The First Word: The Search for the Origins of Language

by Christine Kenneally

  Like Lieberman and others, Arbib disputes the idea that language is one big package, a kind of all-or-nothing proposition. When it is conceived of in this way, he said, “I think you make some very foolish claims.” The alternative is to take the historical point of view. “You can imagine the first protolanguage as ten words or a hundred words. Then a lot of things can occur over the generations and crystallize out. Language becomes very mysterious if you have to make it a single biological evolutionary leap.”

  One of Arbib’s most important points is that language is not inevitable. He encourages thinking about possible stages by stepping away for a while from the end state—the current form of language. We have it today not because we took one crucial turn at some point in the past but because we took hundreds of crucial turns. And for each of these turns, you can’t know that you are going to get language at the end of it. Each step is critical for the value it adds at that point in time. Linguistic evolution was a tumultuous natural experiment that started with a particular brain structure and hundreds of variables—a couple of ice ages, constantly evolving predators and prey, a changing social structure. The process lasted many millennia, there was no control group, there may have been false starts along the way, and the completely unpredictable result of this random experiment was modern language.

  The mirror neurons discovery set Arbib on a course that has most recently ended with his fully articulating an idea that many researchers assume but few have examined in detail; that is, language evolution had to occur in a layering of stages. Arbib calls it an ascending spiral. So far he has proposed about ten different stages, though he warns that even a ten-stage theory is still a long way from accounting for all the steps along the journey to language.

  Initially, he says, our ancestors must have developed a capacity for complex imitation that went beyond that of even modern apes, and greatly increased the possibility of social transmission of novel skills. Beyond that there must have been some kind of gestural protosign that broke through the fixed set of primate vocalizations and was supported by the mirror system. Gesture, in his view, was an ancient scaffolding on which language started to build. You had to use protosign to build the scaffolding, and then sounds became parasitic. Speech did not arrive directly, and the first gestural steps of language would have been quite simple. “It doesn’t make sense to have a full sign language and then go to vocalization,” said Arbib. “It’s hard to build up a rich tradition just through gesture. You need sound to flesh out that scaffolding.” So there were oral and facial gestures as well, maybe some association between lip movements and what lips are often used for, like eating.

  Pantomime probably provided the crucial bridge from imitation of practical skills to imitation of the skills required for proto-sign (and much later for language). “The claim is something like this,” he explained:

  You’ve got a system for primate calls, but it’s closed. You can’t add a new call to it. So you use a different system, you go through a different route, to be able to create new patterns of sound that can be paired with new meanings. And then we eventually get to the stage where you can get those sounds and meanings together to create new meanings on the fly. But there must have been an intermediate time when you didn’t create new meanings like that. My argument is that if you look at the ability of the hands to move skillfully, then you can imagine that there was an evolutionary advantage in being able to imitate patterns of hand movement, and having imitated patterns of hand movement (once you had a brain in place that could do that), it’s a plausible step to begin to use patterns of hand movement for communication—pantomime. And the beauty of pantomime is that if you pantomime carefully, and maybe do it three times when the person doesn’t get it, you can convey novel meanings.

  Why we didn’t ultimately become a species that is constantly engaged in pantomime with no speech, said Arbib, is because “people aren’t very good at recognizing someone else’s pantomime.” As he explained, “It doesn’t have to be that you suddenly have a society in which everybody was doing pantomime and conveyed thousands of meanings, but maybe in a particular year, two or three pantomimes were added to the tribe’s vocabulary by becoming somewhat ritualized to make them easier to perform and understand.”

  After this, Arbib suspects that humans developed protospeech:

  The story goes (if it went anything like that) that in the end you can’t disambiguate pantomime by just doing better pantomime. If I flap my hands to imitate the flapping wings of a bird, do I mean “fly”? Do I mean “bird”? Do I mean “bird flying”? So maybe some genius comes along and invents some way of saying, “Well, if I do this sound and I’m flapping my hands, I mean the bird. If I do another sound while I’m flapping my hands, I mean the flying. You need to make distinctions. So the notion is you got to the stage where a sequence of gestures can convey meaning, and you got across the idea that meaningless gestures are part of conveying meaning. It’s no longer pantomime.

  Vocalization was involved all along, said Arbib. There may have been stages where the pantomime was entirely vocal. “My purely fictional example is that you bite the piece of fruit, it’s sour, you go—” He puckered and made a sucking sound, before continuing:

  The act of genius there is to go from having that as a reaction when it’s too late and you’ve already bitten the fruit, to making that noise, before somebody bites the fruit, to warn them, “Don’t waste that fruit; it’s too sour to eat.” So the notion is that the pantomime would give you the possibility of conveying a rich sense of meanings. The arbitrary gestures would come in to begin to allow you to save effort and avoid ambiguity. The gesture in the end is conventionalized. It doesn’t have to be a fresh pantomime all the time. Then the sound can come into play, and it can begin to become part of an integrated performance. Beyond this, you begin to find certain conventional distinctions that are easier to convey, and then you begin to build a phonology, and then, as you begin to build a phonology, you begin to put those meaningless phonological gestures together to take over more of the conventionalized meaning. I want to claim that this skill was parasitic on increasing manual dexterity and the mirror system that supported it, which increases the cortical motor representation, and then we can expand that to the new use in the vocal system. So I prefer that story at the moment.

  Arbib’s account could be further elaborated by explaining why the pantomimes were taken up and spread throughout the group. Perhaps it was a case of sexual selection, as Pinker and Bloom suggested in 1990. In this scenario, the mime is a male who impresses females with his linguistic skills, thus creating more opportunity to procreate, having more children, and spreading the predisposition for expression. The same principle explains why male peacocks develop such spectacular tails.

  Tecumseh Fitch, on the other hand, argues that sexual selection is particularly unlikely as an explanation for linguistic evolution. As with the peacock, this kind of selection generally results in a marked difference between the sexes with regard to a particular trait. However, not only do men and women both use language, but young females are more adept with language than are young males. Other pressures may have come into play. Perhaps a change in the available game required better hunting techniques, which in turn required more precise language. Maybe the step from one form of protolanguage to another occurred when hominids reached a critical level of population density—just like the orangutans with their Neesia-splitting techniques.

  In an interview Chomsky suggested there had to be a point in time when a rewiring of the human brain that allowed people to use recursion took place. Perhaps sixty to seventy thousand years ago in a small hominid group in East Africa, a single individual was born with a genetic mutation. This mutation would have caused a restructuring of the brain and instantly bequeathed the affected person with the capacity for unbounded thought. Linguistic communication would not have begun at this moment, because the individual with the mutation was the only one with the capacity for it. But even a sligh
t advantage spreads quickly throughout a population, and after this new rewiring was passed on to his or her offspring, the entire group would eventually become language-ready.

  Is it possible that even though some of these accounts appear mutually exclusive, the researchers are actually describing different stages of a cohesive evolutionary narrative? Yes. It’s likely that different parts from many theories will survive in a grand synthesis because within this vast time frame, numerous evolutionary pressures had some effect. Given the way the recent accumulation of data about how the brain works and genetic influences on language have forced researchers to constrain their theories accordingly, a more widespread agreement isn’t out of the question in the near future.

  At any rate, the question of which specific evolutionary pressures were in play at which moment in time is a less prominent consideration in the field. There is less concern about why language came to be because there is so much to say about what came to be and how it came to be—which gene changed, which behavior is ancient, and which ability is new? At this point, we must be content to survey all possible answers and acknowledge that in the last six million years many of them probably played a role. The stories can be especially helpful as spurs to testable hypotheses.

  In addition to examining the specific pressures, incidents, and abilities that have contributed to the story of language evolution, it is also important to look at co-evolution: the way that human language and the human genome have shaped each other. Co-evolution is the least explored aspect of the mystery. For all the difficulty and challenge of tracing language evolution, working out how species and language arise over time and then provide feedback to each other is probably the hardest part.

  Terrence Deacon has grappled with the issue of co-evolution, focusing on the back-and-forth between language and the brain. Recall that he proposed that the beginnings of language and symbol use can be found in the shift from australopithecines to hominids some two million years ago. What preceded this evolutionary shift was the use of flaked stone tools. Deacon argues that it was this tool use that spurred the evolution from one kind of primate to another and, in doing so, created an animal with a predisposition for even more symbol use.

  This kind of change, which is called Baldwinian evolution, occurs when the behavior of an animal actually contributes to the environment in which its genetic evolution is shaped. Lactose tolerance is an example of Baldwinian evolution in humans. The ability to digest dairy products in adulthood is most common in groups of people who have been herding animals the longest. In this case, it’s a behavior—herding—not a climatic change or some other kind of environmental shift, that contributed to the selection pressures in which a predisposition for lactose tolerance improved reproductive success.

  The australopithecine tool use helped to create a world where it was more and more useful to have the genetic predisposition underlying that behavior. The better an individual was at it naturally, the more likely he or she was to survive and have offspring, probably passing this trait on to them, and the more significant that behavior became in the world of the species. It wasn’t that our brains got bigger as a result of bipedalism or dietary changes or any other reason, thereby making us clever enough to invent stone tools; rather, we started to use stone tools that are slightly more complicated than the tools chimpanzees use even today, and as a result our brains got bigger.

  The co-evolutionary story that began at this time and that continues to this day is one in which the Baldwinian interaction between culture and biology played a particularly significant role. Deacon points out that our brains did not get bigger in the australopithecine-hominid transition in the same way that the surface of an inflating balloon gets bigger all over. It was the forebrain, particularly the cerebellum and the cerebral cortex, that ballooned, while the rest of the brain followed the growth rate seen in other primate brains.2 In order to unwind the ways that language and the brain have co-evolved, you have to look at the parts of the brain that got bigger, says Deacon, and you have to look at how they got bigger.

  The prefrontal cortex corresponds to a small section of the developing brain in the human fetus. When the brain is nothing more than a neural tube, the part of the tube that later turns into the forebrain breaks out of the growth patterns that constrain the rest of the brain. This stretch of tube is controlled by the Otx and Emx genes. The developmental clock that signals to every part of the brain and body when to stop growing has been extended for the regions controlled by these genes. The significance of this altered growth pattern, according to Deacon, is not that human brains are faster and better computers; it means that the balance has been shifted in terms of what kind of thinking goes on in the brain. As a result, our learning skills are biased toward certain types of processing and not others.

  Acquiring and deploying the particular kinds of connections and structural patterns that characterize language, says Deacon, pose some very unusual learning problems, and the kinds of learning processes that most mammal brains are specialized for are not well equipped to deal with these problems. However, the neural machine that results from the human combination of body and brain growth patterns is one that rather brilliantly performs the computations that underlie language learning.

  The fact that language arises from dynamically interacting brain regions with their vastly different evolutionary histories (the more primitive and unchanged along with the parts that have shifted more recently) is another reason why we should not think of language, or even other mental abilities, such as mathematics, as monolithic things. Instead, argues Deacon, they arise out of a “delicate balance of many complementary and competing learning, perceiving and behavioral biases.”3

  Upending these assumptions about brain evolution leads us to a startling conclusion, says Deacon. One of the reasons we haven’t been able to work out how language and the brain co-evolved is because we have been asking the wrong question all along. From the beginning, researchers investigating the brain and language have assumed that the brain came first. The usual line of reasoning holds that the brain was selected for increased general intelligence and then it evolved language, which relies on that optimized intelligence. Actually, says Deacon, we should be looking at the effect of language on the brain, as well as the effect of the brain on language.

  Generally, the amount of brain tissue devoted to particular types of processing is proportional to the amount of information being processed. Brain regions that serve seeing, smelling, and touching, for example, are matched sizewise to the amount of information that is filtered through our bodies in these senses. One of the crucial differences between the human brain and other mammal brains is that ours is larger overall relative to the body. This leaves a considerable proportion of the human brain, says Deacon, that is not processing information from the outside world in the way that the visual and auditory cortices are.

  Even though they are not directly processing sights, sounds, and other senses, the unusually expanded prefrontal brain regions look as if they have been “deluged with some massive new set of…inputs.”4 The larger brain region, says Deacon, is “an evolutionary response to a sort of virtual input with increased processing demands.”5 That input, of course, is language.

  In this view, language cannot ultimately be treated as a straightforward example of the capabilities of the brain, and we should not be asking “How did the brain evolve language?” Rather, we should ask, “How did language evolve the brain?” Language is the author of itself, says Deacon, and the brain is the smoking gun for language.

  The result of the co-evolution of the human brain and language is that we now have an overall cognitive bias toward the “strange associative relationships of language.” In this sense our whole brain is shaped by language, and many of our cognitive processes are linguistic. What this means, according to Deacon, is that once we have adapted to language, we can’t not be language-creatures. For us, everything is symbolic.

  Indeed, Deacon explains, t
he virtual world that we inhabit is as real, sometimes more real, than the physical world. Even the tendency to infer the hand of a designer when faced with complex design (whether it is a deity that has designed all of creation or a special language organ that generates human languages) arises from the fact that we are a symbolic species. Ironically, what makes it hard to discern how language evolved is a result of language having evolved. The worldwide web of words and rules that we inhabit is so vast, contracted, and dense, it’s hard to look in from the outside.

  Arbib and Deacon seek to illuminate moments in the last six million (and more) years of human evolution. By comparison, the last ten thousand years is a blip. Nevertheless, it is an interesting period in the history of the human brain and also of language. There is some suggestion that our brains may have changed as little as ten thousand years ago, and in fact, become smaller. It will be some time before data on the trajectory of brain growth at this time are solid enough for us to be confident of this change. Generally, it is thought that within the last ten thousand years there has been no obvious anatomical change arising from the drift and selection of genes in our species. The same goes for language. Most language change in this time frame is associated not with obvious biological change in humans but with the movement of human populations and transformation of their lifestyles.