The First Word: The Search for the Origins of Language

by Christine Kenneally

  Because the revolution in language evolution is so recent, one of the most important messages of this book is the very basic idea that investigating the evolution of language is a good and worthwhile pursuit. Indeed, it’s not possible to fully understand language if you don’t take evolution into account—either you must begin with evolution or you must make room for it.

  I have tried to draw attention to the commonalities among language evolution researchers, and it may turn out that many of the researchers who disagree have more in common than currently seems to be the case. At the moment, most scientists are not particularly concerned with tying the stages they consider important to language evolution to the chronology of our evolutionary history. But as the research becomes more elaborated, what now look like conflicting theories of linguistic expansion may end up as different phases in the same evolutionary account.

  This epilogue is devoted to the differences among researchers. Part of the glory of the language evolution debate, as with all the other big, messy debates, like that about the relationship between mind and brain, is just how many highly trained and really smart people disagree completely with one another.

  It used to be that words like “innateness” and “uniqueness” were sufficient to pinpoint the distinctions between particular scientists or schools of thought. Until recently, a great divide separated those who believe there is some kind of computational mechanism at work in the generation of language and those who think it can be explained only by general principles. But even now, these two positions are becoming more difficult to distinguish. No one serious has taken a stance at an extreme end of the continuum, and each side makes concessions to the other.

  Everyone would agree that our biological endowment and the way that our individual lives unfold cannot be fully disassociated. In fact, we need an easy word to describe what we actually do have: a unity of nature and nurture. Geneticists talk about the phenome, the inextricable mesh of the individual’s genome and the environment that selects and deselects the way the genome gets expressed. Probably the best word for our purposes is just “life.”

  But even without extreme arguments, examining the role of biological endowment and the environment in language learning remains one of the best ways to identify differences between scholars who differ in the relative weight they assign to each. Chomsky once likened the emergence of language to the growth of limbs, implying that language is something that inexorably projects out of the individual without effort or conscious intervention. Other researchers like Philip Lieberman cite cases like Genie, the little girl who was not spoken to as a child and never developed language normally. He maintains that children must be exposed to language in order to acquire it fully.

  I asked the key researchers interviewed for this book to answer the following question (some declined to participate): If we shipwrecked a boatload of babies on the Galápagos Islands—assuming they had all the food, water, and shelter they needed to thrive—would they produce language in any form when they grew up? And if they did, how many individuals would you need for it to take off, what form might it take, and how would it change over the generations?

  Michael Arbib: The closest data that we have on this topic is that of Nicaraguan Sign Language. Here, a group of deaf children, brought together in a school for the deaf in Managua, Nicaragua, spontaneously developed a full human signed language over three “generations” (where a generation was not a biological generation but rather a cohort of children admitted over a ten-year period). Each cohort seemed to plateau in its capability, so that the signing of the first generation was more like pantomime and less like conventionalized sign language. But with each generation, the repertoire of conventionalized signs and the expressivity with which they combined increased greatly.

  Let’s leave aside the fact that babies given food, water, and shelter but without caregivers are unlikely to survive. Assuming they did survive as an interacting group, the data on Nicaraguan Sign Language might suggest that if a group of babies were raised in isolation from humans with language (the Galápagos Islands really don’t qualify), then in three generations (this time, biological generations) some critical mass of children—let’s say thirty or so—would develop language. And presumably, since these children are not deaf, one might well expect the resulting language to combine vocal and manual gestures, as does normal human discourse.

  However, I doubt very much that this would happen. I believe that the brain of Homo sapiens was biologically ready for language perhaps 200,000 years ago, but if increased complexity of artifacts like art and burial customs correlate with language of some subtlety, then human languages as we know them arose at most 50,000 to 90,000 years ago.

  One may either respond by rejecting this idea that it took human brains 100,000 years or more to invent language as we know it or suggest that the Nicaraguan deaf children had an advantage that early humans lacked. I adopt the latter view. And what is that advantage? I claim that it is the knowledge that things can be freely named, and the knowledge that languages do exist. Certainly, the Nicaraguan children could not hear, but they could see the lip movements that indicated that their families could communicate their needs and requests. In addition, they lived in a world of many distinctive objects, both natural and artificial, and could see that something more subtle than pointing could be used to show which object was required. Moreover, some had at least basic knowledge of Spanish and had both seen and performed a variety of co-speech gestures. They thus would be motivated to try to convey something of their needs, or share their interest by pantomime and the development of increasingly conventionalized gestures. Intriguingly, Ann Senghas (an expert on Nicaraguan Sign Language) has told me that the second generation even went to Spanish dictionaries in search of words for which they needed to develop hand signs.

  For us, as modern humans, it seems inconceivable that the very idea of language is something that has to be invented. Yet, to take a related example, we know that writing was invented only some five thousand years ago. Yet once one has the idea of phonetic writing, it is a straightforward exercise to invent a writing system—as has been demonstrated by many Christian missionaries who wanted to bring literacy and the Bible to a people who had language but no writing.

  In view of all this, I doubt very much that a few children on a desert island would develop much beyond a rudimentary communication system of a few vocal and manual gestures and some conventionalized pantomime unless they had hundreds of generations in which to create culture and the means to discuss it. But they would have the brains to support such inventions, whereas other creatures would not.

  Paul Bloom: The answer is: yes, two. This is more than guesswork because a variant of this situation has already occurred, more than once. This is when children grow up without being exposed to a language model, such as deaf children who are raised by adults who don’t use sign language. Such children will sometimes create a rudimentary language, complete with words and some sort of morphosyntax. Over generations, the vocabulary will grow, and the syntax and morphology will become more complex.

  Wolfgang Enard: Yes, if there are more than two.

  Tecumseh Fitch: Yes. You’d need a village worth. They start out with something very basic in the first generation. Then they’d develop a pidgin in the second generation. In the third they’d have a creole, and by the fourth they’d have a fully stable language.

  Marc Hauser: Since the language faculty requires input of some kind in order to be expressed as an externalized or e-language in Chomsky’s sense, there would be highly structured internal thought but there would be no expressed language and no communication, as there would be no one to communicate with.

  Ray Jackendoff: Yes. Only if there were about thirty of them to begin with.

  Simon Kirby: There are two different sources of evidence that we can use to get a handle on these Galápagos children. Firstly, because computational agents aren’t yet complex enough to have acquired any rights as experim
ental subjects, we are free to re-create this Galápagos in computer simulation. This very scenario has been the subject of a good deal of research, particularly over the past decade. One of the problems with this approach is that it can tell us only what might happen, given particular assumptions about how children’s brains work, and that in itself is a difficult research issue. However, we can learn from simulation exactly how little, or how much, of language is required to be pre-wired before a language would emerge in the Galápagos population, and what factors other than individual psychology are important.

  A second source of evidence comes from studies of spontaneous language emergence in real populations. Of particular interest are the indigenous sign languages that have evolved in populations of deaf children who lack a preexisting shared system for communication. The most celebrated case is the language that emerged in Nicaragua around the time of the Sandinista revolution when a large population of deaf children were brought together for the first time. However, a fascinating meta-study conducted by the anthropologist Sonia Ragir has shown that this kind of language emergence is not guaranteed to occur in all populations of deaf children who lack a common language. Just as has been suggested by the computational models, certain features of the population (and its dynamics over time) appear to be critical for a novel language to emerge.

  I would bet that the emergence of what we’d consider a “full” human language in the Galápagos scenario is equally not guaranteed. What would make it more likely is if the population of children was large, if further boats of children arrived at regular intervals (say once or twice a year), and if every member of the community was engaged socially with the group in a way that made linguistic communication relevant. I would expect that language emergence would be gradual, with later arrivals and younger arrivals using the language in increasingly abstract ways and with increasing amounts of what we’d consider linguistic structure. I would expect that for a good number of generations a visiting linguist would be able to tell straightaway that the Galápagos language was unusual. Whilst it would be likely to “obey” some of the key universal principles of language organization—particularly ones that have their explanation in language processing and use—it would lack many of the morphological irregularities and paradigmatic quirks that are common in “normal” languages (i.e., languages with a history). If I were to go out on a limb, I’d say it is at least possible that the language would initially lack some of the features that we take for granted, like potentially unlimited embedding (my brother’s son’s friend’s mother’s fishing net), and perhaps some basic grammatical categories. In addition, we can be almost certain that for a very long time it would lack an extended numeral system or more than a handful of basic color terms.

  I think the Galápagos experiment might also show us that much of the complex structure of human language doesn’t really give us immediate payoffs in terms of any enormously increased chance of survival. I would predict that the vast majority of the survival needs of the nascent community would be served by a much more primitive protolanguage. If a structured, complex language were to emerge, it might not be of any immediate survival benefit (at least as a result of communicative efficacy); rather, structure and regularity would appear purely because of the adaptive dynamics arising from the way the language itself is transmitted from individual to individual. Only much later, as more and more complex cultural artifacts appeared on the Galápagos, would having a complex human language come into its own as a way of transmitting cultural information from individual to individual.

  If this is right, it suggests a problem for explanations of language structure that rely on natural selection pressures arising from communicative needs. Why then would we have the ability to acquire complex language at all? Of course, this is the question we all want to answer. Certainly, we already know that signaling systems with fairly complex structure have evolved a number of times in nature without providing any obvious communicative benefit. Perhaps the visiting linguists would also learn a lot from listening to the songs of the birds on the island, or to any whale song they might hear on their voyage there.

  Chris Knight: The key innate feature of human cognition—the one most relevant to the emergence of language—is in my view the capacity for joint attention and egocentric perspective reversal. We humans possess an inborn capacity to correlate our perspectives on the world, viewing ourselves from one another’s standpoint. As well as being cognitive in the narrow sense, this faculty has, simultaneously, moral relevance. If I choose to have a violent tantrum, I must temporarily shut down my moral self-awareness. I need this kind of awareness only if I am trying to tune my behavior to social requirements.

  For this reason, I imagine the boatload of babies would spontaneously produce some kind of language. But if I think this, it’s because I assume the population would comprise females and males in about equal proportions, and because I assume potentially violent conflicts over sex would be sorted out as these individuals reached puberty. A potentially aggressive, sexually violent male, for example, would soon meet collective opposition. If this was effective, it would force him to view his behavior from the standpoint of others, modifying that behavior accordingly. The idea that language as we know it could emerge wholly autonomously, in isolation from any kind of institutional structure—any kind of self-organized moral regulatory framework—is gravely mistaken.

  Like the use of paper money, linguistic communication depends entirely on trust. If the social and sexual dynamics on the Galápagos Islands obstructed the emergence of sufficient mutual trust, then the emergence of a shared public language in the shipwrecked population would be severely threatened and obstructed. As we all know, chimpanzees, bonobos especially, have considerable innate potential for symbolic communication. Why is such potential not drawn upon in the wild? The reasons are political as much as cognitive. Chimpanzees don’t hold one another to collectively agreed standards of public behavior. Although as individuals they value cooperation and sociability, it soon becomes clear that a kind of internal civil war is the default state in relations between sexually mature adults. Sometimes this civil war is latent; sometimes it explodes into the open. Where public trust is not the default state, individuals have no choice but to fall back on emotionally persuasive, hard-to-fake gesture-calls. This applies even to humans. One theoretically possible outcome in the Galápagos would be something like the scenario depicted by William Golding in his terrifying novel Lord of the Flies. Under violent and inhuman conditions, I imagine that this shipwrecked population would engage in a lot of screaming, crying, and so on, but not a lot of quiet, rational conversation.

  Philip Lieberman: No.

  Gary Lupyan: The emergence of language on the island is by no means certain even with a fairly large number of individuals. I think the emergence of modern language is as much a cultural as a biological phenomenon.

  It’s helpful to think of the emergence of language in terms of other cultural achievements of our species such as the emergence of writing. Biologically modern humans are obviously capable of reading and writing, but if you put a bunch of already-speaking babies on an island, what’s the chance that writing would emerge within their lifetimes? Well, it happened at some point, so it’s not zero. But the chance depends on factors like their motivations, their culture, their technology, and so on. Hunter-gatherers have little need for writing. But given a culture with money, farmers, and landowners, and you can foresee a need to keep permanent records.

  Like writing, the emergence of language depends on motivational and environmental factors. What those are is a bit of a mystery. Group size is an obvious one; the need to organize and cooperate is a likely factor. If food is plentiful on this island and individuals can get by in small groups, the emergence of language is less likely.

  If language emerged, I think it would change radically over the first few generations, demonstrating the kinds of changes languages undergo as they move from pidgin to creole. But language
change is not just something that happens with time—change responds to pressures of the environment and the society. If the people on the island needed ways to talk about time in a precise way, we could expect a complex system of tenses to emerge. If the culture created by the children is very hierarchical, a grammaticized system of honorifics may emerge in the language.

  I believe the idea of language as an instinct is wrong. People who hold this view are so impressed with children’s proficiency in acquiring language that they take for granted that modern children are born into a linguistic environment. The hard part—“inventing” a language—has already been done for them. From the day they are born, children interact with people whose minds and behaviors have been shaped by language.

  In the case of the Nicaraguan deaf children, though they couldn’t understand the words spoken by their hearing parents, the children were interacting with people who did possess full-blown languages. In creating their own sign language, these kids were not copying the languages of the people around them, but this process of creation was in no small part launched by the linguistic environment in which they found themselves.

  Heidi Lyn: No to “language.” I believe they would develop a protolanguage communication system, but if there was no need for cooperation and no culture to model communication for them, I don’t think they would develop full-blown language. I think you would need at least ten individuals for any kind of real communication to take place. I still think this protolanguage would be more developed than what we’ve seen in other species, but with only a few individuals I don’t think there would be any need even for that.