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

by Bickerton, Derek

  As we saw above, more than one species of monkey give predator alarm calls. The vervets’ calls are just more varied than most. They have a call for eagles, a call for leopards, and a call for snakes. Why can’t we say that these calls are in effect the vervet “words” for eagle, leopard, snake?

  Because, as I pointed out in chapter 1, while any word can be used in the absence of what it refers to, no animal call can. Even if a call is used deceptively, to distract aggressors or secure a tasty tidbit, those who hear the call have to assume that a predator’s really there. If they don’t, the ruse won’t work. We may choose to call this “meaning,” but it’s quite different from the way words in any human language work. Realizing this, some people have chosen to call it “functional reference.” That’s a way of saying, well, words refer fully, in the sense that you can use them in meaningful ways regardless of whether what they refer to is there or not. But since the leopard call isn’t used for anything other than leopards, it has the effect of drawing attention to leopards, and thus discharges the most basic function of reference, which is to pick something out and direct your attention to it.

  However, there’s a function more important than reference that these calls perform, and that’s eliciting a specific response from the hearer, as follows:

  Eagle call: look at the sky, get ready to hide in the bushes.

  Leopard call: look around, get near a tree you can climb real quick.

  Snake call: look down at the ground all around you.

  Do these sound like the names of different animals? If we try to translate them into humanese, the animal’s name may not even be included in the translation. Take the eagle call. Does it translate as “Look out, an eagle is coming!” or “Danger from the air!” or “Quick, find the nearest bush and hide in it!”? Any of these three seems more efficient, more functional, than simply “eagle.”

  (Note that even here, the potential ambiguity is not the same as the ambiguity sometimes found in words. Ambiguous words mean totally different things. A “bank” is either a place where money is kept or the side of a river; “rape” is either a violent sexual act or a commercial crop. But translations of animal calls represent various possible interpretations of the same thing. Remember this; it will become important in the next chapter.)

  What do my three translations of the eagle call have in common?

  They are all complete utterances—complete in themselves.

  How does the word “eagle” differ from that?

  It isn’t complete in itself. It tells us something, but not enough. Is there an eagle here right now, or are you talking about yesterday, or the chance we’ll meet one tomorrow? Are you making a general statement about eagles, or a particular statement about one eagle, or just listing major bird species? It could mean any or all or none of the above.

  For me to know what you’re talking about, you have to predicate. You have to combine “eagle” with some other word or words that will tell me which of the many possible things you mean. But for me to know what an alarm call means, you don’t have to predicate. The call is enough. I’m up the tree or into the bushes already.

  Now we can see why signals in ACSs are never combined.

  It would make no sense to combine them. They’re not words that have to be combined to form a particular meaning. They’re specific responses to specific situations, complete in themselves, and more than that, they’re responses that have had, in the past, a demonstrated capacity to improve the fitness of those that used them. If those responses hadn’t produced longer lives and more offspring for their users, evolution would have erased them.

  It’s not that animals are too dumb to put things together. Just that the calls and signs and all the other things they communicate with weren’t designed to be put together. And if you did put them together, one wouldn’t “modify” the other; together, they would mean exactly what they meant separately. One wouldn’t change or affect the other in any way.

  This hasn’t always been apparent to everyone. In 1964 the journal Current Anthropology published an article called “The Human Revolution,” by Charles Hockett, one of the leading linguists back then, and his colleague Robert Ascher; the journal thought so highly of this article that it was reprinted, unchanged, twenty-eight years later. (Until 1990 or thereabouts, the pace of change in language evolution studies was, indeed, glacial.) Hockett’s intuition was that language began when some protohuman, encountering a situation in which there was both food and danger, blended the food call with the danger call. Then this, the first combination of meaningful units, led to more of the same, and language was up and running.

  Hockett’s analysis ignores the following facts:

  Words combine as separate units—they never blend. They’re atoms, not mudballs.

  To a naive animal, a blended call would probably be meaningless.

  Even if the blend had been interpreted, how could it have made sense? If it had been a predication, there are only two possibilities:

  “Dangerous food”? Unlikely; danger calls, as we’ve seen, at least roughly specify the source of the danger without further addition. No animal I’ve ever heard of has an alarm call for poisonous substances.

  “Edible danger”? Come on!

  All the blended call could have meant was, “There is food, but there is also danger.” But, just as I said, this is no more than the sum of what the two calls mean in isolation. As such, it would have moved us not one inch closer to anything you might call language.

  The dream of strong continuists is to find precursors of words and precursors of syntax among other species. That would be the easiest and most obvious way to establish true continuity between ACSs and language. But it isn’t the right way, simply because while words (or manual signs, or any similar kind of linguistic unit) have little meaning until they’re combined with other words, animal calls (or any other ACS units) mean no more when they’re combined than they mean in isolation.

  So why would any animal in its right mind even want to combine them?

  It’s a waste of time looking among other species for precursors of words or precursors of syntax, because animal communication was not designed by evolution as an inferior substitute for language. It wasn’t that animals were slowly and stumblingly trying to get nearer to language, but didn’t quite know how. What we’ve been looking at as if they were ACS limitations are really only limitations from our own peculiar perspective. For other animals, ACSs do their job just fine. It was only some aberrant ancestor of ours that needed something a little different (and got much, much more than it had bargained for).

  So if we want to demonstrate real continuity in evolution, we should be looking not for linguistic precursors, but for some point of flexibility in ACSs, some point where the right selective pressure could force a distortion that might ultimately lead to the creation of words, and, later on, the creation of syntax. Because these things—words and syntax—are total evolutionary novelties, things useless and meaningless outside language. Things whose like had not been produced by evolution in all the three-billion-plus years it had been working—not because, in all those years, evolution had “failed to produce language,” but because it had succeeded in producing something wholly different from language. Not some poor limping thing longing to be language, but a tool that served the needs of its users perfectly well.

  Talk about primate-centric—people who look for precursors of language are homocentric. Instead of looking at communication objectively, from a neutral perspective, they seem like they’re strangled by language—locked into the worldview of one rather peculiar species.

  ESCAPE FROM HERE, ESCAPE FROM NOW

  To find where ACSs are flexible, we must yet again compare them with language—not so we can disparage them, but so as to better understand the very different ways in which they work.

  One of the things ACSs don’t do but language does is refer to anything that isn’t right there, at the moment you make the call, immediatel
y within the range of your senses. Once more we must ask the question: Is this accidental or is there a principled reason that things are the way they are, and not otherwise?

  Philosophers of language might say, it’s because the signs of ACSs are indexical, not symbolic.

  An indexical sign is one that points directly at its referent. Vervet predator warnings are good examples of indexical signs. A symbolic sign, however, can stand in place of its referent, even when that referent is thousands of miles away or thousands of years back in history.

  But this just names the difference; it doesn’t explain it.

  We could ask, why are ACS signs indexical, rather than symbolic? But a more revealing question would be, which comes uppermost, the informative or the manipulative?

  We must tread carefully here. All communicative acts are informative, in some sense, and in that sense both ACSs and language are both informative and manipulative. Body language—part of the human ACS—is informative; if in spite of your placatory words your body language shows me that you’re angry, that’s information, a kind of information I wouldn’t have had if you hadn’t used body language to express your anger. Similarly, any linguistic act can be manipulative; a purely factual statement about the weather could be aimed at convincing you to stay home with me, rather than go out with someone else. So it would be easy, and true, even if quite uninformative, to say: ACSs are both informative and manipulative, and so is language. What’s the difference?

  The difference is that an ACS is primarily manipulative and only secondarily informative, whereas language is primarily informative and only secondarily manipulative.

  ACSs may provide information, but that information is merely a byproduct. Their primary function is to get you to do things that enhance my fitness. (If they enhance your fitness too, you’re just lucky.) But if ACSs are designed to respond to situations and manipulate other individuals, you can see why they have to be bound to the here and now. You can’t respond to a situation that’s distant in time or space (at least you couldn’t until they invented TV). You can’t manipulate somebody who isn’t there, or do it at any time other than the immediate present. What looks to us like a limitation is, in ACS terms, simply a logical necessity.

  Language, however, puts information first and manipulation second. Suppose I were to explain to you Einstein’s theory of relativity, or Chomsky’s theory of a biologically based language organ. My purpose in giving you this information might well be that of impressing you or even ultimately mating with you (although anyone who could be induced to mate by such means would have to be pretty weird). But I would be trying to manipulate you by means of information rather than just inadvertently giving you information in the course of manipulating you.

  It follows that language doesn’t need to be bound by the here and now. Information (whether used to manipulate or not) can be about things that have already happened or things that might happen but haven’t done so yet. It can be about things before your very eyes, but it’s much more likely to be about things that aren’t, because an important—maybe the most important—feature of information is its novelty. In most contexts, old information is plain boring. (The big exception is bonding, whether of lovers or party members—have you ever heard a politician’s speech that contained anything you hadn’t heard umpteen times already?) In contrast, ACSs endlessly repeat the same old signs for the same old situations—novelty would be disruptive, dysfunctional. And if the situations weren’t ones that repeated endlessly, evolution would never have gotten around to making signs for them.

  By now it should be clear why ACS units are indexical and language units are symbolic.

  ACS units are indexical because they’re designed to manipulate others. Those others have to be right there in the present time at the present place if they’re going to be manipulated. So even if information is exchanged, it can only be information about the here and now.

  Language units are symbolic because they’re designed to convey information. Information can be past, present, or future, here, there, or anywhere. But to the extent—a very considerable extent—that its value lies in its novelty, it had better not be about the here and now.

  But this, of course, is no help at all in explaining how anything could have come to be symbolic in the first place.

  WHICH WAY TO THE RUBICON?

  A decade ago, Terrence Deacon (then at Boston University, now at UC Berkeley) published a widely acclaimed book called The Symbolic Species. In it he claimed that what most sharply distinguished humans from other species was the capacity to create and use symbols. In reviewing it, I said I thought he was wrong and that the really distinctive thing was syntax. And later on, we debated the whole thing publicly, twice (in Seattle and in Eugene, Oregon). But quite recently I came to the conclusion he was right and I was wrong—about symbolism versus syntax, at least.

  Certainly, the reason I criticized his book was not the right one. The right one, as I now see with the twenty-twenty vision of hindsight, was that he didn’t deliver on his implied promise. There are chapters and chapters on why animals don’t get symbols and why we must have gotten them in order to have become the kind of creatures we are. But there’s nothing about how we got symbolic words. About how we got symbolism, sure: it was through ritual. And which ritual, in particular? Would you believe marriage?!

  No, to be fair to Terry, he wasn’t suggesting that the first words were “Do you take this woman . . .” He had in fact what was a rather good argument, at least from an anthropological perspective. It was that among prehumans, where men went off to get meat and women stayed close to home collecting the veggies, there was always the chance some guy would sneak back and mate with your mate. Since the meat you brought back was to share with her and her children, you ran a substantial risk that all your efforts would go to promote your cuckolder’s genes at the expense of your own. So, to avoid all the stresses and strains, jealousies and conflicts, that would result, some form of generally accepted pair-bonding ceremony had to evolve. And indeed, marriage of some sort seems to be universal in human societies (though I doubt it reduces promiscuity all that much).

  But the nearest Terry got to explaining how symbolism went from (admittedly wordless) rituals to actual words was to claim that although “vocalizations” were found along with all these “ritual gestures, activities and objects,” “probably not until Homo erectus were the equivalents of words available.” How did they become “available”? How did anything get to mean anything? Not a word.

  Yet I’m convinced now that Terry was right in his main contention, that symbolism was the Rubicon that had to be crossed for our ancestors to start becoming human. I had argued for syntax because, while trained apes could be taught things that in meaning were roughly equivalent to words, and while they could (apparently without much if anything in the way of further instruction) string these words together into a kind of protolanguage, they’d never acquired anything you could call syntax, even though, in at least one experiment, simple elements of syntax had been explicitly taught to them. But syntax, I began to realize, may have become possible only because two million years of protolanguage use brought about significant changes in its user’s brain. If that was so, it was ridiculous to treat something apes never had a chance to get to as the main distinction between them and us. It made far more sense, as Deacon had proposed, to see the main distinction as arising at the very beginning of language—at the first step, the creation of symbols, that set the whole process in motion.

  For if it’s useless to look for precursors of words, or precursors of syntax, there’s nothing left to do but look at the units found in ACSs and see if there are any, anywhere, that might, under special circumstances, take on at least one of the properties that symbolic units—words, or the signs of manual sign languages—possess.

  And as we have seen, the most salient characteristic of symbols is that they can refer to things outside of the here and now. This capacity is something linguists
generally refer to as “displacement.”

  So let’s review Marc Hauser’s tripartite division of ACS units into social signals, mating signals, and survival signals. Which of these classes is the likeliest to contain something capable of displacement?

  We can quickly dispose of the first two. Social signals wouldn’t be social if they didn’t involve manipulating the actions of other group members, something that can only be done in the here and now. Mating signals, apart from those that merely indicate species, sex, and/or availability, consist of advertisements of the fine genetic stock the advertiser comes from—displays of fine feathers, aerobatic skills, capacity to bear handicaps, ability to defeat rivals, or whatever. These qualities can only be exhibited in the present; there’s no way any animal can convey any equivalent of “I may look a bit run-down at the moment, but you should have seen me last week.”

  This leaves us with survival signals, which again fall into two broad classes: alarm calls and food calls. Alarm calls we have already dealt with at length, and seen that they are inextricably linked to the appearance of predators or at least their presumed appearance (presumed by both caller and receivers if the caller mistakenly believes there’s a predator there, and by the receivers even if the caller is lying to them). Food calls are mostly immediate reactions to the discovery of a food source, intended to be audible (or if signed, visible) to group members in the immediate vicinity. None of these is a likely candidate for displacement.

  However, suppose that the food was some distance away from any other group member, and suppose that a measurable length of time had to elapse between the discovery of a particular food source and the transmission of news about that discovery to other animals. If any animal signal could be used in this kind of situation, wouldn’t it qualify as an escape from the prison of the here and now, the first true case of displacement?