Adam's Tongue: How Humans Made Language, How Language Made Humans

by Bickerton, Derek

  After all, hadn’t apes, once they got over their initial bafflement, proceeded to pick up signs on a minimal exposure? If they could do this with brains less than a third the size of ours, why couldn’t our ancestors, with brains up to twice as big as those of apes, have done the same?

  The adaptive benefits were beyond question. All the things I’ve dismissed as possible selective pressures for actually starting language—instructing the young, competing socially, displaying sexually, making artifacts, gossiping, performing rituals, and so on—were things you could use language for, once you had it. All these activities would be enhanced, some a little, some immeasurably, by a species whose discourse could range back and forth in time, hither and yon in space. Surely one after another, each of these activities would acquire its appropriate words. And while the absence of any kind of regular structure might keep actual utterances down to a few words at a time, you should soon have quite a respectable protolanguage.

  I cannot now see a single compelling reason for believing this, and there is much that points in a very different direction. The most impressive evidence for a long, slow gestation of language can be found in the most tangible data we have—the fossil and archaeological record of our ancestors over the past two million years.

  THE LONG STAGNATION

  When paleontologists describe those two million years in books aimed at a popular audience, here’s the kind of picture we usually get:

  Long-continued increase in size and complexity of the brain was paralleled for probably a couple of million years by long-continued elaboration and “complexification” . . . of the culture. The feedback relationship between the two sets of events is as indubitable as it was prolonged in time [Phillip Tobias, South African paleontologist].

  The fossil and archaeological record for Homo picks up around two million years ago in East Africa. And what a record it is! Brain size “took off” and subsequently doubled from approximately 700 cubic centimeters to 1400 cubic centimeters . . . Recorded tool production also accelerated in Homo, spanning from initial clunky stone tools to contemporary computer, space and biological engineering [Dean Falk, professor and chair of anthropology at Florida State University].

  Tell that to a Martian, and he’d probably assume that things like stone bridges were invented about a million years ago (it was the Stone Age, wasn’t it?), the wheel (Flintstones-style) maybe half a million years ago, and steam trains about a hundred thousand years ago. I’ve never understood why reputable experts in the field say things that they know perfectly well are not true. Is it because the facts about human evolution are so totally at variance with anything you’d expect that paleontologists simply don’t want to admit them? I can’t think of any other explanation.

  Tobias is right that the increase in brain size went on for the best part of two million years, but entirely wrong when he claims that “the culture” elaborated and complexified. Falk is right that brain size doubled—actually, more than doubled, in Neanderthals—but mislead-ingly suggests that the progress from “clunky stone tools” to modern technology was smooth and evenly paced. It wasn’t.

  As I mentioned in chapter 7, the standard tool of Homo erectus was the symmetrical, pear-shaped object known as the Acheulean hand ax. For more than a million years, this remained unchanged. And the few other tools that developed, the so-called borers and scrapers, were basically variations on this tool, making it thicker or thinner, more pointed or more long-edged. There were no notched or tanged points, like the Aterian points described in chapter 10. There were no hafted tools—tools that required two or more parts to be joined. There were no artifacts made from bone or ivory—nothing but variations on the one-stone tool. During the entire period, about the only innovations were the taming of fire, the invention of spears, the erection of primitive shelters, and the start of serious big-game hunting. Even for these milestones, firm evidence comes only from the latter part of the period.

  Yet more than a million years ago there were already plenty of hominids whose brain sizes, more than 1,000 cubic centimeters, fell within the normal range of modern humans. If increased brain size spells increased intelligence, and if people with brains smaller than those of some members of Homo erectus can talk and write and do science and invent stuff, how is it that the species that preceded ours endured the same harsh conditions of life, without any serious attempt to improve its lot, for the best part of two million years? For far from the smooth progression Falk claims, everything in human civilization—herding, agriculture, cities, industrialization, and the exploration of the solar system—was squished into a bare two-hundredth part, 0.005 percent, of that period.

  If language is indeed what drives human thought, and if language began two million years ago, how can such things be?

  Well, you could argue that I’m wrong, that language did not start two million years ago, that it did not start until much later. But then we’re back with the problem, what else could have started it? I know of no development in prehistory later than power scavenging that could have served as a trigger. Of course that doesn’t mean that there wasn’t one. But there isn’t even a promising candidate in sight. Unless one turns up, power scavenging remains the likeliest cause.

  Besides, the later the birth of language, the harder it becomes to find time for the extensive rewiring of the brain that language required. Short of some magic mutation, that process must have taken quite a while. Moreover, as we’ll see in subsequent sections of this chapter, there were hurdles over and above brain rewiring that would have delayed things further. It’s tempting to hypothesize a rapid and relatively recent coevolution of language and culture, starting from scratch, over the last hundred thousand years or so. Such a scenario would feel comfortable, would fit well with contemporary thinking. But if we ask, “What evidence supports it?” the answer is “None.”

  There is still another alternative. Some think that language developed fully at a much earlier date. They will point to some of the hunter-gatherers who still survive in the modern world. In language and intelligence, these are completely normal modern humans. Yet their tool kits and all the things in their material culture are hardly more complex than those of Cro-Magnons. So why shouldn’t our remote ancestors have sat around the campfire hundred of thousands if not millions of years ago, with a modern language capacity or something not too far short of it, telling stories, politicking, wooing one another with words, happy as Larry in their free hunter-gatherer lives, simply preferring not to weigh themselves down with all the massive impedimenta of civilization?

  Because hunter-gatherers lie at the conservative end of a spectrum of behaviors—a tiny minority who, by chance or by choice, failed to innovate in the way the vast majority of our species did. Previous hominid species did not show any spectrum of behaviors. Their behaviors varied as little as do those of chimps or vervets or any other nonhuman species, whether over eons of time or thousands of miles of space. The idea that a species could have a highly adaptive capacity that none of its members ever exercised goes against all we know about how species construct their niches. To the contrary, species exploit their capacities to the full, even expand on those they already have. Any species pushes the envelope to the extent that its genes and the phenotype they produce allow it. It’s unthinkable that any species should have the power to radically alter its behavior but that no member of that species would ever use that power. So the idea of a linguistic species with next to no technology is as problematic as that of a species developing language from scratch at a very recent date.

  So, on balance, I’m convinced that an early start coupled with a painfully slow development of language fits better once all the evidence we have today is taken into account.

  BARRIERS TO LANGUAGE

  To understand why that start was so slow and difficult, we should look a little more closely at what the recruitment procedure actually provided. It was not a sapling, not even a young shoot. It was more like an acorn—something that, with
good luck and nourishment, might one day expand into a tree. But the future shape and form of language, or for that matter even of protolanguage, was no more visible in the handful of recruitment signals than the shape of the oak tree it will one day become is visible in an acorn. Recruitment had broken the mold of the ACS; that was the crucial step. And it had broken that mold not in a bee or an ant or any species with a microscopic brain, but in the one that currently had the highest ratio of brain to body of all the species on earth.

  But that species wasn’t smart enough to know what had happened to it. Its members, unlike the trained apes, were not surrounded by another species, one that already had language, one that was single-mindedly determined to teach language to them. They were pioneers. They were alone in the universe. They couldn’t have had the faintest inkling of the possibilities their discovery had opened up for them, and there was no one there to take them by the hand and show them those possibilities. They were probably not even aware that they’d done anything new.

  So let’s look more closely at recruitment signals, see what they had and could do and what they didn’t have and couldn’t do. The signals

  had “functional reference” in that they specified one or more—probably several—megafauna species whose exploitation required recruitment. (Functional reference had, you’ll recall, already been achieved by the specific warning signals of vervets and other primates, although not by any of the great apes.)

  had displacement, something that hadn’t been achieved by any primate species; they contained information obtained well ahead of utterance, about things well outside the recipients’ sensory range.

  were created and learned, rather than hardwired.

  contained protonouns—names of species—and probably also protoverbs: noises and gestures that could be interpreted as “Come!” or “Hurry!”

  That’s the plus side. However, in the first chapter I pointed out three further properties that had to be achieved before even a protolanguage could begin. Signs would have to be uncoupled from situations, from current occurrence, and from fitness. In fact, signals were still

  coupled with situations; certainly at first, perhaps for a long time, they were used only for recruitment, and thus made sense only when there was a large dead herbivore not too far away and other group members had found it and were trying to get attention.

  coupled with current occurrence; even though displacement had been achieved, you couldn’t yet talk about the large dead herbivore you found last month, or suggest strategies for finding more such in the future. You couldn’t talk about such things until you had words of some kind showing that you were talking about the future or the past.

  coupled with fitness; even though power scavenging involved cooperating with non-kin, it still contributed to individual fitness, precisely because, without cooperation, each individual would be deprived of possible food, and with it, each individual would benefit.

  Breaking those couplings was an essential prerequisite for the development of even the simplest kind of protolanguage. But as we saw in the previous chapter, the couplings weren’t there by accident, each with its own connections that could be easily snapped. They all stemmed from one cause—the inability of prehuman minds to deal with anything other than the animal’s immediate circumstances. In other words, they were incapable of displacement.

  Let me dwell on this a while, because it took me a long time to fully grasp it and I suspect I was in good, perhaps almost universal, company.

  When most people look at language and contrast it with ACSs, displacement is not often picked as its most salient feature. People think of learning; ACSs are innate, language has to be learned. People think of arbitrariness; ACS signals often show a direct relationship to what they signify (cringing postures to indicate submission; loudness, repetitiveness, or intensity of vocal calls to indicate firmness of purpose). In language, however, words of similar meaning—dog, chien, perro, Hund—bear no obvious relationship to what they describe or even to one another. People think of combinability; ACS signals won’t combine at all, while words, phrases, and clauses will combine without limit. People think of complexity; while language has an intricately layered structure of sounds and units of meaning and syntax, ACSs are single-level, what you see is what you get. Displacement gets second billing, if it gets any billing at all.

  It’s only when you fully appreciate what displacement means, how the absence of displacement is not just a casual feature of ACSs but a crucial defining feature of prehuman minds, that you can start getting the complete picture. That picture shows you two complementary things. It shows you how achieving even the rather superficial form of displacement found in recruitment signals was the greatest single step any communication system could have made in the direction of language. But it also shows you the immense difficulty, even when that step had been taken, that’s involved in creating true displacement, true escape from the here and now in which all species had hitherto been trapped. To do that, you had first to make concepts, mental symbols of reference no longer bound by particular instantiations of the things referred to. Only with such abstract symbols could you roam mentally, freely through space and time as we do today, in both language and thought.

  Still, the displacement in recruitment signals formed a wedge already driven deep into the status quo—a wedge without which we would either still be wandering houseless across the savannas, or more likely would long ago have gone extinct.

  Let’s see how that wedge could have worked.

  FROM SIGNAL TO WORD

  In the initial, recruitment phase of protolanguage, there were, properly speaking, neither concepts nor words. Recruitment signals weren’t words. They were iconic and/or indexical signals that, to those who used them, were no different from all the other ACS signals they already had. Signals had to become words and words had to give birth to concepts before anything you could even call a protolanguage could be born.

  The signals associated with recruitment were the only signals in the protohuman ACS that had displacement, and in the beginning they were tied to what had happened or was about to happen. What you might loosely want to call the “mammoth” signal might have been better interpreted as “We’ve just found a dead mammoth and we want you to come help us butcher it.” But precisely because the signal—pantomime of the live animal’s looks or movements, vocal imitation of the noise it made, or whatever—drew attention to the nature of the animal rather than simply pointing to an actual appearance of that animal, it became available for use in other circumstances that had to do with mammoths.

  There’s no recourse here but to tell just-so stories. Walking along a dried but still muddy watercourse, brief fruit of rare rains, an older and a younger individual see a set of deep footprints. The older one points and gives the mammoth signal.

  A bunch of young ones, some little more than infants, are playing. One or two are old enough to have gone out on megafauna butchery expeditions. (How old did that have to be? My guess is, they needed all hands, so as soon as you could run fast enough and throw far enough, you were on.) The older ones mime the expedition, interspersing their mammoth signal with boastful gestures and noises. The younger ones listen and imitate; in their still-plastic brains, sounds morph into images and out again.

  A group comes upon a heap of big, stripped bones. This was one they didn’t find in time. Some of them, in tones of disappointment or anger, make the mammoth sound as they turn the bones over, looking for scraps the other scavengers may have left.

  Gradually the sound is getting divorced from the situation that gave rise to it. Put something into enough different contexts and its particular details become blurred; it gets closer to becoming an arbitrary symbol.

  At the same time, a representation is formed in the brain: a representation of the mammoth sound. How is this different from the representation of all the other ACS signals? At first, not at all—except that it’s been learned; it’s passed from generation to generation.
For the moment, that doesn’t carry any consequences. If there are other signals that go with it—signals that might mean “Hurry!” or “Come!”—they go with it only in the recruitment scenario, or in the pantomime reenactment of it (which might have been done, as ritual, by adults as well as children, but we’ve no idea when rituals began). However, as the intimate connection between signals and situations begins to erode, those signals become increasingly more wordlike, and increasingly available to the combinatorial process that is essential, even in protolanguage.

  But what we must bear constantly in mind is that we’re not dealing with a species that was sitting around passively waiting for happenstance or genetic drift to carry it along. We’re dealing with a species that was actively carving out the niche of high-end scavenging, and this process in turn fed into the growth of protolanguage.

  NICHE CONSTRUCTION DRIVES LANGUAGE

  Think about it. Here’s a species that’s developing a new niche. Alone among species, it can drive other scavengers away from dead megafauna long enough to access the carcass and make off with the best cuts. This species has the biggest brain going and the carcasses form its richest possible supply of food. So this species is going to go on, millennium after millennium, scanning the skies for cruising vultures but otherwise just accessing whatever carcasses it happens to stumble upon. Whoops! Look at that! Why, I do believe it’s a dead giant hippo!

  I don’t think so.

  I think, before very long, it occurred to some brighter-than-average erectus to take a more proactive approach. After all, wherever the megafauna wandered, they left a pretty big signature—dung heaps, trampled grass, torn-up vegetation, footprints around streams and rivers. Some maybe had regular migration patterns. The logical thing to do was not to wait until you stumbled on a dead one, but to track the herds, pick out sick or injured ones in advance, be there when the dying commenced, and get to work probably before the moment of death, as soon as the animal became unable to defend itself.