguan Sign Language, invented by children in a Nicaraguan school for the deaf. The other, Al-Sayyid Bedouin Sign Language, was developed by members of a large Bedouin clan who live in a village in the Negev desert of Israel.

The Nicaraguan case began when children were brought to a school for the deaf founded in 1977 by Hope Somosa, the wife of the Nicaraguan dictator. ⁴⁷ Instructors noticed that the children had learned little from their Spanish lessons but had developed a system of signs for talking to one another. Each generation of kids taught it to the next, and the language has rapidly evolved from a set of gestures into a sophisticated language with its own syntax.

The Al-Sayyid clan consists of some 3,500 people descended from a single founder who arrived 200 years ago from Egypt and married a local woman. Since the third generation, marriage within the clan has been encouraged, so there is a considerable level of inbreeding. Two of the couple’s five sons were deaf, as are about 150 members of the community today. The clan’s village is isolated in part by geography and even more by social barriers, since other Bedouin look down on them. Its deaf members did not go to school until recently and so were not exposed to either Israeli or Jordanian sign languages. They developed their own, and the language is also used by their hearing relatives to communicate with them.⁴⁸

According to a famous story in Herodotus’s History, an Egyptian king tried to ascertain the nature of the first language by isolating two children from birth and waiting to see in what tongue they first spoke. Study of the two new sign languages confirms that King Psammetichus’s experiment was misconceived: it is not specific words that are innate, but rather the systems for generating syntax and vocabulary. Among both the Nicaraguan schoolkids and the Al-Sayyid clan, a spontaneous sense of syntax has developed, specifically the distinction as to whether a word is the subject or object of a verb.

In addition, the Al-Sayyid signers have developed the preference for a specific order of words in a sentence, that of subject-object-verb. The Nicaraguan children, in contrast, have developed the signed equivalents of case endings for words. Since these indicate whether a word is subject or object, word order is not so important and keeps changing with each cohort of children.

The apparently spontaneous emergence of word order and case endings in the two sign languages strongly suggests that the basic elements of syntax are innate and generated by genetically specified components of the human brain. The Al-Sayyid sign language has been developed through only three generations but some signs have already become symbolic. The sign for man is a twirl of the finger to indicate a curled mustache, even though the men of the village no longer wear them. Change is also brisk in Nicaragua. At first the children represented the number twenty by flicking the fingers of both hands in the air twice, says Ann Senghas, a linguist who has been studying their sign language for 15 years. But the sign was too cumbersome and has now been replaced with a form, signable with one hand, that looks nothing like a 20 but can be signed fast.⁴⁹

Sign languages emphasize an often overlooked aspect of language, that gesture is an integral accompaniment of the spoken word. The human proto-language doubtless included gestures, and could even have started with gestures alone. Michael Corballis, a psychologist at the University of Auckland in New Zealand, argues that language “developed first as a primarily gestural system, involving movements of the body, and more especially the hands, arms and face.”⁵⁰ Speech would have evolved only later, he believes, because considerable evolutionary change had to occur to develop the fine muscles of the tongue and other parts of the vocal apparatus.

Corballis’s idea has several attractive features. It would explain why word and gesture are so well integrated, and why people even gesticulate when talking on the telephone despite the fact that their listeners cannot appreciate the performance. But critics of the idea note that gesture-based languages would be useless in the dark and that they require those conducting a conversation to be looking at each other all the time. Spoken language suffers from neither constraint.

Evolutionary Pressures for Language

Once language started, whether in the form of word or gesture or both, its further evolution would doubtless have been rapid because of the great advantages that each improvement in this powerful faculty would have conferred on its possessors. Even while still in its most rudimentary form, language would have made possible a whole new level of social interactions. Precise and unambiguous thoughts could at last be shared among members of a community, whether for making alliances, indicating intention, describing people and places, or transmitting knowledge. Moreover each small improvement in the overall system, whether in precision of hearing or articulation or syntax formation, would confer further benefit, and the genes underlying the change would sweep through the population.

But easy as it is to see how a simple form of language might have evolved into a complex one, that doesn’t answer the question of what particular stimulus brought language into being in the first place.

Language now plays so many roles in human society that it’s hard to arrange them in some hierarchy and say one role was the root and the others its branches. But evolutionary psychologists have come up with several interesting suggestions about the possible pressures for language to evolve. Robin Dunbar at the University of Liverpool in England has proposed a “social grooming” theory of language. He notes that monkeys and apes spend an inordinate amount of time grooming each other’s fur. This activity, besides curbing parasites, serves to cement social relationships. But social grooming sets a limit on the size of a monkey group, because members will have no time to search for food if there are too many acquaintances whose fur must be rubbed the right way.

In practice, different monkey species spend varying amounts of time on grooming one another, up to a maximum of 20% of their waking day, and this is among species whose typical group size is about 50 members. The maximum time available for social grooming, Dunbar argues, has effectively capped the size of monkey social groups at 50 members. How then did the typical size of hunter-gatherer groups grow to 150 members, a number that would in principle require everyone to spend 43% of their waking hours on social grooming, or its human equivalent? Because of language, Dunbar suggests. Language is so much more efficient a way of establishing and confirming social bonds that the requisite amount of social grooming could be cut way back. In a wide range of human societies, it so happens, the amount of time people spend in social interaction, or conversation, is 20%. The driving force behind the evolution of language, in Dunbar’s view, was the need to bond people in larger social groups.⁵¹

A quite different explanation has been advanced by the evolutionary psychologist Geoffrey Miller. He believes that sexual selection—Darwin’s theory that the peacock’s tail is the evolutionary product of peahens’ choices—is what has driven the evolution of language. Just as the richness and symmetry of the peacock’s tail signals its freedom from parasites, so eloquence and articulate speech signal the quality of an individual’s mind, and will be highly favored by both men and women in their sexual partners. Language is a device that lets us learn about potential mates more thoroughly than any other method, Miller writes.⁵²

The Dunbar and Miller hypotheses are both evocative and each may hold some measure of truth. But it’s not clear if either really accounts for the richness and precision of language. Most adult speakers of English have a vocabulary of 60,000 words, though the top 4,000 words account for 98% of conversation. Does one really need 60,000 words, or even 4,000, for the purposes of social grooming, or even impressing one’s inamorata? Miller’s answer is that excess is the hallmark of sexual selection—once selection has started, the character under selection is taken to extremes, like the stag’s enormous antlers.

But for linguists, the essence of language is meaning and communication, and it seems unsatisfactory to explain its evolution on any other grounds. Pinker argues one should take into account the new ecological niche that humans had moved into, which was in fact a knowledge-laden environment requiring a wealth of new information about plants and animals, about how to make tools and weapons, and about goings-on in one’s own society. People’s longer life span made it worthwhile to gather information and transmit it to one’s children and grandchildren. “Language,” Pinker says, “meshes neatly with the other features of the cognitive niche. The zoologi cally unusual features of Homo sapiens can be explained parsimoniously by the idea that humans have evolved an ability to encode information about the causal structure of the world and to share it among themselves. Our hy persociality comes about because information is a particularly good commodity of exchange that makes it worth people’s while to hang out together.”

Pinker concludes that know-how, sociality and language are three key features of the distinctively human lifestyle and that the three factors coevolved, each acting as a selective pressure for the others.⁵³

It would be easier to pinpoint the most likely stimulus for the evolution of human language if one could identify when language emerged. Obviously the joint human-chimp ancestor did not speak, or chimps would too. And all human races can speak equally well, so that fully articulate, modern language must have evolved before modern humans left Africa. This means language would have emerged after 5 million years ago and before 50,000 years ago. Paleoanthropologists have made strenuous attempts to pin down the development of language through anatomy, by looking at the shape of the brain as implied by interior casts of old skulls, or features such as the hy oid, a U-shaped bone that supports the tongue muscles, and the hypoglossal canal, a passageway through bone for the nerve bundle that wires up the tongue muscles. But these studies have not yet brought a great deal of clarity to the problem.

Paleoanthropologists have tended to favor the idea that language started early, with Homo erectus or even the australopithecines, followed by slow and stately evolution. Archaeologists, on the other hand, tend to equate full-fledged modern language with art, which only becomes common in the archaeological record some 45,000 years ago. Their argument is that creation of art implies symbolic thinking in the mind of the artist, and therefore possession of language to share these abstract ideas.

Other archaeological facts favor a late start for language. To look at the rough stone tools of the Olduwan (made between 2.5 and 1.7 million years ago) they seem to be just chipped pebbles, made with no particular design in mind. But the tools of the Upper Paleolithic, which began 45,000 years ago, are precisely shaped and so well differentiated from each other that it seems plausible their makers had a different word for each, and therefore had language. “It is as though Upper Paleolithic flint workers were saying ‘This is an end-scraper: I use it as an end-scraper, I call it an end-scraper and it must therefore look like an end-scraper,’” writes the archaeologist Paul Mellars. He argues that the makers’ evident emphasis on the precise visual shape of their tools “is probably exactly what one would anticipate if Upper Paleolithic groups had a much more complex and highly structured vocabulary for the different artifact forms.” Given their much cruder tool kit, the Neanderthals might also have had language, Mellars thinks, but with a much simpler vocabulary.⁵⁴

If fully articulate modern language emerged only 50,000 years ago, just before modern humans broke out of Africa, then the proto-language suggested by Bickerton would have preceded it. When might that proto-language first have appeared? If Homo ergaster possessed proto-language then so too would all its descendants, including the archaic hominids who reached the Far East (Homo erectus) and Europe (the Neanderthals). But in that case the Neanderthals, to judge by their lack of modern behavior, appear never to have developed their proto-language into fully modern articulate speech. That might seem surprising, given the advantage any improvement in the language faculty would confer on its owner, and the rapidity with which language might therefore be expected to evolve. So perhaps the Neanderthals didn’t speak at all.

Discovery of a Gene for Language

A remarkable new line of inquiry bearing on the origins of language has recently been opened up by the human genome project. This is the discovery of a gene that is intimately involved in many of the finer aspects of language. The gene, with the odd name of FOXP2, shows telltale signs of having changed significantly in humans but not in chimps, exactly as would be expected for a gene serving some new faculty that had emerged only in the human lineage. And, through the ability of genetics to reach back into the distant past, the emergence of the new gene can be dated, though at present only very roughly.

FOXP2 came to light through the discovery by Jane Hurst, an English geneticist, of an unusual London family whose existence she reported in 1990. The family consists of three generations. Of the 37 members old enough to be tested, 15 have a severe language deficit. Their speech is hard to understand, and they themselves have difficulty comprehending the speech of others. If asked to repeat a phrase like “pattaca pattaca pattaca,” they will stumble over each word as if it were entirely new. They have difficulty with a standard test of the ability to form past tenses of verbs (“Every day I wash my clothes, yesterday I_____my clothes”; four-year-olds will say “washed” as soon as they get the idea). They have problems in writing as well as speaking. The affected members of the family have been given intensive speech training but mostly hold jobs where not much talking is required. “Their speech is difficult to understand, particularly over the telephone, or if the context is not known. In a group of family members it is hard for you to pick up the pieces of the conversation, which is difficult to follow because many of the words are not correctly pronounced,” says Faraneh Vargha-Khadem of the Institute of Child Health in London.⁵⁵

Some of the first linguists to study the affected family members believed their problem was specific to grammar but Vargha-Khadem has shown that it is considerably wider. Affected members have trouble in articulation, and the muscles of their lower face, particularly the upper lip, are relatively immobile.

It could be argued that their defect stemmed from some general malfunction in the brain, which was not specific to language. But the IQ scores of the affected members, though low, fell in a range (59 to 91) that overlapped with that of the unaffected members (84 to 119).⁵⁶ The core deficit, Vargha-Khadem concluded, is “one that affects the rapid and precise coordination of orofacial [mouth and face] movements, including those required for the sequential articulation of speech sounds.”⁵⁷

The affected members of the KE family, as it is known, have each inherited a single defective gene from their grandmother. They provide the results of an experiment that no one would even contemplate doing in humans, but which nature has performed nonetheless—what happens if you disable a critical speech gene? And the one disabled in the KE family seems to operate at such a sophisticated level that it looks as if it were one of the last genes to be put in place as the faculty of language was perfected.

In 1998 a team of geneticists at Oxford University in England set out to identify the defective gene by analyzing the genome of KE family members. Their method was to look for segments of DNA that the affected members shared and the unaffected lacked. The Oxford team soon narrowed the cause of the problem to a region on chromosome 7, the seventh of the 23 pairs of chromosomes in which the human genome is packaged. Within this region lay more than 70 genes, and it seemed that it would take several years to study each gene and see which one was responsible. But Hurst then turned up a new patient with the same rare set of symptoms. The patient, a boy, had a break in his chromosome 7 that disrupted one of the genes in the section the Oxford team was studying. It was an easy task to identify which of the new patient’s genes had been broken. It was a gene known as forkhead box P2, or FOXP2 for short.⁵⁸

The Oxford geneticists, Cecilia Lai, Simon Fisher and Anthony Monaco, then analyzed all 267,000 DNA units in the FOXP2 genes of the KE family members. In all the affected members, and in none of the normal members, just one of these letters was changed fromaGto an A (the four different kinds of chemical units in DNA are known for short as A, T, G and C). The switch to an A at this site in the gene meant that in the protein molecule specified by the gene, a unit that should have been an arginine was changed to a histidine (proteins are made up of 20 different kinds of units, known as amino acids, of which arginine, and histidine are two).⁵⁹

How could a single mutation in a gene cause such a wide range of effects? The FOX family of genes makes agents known as transcription factors, which operate at a high level of the cell’s control system. The agents bind to DNA and in doing so control the activity, or transcription, of many other genes. FOXP2 is active during fetal development in specific parts of the brain, and the protein transcription factor it makes probably helps wire up these brain regions correctly for language. Brain scans of affected KE family members seemed normal at first glance but a more sophisticated type of scan has shown they have considerably fewer neurons than usual in Broca’s area, one of the two brain regions known to be involved in language, and more neurons than usual in the other region, known as Wernicke’s area.⁶⁰

FOXP2 is an ancient gene, and even mice possess a version of it. If the human version of the gene is intimately involved in the language faculty, then the gene would be expected to have changed in some significant way in the human lineage. Svante Pääbo and colleagues at the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany, analyzed the sequence of the FOXP2 gene in mice, in the great apes, and in people from the major continents. Some genes change quite rapidly over evolutionary time but FOXP2, they found, is highly conserved. Chimpanzees and gorillas carry the identical version of the gene, which must be the same as that possessed by the joint ancestor of chimps and humans who lived 5 million years ago. That version makes a protein that differs in only one of its 715 units from the version carried by mice, which shared a common ancestor with humans 70 million years ago. This means that from 70 million to 5 million years ago, a span of 65 million years, the FOXP2 protein underwent only a single change.

But its evolution suddenly accelerated in the human lineage after the human and chimp lineages diverged. The human version of the FOXP2 protein differs in two units from that of chimps, suggesting it was subject to some strong selection pressure such as must have accompanied the evolution of language.

All humans have essentially the same version of FOXP2, the sign of a gene so important that it has swept through the population and become universal. By analyzing the variations in the FOXP2 genes possessed by people around the world, Pääbo was able to fix a date, though rather roughly, for the time that all humans acquired the latest upgrade of the FOXP2 gene. It was fairly recently in human evolution, and certainly sometime within the last 200,000 years, he concluded.⁶¹

Language is such a complex faculty that it must be mediated by a large number of genes and have developed in several stages. Given the observation that the KE family’s deficit seems to be in the power of fluent, articulate speech, Pääbo thinks FOXP2 may have been one of the last genes recruited to the language function, perhaps the final step in the development of modern human speech. “Maybe it made the last perfection of language, made it totally modern,” he says.⁶²

Pääbo regards the dating of the gene as compatible with Klein’s argument that modern language evolved very recently and was probably the spur to the human behavioral changes seen in the archaeological record 50,000 years ago.

Societies with two kinds of people, of greatly differing language abilities, may have existed during the evolution of language. As each new variant gene arose, conferring some improvement in language ability, the carriers of the gene would leave more descendants. When the last of these genes—perhaps FOXP2—swept through the ancestral human population, the modern faculty of language was attained.