The First Word: The Search for the Origins of Language (25 page)

It’s common knowledge that the brain is divided into two hemispheres, each of which normally controls the opposite side of the body. Roughly speaking, the right side of the brain controls the left hand and leg, and vice versa. It’s also the case that particular functions and behaviors can be dominant in one hemisphere, and a significant amount of language function seems to be represented on the left side. Brodmann’s area 44 is larger on the left side than on the right in humans. So far so good—we’ve known this for a long time. But Cantalupo and Hopkins showed that the area corresponding to Brodmann’s area in ape brains is much larger on the left side as well.

Why would this be the case? Apes don’t speak. And if spoken language is a purely human phenomenon, this finding makes no sense. It does make sense, however, if we think of linguistic ability as having a heterogeneous structure. If this ability has developed piecemeal over time, then ape brains should share some of the same structures we use for language. The ape asymmetry also means, wrote Cantalupo and Hopkins, “that the neuroanatomical substrates for left-hemisphere dominance for language were evident at least five million years ago and are not unique to human evolution.”

But still, apes don’t speak. What purpose does a larger left area 44 serve for them? Cantalupo and Hopkins suggest that apes are controlling gestures with this part of the brain in a languagelike way. Humans evolved the ability to point intentionally with their body parts and then with words. Captive apes are known to point at objects with intention, and in the apes observed by Cantalupo and Hopkins, a preference was exhibited for doing so with the right hand. Since the right hand is controlled by the left hemisphere, Brodmann’s area 44 may be controlling the ability to flexibly refer to objects in the world, an ability that underpins verbal and gestural communication.

It is also the case that the apes’ bias for using the right hand was consistently greater when they were vocalizing at the same time as they were pointing. In evolutionary terms, say the researchers, this means the “brain area may be associated with the production of gestures accompanied by vocalizations.” So what started out as a meaningful gesture plus screech in apes, according to Hopkins and Cantalupo, likely became selected over time for speech and modern language in the human species. In 2002 Elizabeth Bates and Fred Dick reviewed the work done on gesture and language and found that as a child grows, these components develop at the same time in the same places in the brain.
²⁶

Another extremely striking finding about these shared brain bases came from Michael Arbib and Giacomo Rizzolatti, who discussed mirror neurons as the first real evidence of the neurological underpinnings of imitation in 1997. Mirror neurons are specialized brain cells that fire if you, say, grasp a pen; they also fire when you see someone else grasp a pen. In some sense, the brain interprets these actions as the same thing by mapping them in the same way, meaning that what the monkey can do, the monkey can see. Arbib and Rizzolatti argued that the evolution of minor neurons allowed humans to be skilled imitators: what the human can see, the human can, within reason, do. They help explain why speech is rooted in gestural communication.

Over the course of his career, Arbib’s research has involved developing computational models of the brain mechanisms that underlie language and getting them to sync with findings in psychology, philosophy, and linguistics. In the 1980s he began a research program at the University of Southern California for computational modeling of mechanisms in the monkey brain and started collaborating with Giacomo Rizzolatti on how the brain used vision to control hand movements. He was thus on the spot when Rizzolatti’s research group in Parma, Italy, discovered mirror neurons. It was this work that led him to mirror neurons in monkeys. Arbib began a collaboration with Scott Grafton, a colleague who was an expert in PET imaging, and together they ran some PET experiments to look for mirror neurons in humans, which they eventually found in many regions of the brain.

At first, mirror neurons were thought to underlie only visual recognition of hand actions. But then Evelyn Kohler and others in Parma began to look at their use in the auditory domain, finding that the monkey mirror system was much more sophisticated than originally thought. Mirror neurons fire when stimulated by distinctive sounds as well. For example, if a monkey sees another cracking a nut, certain neurons will fire. If the monkey only hears the breaking shells, some of the same neurons—the audiovisual mirror neurons—will fire. This is a long way from speech, but it does show that mirror neurons can link to auditory input, so some basic mechanisms for grounding the evolution of speech analysis were, presumably, already in place in the brains of our common ancestor with monkeys, who lived twenty million years ago. One aspect of language for which the mirror system may be responsible is the repetition of pronunciation and words. It may also be a foundation for word acquisition, in which repetition is a relatively stereotyped performance.

In his comparative work on mirror neurons, Arbib said his challenge is to ask, “What is the minimal set of requirements for our brain which would make it possible for us to acquire language?” He uses the slogan “the language-ready brain” to suggest that a brain “might not have language, but might be ready to learn it—just as we have computer-ready brains and today we can use computers.” He added, “Nobody would claim that our biology was in any way influenced by the use of computers.”

So far most researchers have studied one relatively small local area of the brain. Arbib has examined the interaction of mirror neurons in the neocortex, and he’s done a fair bit of work on the basal ganglia, the same area of the brain that fascinates Philip Lieberman. Lieberman argued that the kind of sequencing that the basal ganglia controls is as fundamental to language as it is to dancing. And Arbib is inclined to agree. “The mirror system won’t explain all of language,” he said. “The next big step is to pull together all these brain areas that are very important for language, and in particular for understanding how language is created and understood on the fly. The brain is a big place.”

Currently, Arbib is working on a scene description study. “I’m asking, ‘How do you look at a scene, where you do start?’ If I give you a video clip, you’re going to pay visual attention to it, and you’re going to create a visual representation. Then you’re reading part of it out as a sentence as you describe it to me.” When people do this, there’s no sense that they are developing a syntactic structure first and then popping words into it as they describe it, but are literally making it up as they go along. The subjects have a very complex mental picture, and they have to translate from the mental picture to the meaning and then to the words of language.

“It’s not just the sequence but the skill,” said Arbib. As he reached for a cup of coffee, he said: “Take an example from manual skill. We’ve actually done models of the cerebellum where we reach for a cup. What you’ll see is just one smooth movement where my opposing fingers reach the cup at the same time.” But he explained that if the cerebellum was damaged, it would not be so easy:

 

You’d have to decompose it in two movements, because you can’t coordinate the timing. So if you tried to do it, you might end up having the fingers too close when you hit the cup, or too far apart when you reach the cup. In other words, you run the risk of knocking the cup over. So instead, what do you do? You very quickly compensate for your understanding of your deficit, and you reach out and you get, let’s say, thumb contact, and then you will close the hand. In other words, you break the thing down into pieces that you know you can succeed with, and then you resynthesize the sequence that will get you to your goal. But each gesture is itself less skillful than it would be if you executed it with [an undamaged] cerebellum.

 

This implicates yet another part of the brain. “You can get the sequence right without the cerebellum, but if you want a smooth performance, you need a cerebellum. It cues each movement, and it coordinates the movements,” he said. “You can’t do language without a cerebellum.”

T
here is a family in England known in the medical literature as the KE family. Its twenty-nine members are spread over three generations, and fourteen of them have severe difficulties with speech and language, as well as some general cognitive problems that are less severe. Faraneh Vargha-Khadem, the cognitive neuroscientist at the Institute of Child Health in London who has studied the family for over two decades, explains that their disorder causes immobility in the lower portion of the face, including the lips, tongue, and mouth.

As a result their articulation is greatly impaired, but the problem is more than one of just motor control. In simple repetition tests, affected individuals have trouble reproducing sounds and words in the correct sequence, selecting the right sounds for words, and maintaining an appropriate rhythm. Multisyllabic words like “hippopotamus” can be particularly difficult, and in general, the more unfamiliar a word, the more trouble they will have saying it. Their speech is sometimes unintelligible.

As babies, the affected family members behaved somewhat like deaf children—they were quieter than the average infant. Because the lower part of their face was relatively immobile, they had a limited array of facial expressions, which in general were not as spontaneous as those of the unaffected members of the family.

In the affected family members, structural and functional brain scanning shows changes in the speech and language areas. For example, when you’d expect Broca’s area to be active, the affected KE members show a scattered pattern of activation in regions of the brain that wouldn’t normally be active during language processing.

Vargha-Khadem discovered the family when one of the affected children was seen because of speech and language-related problems. Consequent to this meeting, other members of the family were also assessed, and the profile characteristic of the affected individuals was identified. The disorder, she said, involves a complicated circuit that regulates the movement of the muscles of the lips, tongue, and lower face used in speaking and the hardwiring of the brain structures that are typically used for language. It’s unknown whether the problems begin with the physical challenges that the affected family members have in producing, and to a lesser extent comprehending, speech, or whether the fundamental obstacle lies in the creation and understanding of language in the brain.

Vargha-Khadem asked a group of geneticists at the University of Oxford to see if they could identify the defective gene causing the disorder. The team spent several years closing in on the gene responsible when their search was given a boost by a similar speech and language problem in an unrelated child. That child had problems very much like those of the affected KE family members, and between the two different sets of data the geneticists were able to narrow their focus and find the problem gene, dubbed FOXP2. It was the first, and so far only, time that a single gene was linked to an inherited speech disorder.
¹

The FOXP2 gene is located on chromosome 7. Because all normal people have two copies of a chromosome, every individual should have two copies of chromosome 7 and two copies of FOXP2. In the KE family affected members have a mutation that leaves them with only one working copy of FOXP2. The result is what geneticists call a dosage effect: If you have two normally functioning copies of FOXP2, brain and language develop normally, as is the case for the unaffected members of the KE family. If you have only one working copy of FOXP2, you are going to have an array of difficulties with language and speech. No living individual with two malfunctioning copies of FOXP2 has ever been found.

FOXP2 is expressed in several organs of the body, including the brain, where its pattern of expression appears to be specific to regions involved with the development of motor control.

Twin and other developmental studies have demonstrated a strong link between genes and disorders of speech and language, but most of these findings have presented a very complicated picture and it is suspected that language impairment is related to many genes. The KE family is the first clear demonstration of a single gene affecting language ability and speech articulation. It is a landmark case and may yet prove to be the twenty-first century’s Phineas Gage in its being a foundational case study for more than a century’s worth of neuroscience.

The announcement of the discovery of FOXP2 inspired a debate about what role the gene would normally play in language function—whether its main function is to process and produce the sounds of language or specific parts of language, like grammar. Initially, it was hailed in the popular media as proof of the existence of a language gene, or even a grammar gene.

Even before the discovery of FOXP2 some researchers argued that the KE family proved the existence of a grammar gene. Why is the idea of a specific language or grammar gene appealing? Why would isolating a gene that controls language and that controls only language be such a coup? First, if such a language gene existed, you could track the development of language very finely over time. The beginning of language in the human race could, theoretically, be exactly pinpointed. Second, if a language gene like this was discovered, it would give great weight to the theory that language appeared with a big bang. Finally, a language gene that was possessed by humans and no other animal would provide compelling evidence for the traditional claim that language is a discrete mental trait unique to our species. Indeed, in 1990 the linguist Derek Bickerton proposed that language evolved because of a single genetic mutation.

One criticism of Chomsky’s view of evolution was that it was almost creationist and that it required some kind of miraculous genetic big bang. The defense to this criticism had always been that Chomsky’s ideas about language didn’t implicate evolution one way or the other, and yet Bickerton made explicit what critics said was implicit all along. Bickerton proposed that in a single female who lived approximately 220,000 years ago, a genetic mutation resulted in changes to the vocal tract and skull, as well as a rewiring of the brain for syntax, thus giving rise to language.

Bickerton’s proposal was vociferously criticized by evolutionary biologists, and he has since modified his position.
²
FOXP2-based claims for a grammar gene have likewise petered out. Why? They depend on a view of genes only as atomistic building blocks and the genome as a blueprint for the organism, and neither of these ideas has held up. While few researchers would claim that language and genes are not related, there has been little evidence that language is genetically encoded. Certainly, there is no direct relationship between possession of the FOXP2 gene and fully having language.

This thread in the development of the language gene story runs strongly parallel to all the ideas regarding comparative animal work on gesture and cognition that have so far been discussed.
³
The genetic mutation idea echoes all the other suggestions that the extremely complex apparatus that allows you to learn language evolved as a discrete and singular entity—a language organ—that arose without any antecedent.

When Darwin described natural selection a century and a half ago, he was essentially describing a genetic process (the way that genes throw up random mutations and then propagate over time). Today we know that all normal humans have twenty-three pairs of chromosomes, which reside in the nucleus of cells. Chromosomes are made up of DNA, which in turn is made up of four nucleic acid bases, adenine, thymine, guanine, and cytosine. The bases are most commonly designated by their first letters, A, T, G, and C. Stretches of DNA along the chromosome constitute a code for specific proteins, so when molecular machinery reads these segments of DNA, proteins are made in the cell. These segments—these units of code—are called genes. A gene is expressed when the protein it codes for has been produced.

In between the genes, there are stretches of nucleic acid bases that do not code for proteins. These strings of A, T, G, and C, junk DNA, can randomly vary without affecting the organism. The genome of an organism, then, is the entirety of its DNA, junk and genes.

During reproduction, genes are duplicated and carried forward, sometimes having no effect whatsoever. Other times, genes or larger groups of genes get flipped and reinserted in the process of duplication, possibly into the same spot, or they might get moved. This rearrangement occurs at different rates in different species (it is a process we don’t fully understand).

In the last few years our ability to describe what the units of evolution look like and do has culminated in the sequencing of the human, mouse, rat, fruit fly, and chimpanzee genome, among others. We have discovered that our genome is not nearly as large as we thought, and once we got over the shock of this, our understanding of how genes actually work has grown immeasurably more sophisticated. The sense that a huge gap existed between animals that produced language and animals that did not arose in large part from our narrow view of the abilities of nonlinguistic animals. Now that we are crediting them with greater mental skills, we can see more clearly how the language we have arises from the platform of thinking and communication that we share with them (or, if you want to cut it more finely, from the many platforms we share, each resting on the other, mammalian arising from reptilian, and so on). The common platform arises from common genes.

Since Darwin’s time we have come to understand that not only has all life descended from the same ancestor but many features that arose in more recent ancestors are still shared between us, being built by the same genes. We can see that biologically we are basically African apes who only recently left the motherland. And our most distant human ancestors have been located in time and space. All of us alive today share at least one grandmother who lived 150,000 years ago in East Africa.
⁴
We also share at least one grandfather, an African man who lived 60,000 years ago.
⁵

We see today that differences in complexity between life-forms arise more from the way that genes interact with one another than from their raw number. It’s clear that the notion of a genome as a blueprint—so popular only five years ago—is at best inadequate and at worst completely misleading. Instead of following straightforward predetermined plans, genes operate in a dynamic fashion. Many genes respond to the experience of the organism they are building, and they can be switched off or on by other genes or by the effects of the environment.

If a gene comes on in the right cell at the right stage of development, it has a beneficial effect. The same gene acting at the wrong time or in the wrong place can be devastating. Vision, for example, doesn’t just unfold automatically in certain animals. The animals need to be exposed to light for the right gene to start building the ability to see. Moreover, different genes have dominion over different body parts. HOX genes divide up the body plan of organisms, with each affecting a certain segment. Some genes are noted for their effect on other genes. These manager genes turn numbers of other genes on and off, and in this way changes in a single gene can cause chain reactions of gene expression.

What we have learned about genes has allowed us to understand that we are not so much things merely existing in the world as beings in constant interaction with the world. If you took this idea to an extreme and imagined that you grew up on another planet, the essentially dynamic nature of animal building by genes
and
environment might mean you’d look very different. Cloned plants that have exactly the same genome can look like very different specimens if planted at different altitudes. In the same way, if you had grown up on a planet with lower gravity or one that was more distant from the sun and had a lower oxygen concentration, you might be incredibly tall, or short, or weedy, or blind…or maybe you’d have a supersized brain. If you took your African ape genome and cultured it on yet another planet, maybe the resulting
you
would have translucent skin. The point is that although we experience ourselves in some sense as finished or perfected, we are not in any way
intended.
There is no blueprint for what humans are meant to be. And as this moment is merely one moment in the past and future history of our evolutionary lineage, your life right now is merely an instant in the past and future history of the interaction between your genome and your environment.

At the time the discovery of FOXP2 was announced, Faraneh Vargha-Khadem said that she didn’t believe it was accurate to call it a language or grammar gene. As she explained, “The core deficits of the FOXP2 gene have much more to do with speech and articulation than with the more complex aspects of language.” Certainly it has turned out to be much more complicated than a single-function grammar gene.

FOXP2 is the kind of gene that turns a tree of other genes on and off, so there is no one-to-one correspondence between it and a single trait. As mentioned earlier, it is also expressed in the heart, lungs, and other tissues.
⁶
For that reason calling FOXP2 a language gene is a little like calling gravity a force that makes apples fall from trees. It’s true enough, but it’s hardly the whole story. This fundamental truth about genes, in addition to the way that some genes produce cascading changes in others (as opposed to the purely atomistic “gene + gene + gene = discrete trait” idea), has made it increasingly difficult for skeptics to resist new ideas about language evolution.

One of the most exciting things about the FOXP2 discovery was that it seemed to be more than just a gene that could block normal language development (in the same way that, hypothetically, if your mouth didn’t form, you wouldn’t be able to speak). It looked, rather, as if it had some role in actually building language. In the ensuing years evidence for this has accumulated as other groups have begun to study the effects of the gene in different animals. Although our version of FOXP2 is unique to us, it is highly conserved between species, and in fact predates the dinosaurs. Though there is no direct relationship between possession of the gene and fully having language, FOXP2 does play a role in the communication of a number of different animals.