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

by Bickerton, Derek

  Once it was thought that things like warning calls were entirely automatic, like the way you blink if someone pokes a finger in your eye. The poor animal saw a leopard and yelped; it just couldn’t help it. Now researchers have found that animals, though unable to speak, aren’t that dumb. If they’re alone, they don’t call. If they’re not with close relatives, they’re less likely to call than if they’re with immediate family.

  Food calls—signals that announce the discovery of food, sometimes even its kind or quality as well as its location—can again be unhelpful to the individual if it means sharing a tasty tidbit with others instead of hogging it alone. But the same standard of inclusive fitness applies: benefit your brother and you’re giving your own genes, at least some of them, a boost. So all survival calls relate directly to increasing fitness.

  Now take reproduction signals. These may involve advertisements of immediate availability, such as the swelling of female genitalia in some primates when they are in estrus, or they may merely announce “I am a male/female of species X.” At the other extreme, they may involve elaborate courtship dances, or the construction of complex artifacts (such as a decorated bower) to attract females. The simpler signals merely ensure that the right sexes of the right species get together at the right times. Clearly, if at one time there were animals that signaled their sex and their species alongside other animals that didn’t, the first lot would meet and mate oftener than the second, so eventually every member of the species would make those signals. The more complex signals indicate not only that a mate is available but suggest that a mate of the highest quality is available. As Darwin long ago pointed out, female choice—the desire of any female to secure the best breeding stock, to bag a mate that will send her genes further into futurity—forms one of evolution’s most powerful engines. So again with reproduction we have a set of signals that directly increase fitness.

  Finally, let’s turn to social signals. These don’t have to be social in the sense of friendly; they don’t even have to be limited to social species. They can relate to any kind of interaction between members of the same species. Take for example solitary birds that defend a territory, perhaps with a single mate. The signals they send to discourage intruders fall into the social category just as much as more intimate signals like the “nursing poke” used by infant apes to get their mothers to feed them. While the increase in fitness that results from these signals may not always be as direct or as obvious as with survival or mating signals, it is not insignificant. The animal that makes a rival back off without having to fight avoids possible injury or death. The animal that encourages others to groom it benefits from more than the elimination of vermin. Given that the favor is returned, affiliative bonds are then established, status within the group is enhanced, access to sexual or nutritional opportunities is increased. Better living means longer living and more progeny—once more, increased fitness.

  WHY LANGUAGE IS SO ANOMALOUS

  Now that we’ve established two of the most basic features of ACSs—that they grew from behaviors not originally meant for communication and respond only to situations that directly affect fitness—we can begin to realize the enormous size of the problem that language poses for the biological sciences.

  People often think that the core of this problem is the uniqueness of language. It isn’t. Lots of things about humans are unique: bipedalism and absence of body hair (among terrestrial mammals, anyway), the precision grip of thumb and forefinger, even the whites of our eyes. Lots of other species have unique features too: the elephant’s trunk, the giraffe’s neck, the peacock’s tail. And the hammering of woodpeckers, the heat-sensing of pit vipers, the trap-digging of ant lions are behaviors as unique as the physical forms of elephants, giraffes, or peacocks. But no other unique feature of any species is as isolated from the rest of evolution as language is.

  Bipedalism isn’t all that special. Birds managed it. Kangaroos come close. Closely related apes get up and hunker around on their hind legs from time to time. Our grip differs from that of ape fingers only in its greater range and degree of control. Hairlessness is unique, in our case, only because it’s lifelong; the young of many mammalian species emerge naked from the womb and only later grow hair.

  Instead, let’s compare another feature that’s not human but is genuinely unique: the elephant’s trunk. In his book The Language Instinct the psycholinguist Steven Pinker actually uses the elephant’s trunk to make language seem less of an anomaly than it really is. He asks his readers to “imagine what might happen if some biologists were elephants”—as with language, some would say the trunk was too unique to have evolved, others that it couldn’t really be unique at all. But unique stuff can evolve through natural selection, so Pinker insists that “a language instinct unique to modern humans poses no more of a paradox than a trunk unique to elephants.”

  He’s wrong. An elephant’s trunk results from hyperdevelopment of the nose and adjacent parts of the face in the common ancestor of elephants and hyraxes, and anatomists can point to the exact physical ingredients that went into its makeup. But Pinker doesn’t tell you what ingredients went into the making of language. (And anyway, isn’t it a bit weird to compare a behavior with a body part?)

  Uniqueness isn’t the issue. Unlikeness is the issue. And that is something Pinker, like everyone else who writes on language evolution, doesn’t really tackle. For every other “unique” thing that’s evolved, you can see what was there before it, what evolution had to work on in order to produce it. Not with language.

  Take what looks on the face of things the best, if not the only candidate: the communication system of the last common ancestor. But to get from any ACS to language would involve two tasks, just for starters. First, evolution would have to find the raw material—some already existing behavior that could be taken and twisted and refined into an appropriate medium. Second, and this is a task orders of magnitude harder, it would have to uncouple this new system from currently occurring situations involving fitness.

  That’s actually three tasks in one. The system would have to be uncoupled from situations, from current occurrence, and from fitness. Let me explain.

  ACS units—all the calls, flashes, and gestures that constitute ACSs—are all anchored to particular situations: aggressive confrontation, search for a sex partner, appearance of a predator, discovery of food, and so forth. They would be meaningless if used outside those situations. Language units—words, manual signs—are not. They’re meaningful in any situation. If I say, “Look out, a tiger is about to jump on you,” you may know I’m just kidding, but you know perfectly well what the words mean—they mean exactly what they’d mean if a tiger really was about to jump on you.

  Some linguists and philosophers may still tell you that words relate directly to individual objects in the world—dogs, chairs, trees—but they don’t even do that, or rather they do so only indirectly, via the concepts of these objects that we have in our minds. If I say, “Dogs bark,” what actual dogs am I referring to? Big dogs? Brown dogs? The dogs down the road? Obviously not. All dogs, then? Not necessarily. I didn’t say all—my statement cannot be refuted by a barkless dog. What it means is, “Dogs as a general rule bark,” or “Barking is a fairly reliable sign of dogginess.” Well, just point out to me “dogginess” or “dogs as a general rule.” You can’t; there’s no such critters. We have what may be vague but are fully functional ideas of what dogs are like, and that’s what we’re referring to. If we want to refer to a specific dog or dogs we can’t just say “dog” or “dogs”; we have to say “this dog,” “those dogs over there,” “the dog with the waggly tail.” So in order to get to language, the reference of meaningful units—signs or words—has somehow to be shifted from concrete situations to the concepts we have of particular things in the world.

  But what ACSs are grounded in aren’t just any old situations. They’re situations that are occurring right now, at the very moment the ACS signal is being waved or flashed or
yelped. No animal can use a predator alarm call to remind its fellows about the predator that appeared yesterday, or the predator that often hangs around the water hole. No chance of an advance warning, no reprise of what went wrong last time. Each utterance of an ACS unit is tied to whatever is going on in the immediate vicinity right at that moment. Words, on the contrary, are relatively seldom used about what’s going on before our eyes. We can usually see that for ourselves, so what would be the point? We still have body language; for things like showing how far we’ll push a confrontation, or how strong our sex desire is, good old body language works as well for us as for any other species, often much better than words. On the other hand, with words we can do stuff way beyond the here and now. We can exchange ideas about things infinitely remote in space and time, things we may never have seen, even things like ghosts or angels that may not exist. So somehow communication has to be released from bondage to what’s happening right now.

  Finally, there’s freedom from fitness. We have seen how the function of ACS units is to improve fitness; no unit even comes into existence unless it improves fitness in some way. Some people have conjectured that language as a whole increases fitness. Now it may well be that at some stage in evolution, ancestors of ours who had more developed language skills left more offspring than those whose skills were less developed. But though this is a plausible conjecture, there’s zero evidence for it, and in any case it’s a totally different issue. The point is that no ACS signal occurs in any situation that doesn’t directly involve fitness. And this certainly isn’t true of words or signs. They can refer to anything at all, whether it has any connection with fitness or none. And, leaving aside one or two exceptions like “Fire!” or “Help!,” a word can’t, in itself, by itself, contribute in any way to fitness. And these exceptions, when you come to think about them, are more like ACS calls than regular words—they’re tied to situations in just the way ACS signals are. If you doubt me, try shouting “Fire!” in a crowded theater, or ask yourself whether “fire” works the same way in “Help! Fire!” as it does in “There’s nothing like a nice warm fire on a winter’s evening.”

  Let’s do another thought experiment. There must have been a time when the first system that broke the ACS mold—the first protolanguage, let’s call it—had ten units or fewer. So think of any ten words or signs that, singly or in combination, would increase the survival chances and/or procreative capacities of their user.

  There are some constraints on this exercise, of course. There’s no point in saying things that could equally well be conveyed without words. Expressions like “I’m hot!” or “Check this for size!” don’t qualify; nonlinguistic means can express them more than adequately. Also, first words have to be plausible as first words; they can’t be abstract, but must be things whose meaning could easily be demonstrated, by mimicry, pointing, or whatever. Finally, their message can’t depend on the way they’re assembled; most people in the field agree that words came before syntax. So while you’re allowed to string words together in an ad hoc fashion, the final meaning can’t depend on the positions the words hold with respect to one another.

  No prizes, I’m afraid. If I were to offer prizes, you’d have to swear you hadn’t read past chapter 5, and I’d have to believe you.

  Why is this experiment important? Why “ten words or fewer”? Why not “twenty” or “fifty” or “a hundred”? I mean, give language a break; what use would you expect fewer than ten words to have?

  Well, the point is that if the first few words did not have some immediate and tangible payoff that couldn’t have been obtained by simpler means, language would never have gotten past ten words, or even that far. Evolution has no foresight. It doesn’t think, well, if we can crank language up to say fifty or maybe a hundred words, here’s all the nifty things we can do with it. Actually I’m being generous with ten. From word one, language had to pull its adaptive weight, confer some kind of benefit. If not, then nobody would have bothered to invent any more words.

  ISSUES OF NEED AND UTILITY

  So to obtain freedom from fitness, freedom from situations, freedom from the here and now all in one package, so to speak, represents an enormous task, a task without precedent in the three billion years since the first primitive life-forms emerged on this planet.

  Think of it. Think of all the billions of species there have been, in all those years. Not a single one of them failed to get by with a standard ACS. All the things they needed to do could be done with that ACS. And in ACSs in general, there was nothing you could call progress.

  You might think that, since chimps are more complex than dogs, and dogs are more complex than crickets, chimps would have more complex ACSs than dogs, and dogs would have more complex ACSs than crickets. It’s true that there’s a rough—a very rough—correlation between the complexity of a species and the number of units its ACS has. Fish tend to have more than insects, mammals more than fish, primates more than other kinds of mammal. But that’s on average: ranges overlap, and the systems themselves, for all the very different means they exploit, are all startlingly alike. They all share the same limitations: they all consist of single, unrelated signals that can’t join with one another to make more complex messages, can’t be used outside of particular situations, can’t do anything but react to some aspect of the here and now.

  If all species but ours get along with such systems, there can be only one reason. Other animals didn’t get language because, bottom line, they didn’t need language.

  I can hear people saying, “No, no! Their brains weren’t big enough!” I’ll deal with big brains in a minute. For the moment, let me point out that the following species have been proven experimentally to be capable of learning very rudimentary forms of language: chimpanzees, gorillas, bonobos, orangutans, bottlenose dolphins, African gray parrots, sea lions—all of the species most closely related to us and some that are more distantly related. And that’s pretty much all the species people have tried to teach language to. I don’t know of any cases in which people have tried and failed, though I wouldn’t bet on your chances with frogs. Seriously, it looks like any species with a sufficiently complex brain (“sufficiently” still being a black box) can acquire some kind of protolanguage, so it’s need, not brain size, that’s the most crucial factor.

  It follows as night follows day that if humans got language, they can only have got it because they had some pressing need for it. Some need, moreover, that no other species (or at least no other species of even remotely comparable complexity) ever had. There must have been something they needed to do in order to survive that they couldn’t do within the limits of a standard ACS.

  People have always wondered how language began. Only since Darwin have they rephrased the question as “How did language evolve?” But even after Darwin the idea lingered—seldom explicitly stated, but almost always implicit—that we could have gotten into language through things we already did, simply because we could do them better with language. People seem to have thought: Here’s all these other animals, all trying to communicate as best they can, then here’s us with our bigger brains doing it better—end of story. Hardly anyone seems to have realized the immense uniformity of ACSs, under their superficially different guises, or how closely connected ACSs are to specific requirements of situation and fitness, and hence how radical a departure language was.

  What we are being asked to believe is the following:

  Every other species took countless millions of years to develop rudimentary communication systems tied inexorably to the things they needed to do in order to survive.

  Our species in a tiny fraction of that time developed a vastly more complex system just so that we could do things we already did, and that other species did too, better than those other species could do them.

  Put in such stark terms, no one who believes in evolution is going to buy this. Such beliefs survive precisely because they are seldom made as explicit as this. Yet they underlie probably a large m
ajority of explanations as to why and how language began.

  For instance, a generation ago the belief was widespread that language had to do with tools—with the making of tools, or perhaps the teaching of others how to make and use tools. Then we found that chimps make and use tools: sponges of leaves to soak water from hollows, trimmed sticks to fish for termites in termite mounds. Christopher Boesch showed that chimps on the Ivory Coast not only used tools to break up palm nuts but showed their children how to do it. Admittedly, their tools were pretty simple, but so were those of our ancestors more than two million years ago. If apes can do this much without language, why would we have needed such a revolutionary step to do similar things?

  Then there were those who claimed hunting was the decisive pressure. This idea really never had the plausibility even of the tools idea. First, there’s no evidence that our early ancestors hunted, except sporadically, when the opportunity arose, and even then they didn’t have the weaponry to deal with anything much above rabbit size. All those fanciful pictures you see of shaggy fellows poking spears into mastodons relate to relatively recent history, probably to our own species, which is less than 200,000 years old (the last common ancestor was a minimum of five million years ago—and quite likely six or seven million). Second, lots of other species (wolves, jackals, lions) hunt cooperatively, and get on fine without a word between them. The clincher came when chimpanzees were observed hunting colobus monkeys. It was as if they were saying, “Hey, you go this way, I’ll go that, Bill can block him over there while Fred cuts him off at that branch.” But they weren’t. They weren’t saying a word, but the monkey got caught and eaten just as efficiently as if they had been.

  By the 1990s tools and hunting had fallen into disrepute. Now all the talk was of social intelligence. Ethological studies over the previous two decades had shown that the social intelligence of primates, particularly of our nearest relatives, the great apes, was pretty high. Apes formed alliances, played politics. They schemed with and against one another to get access to the most desirable females. They engaged in what researchers Richard Byrne and Andrew Whiten called “Machiavellian strategies,” faking one another out, making phony alarm calls—in effect, lying even without words—as they struggled to enhance their status within their group. Indeed, their social life wasn’t all that different from the social life of humans. So something involving social interaction must have been the pressure that selected for language.