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

by Bickerton, Derek

  We cannot simply assume that what we might choose to describe as “the eagle call” or even as “something like a word for eagle” necessarily refers to eagles. ACS units aren’t designed to refer; they’re designed to get other animals to do things. Such units don’t really translate into human language. We can give an approximate meaning, or several possible meanings, in terms of our language, but the idea that underlying both is the exact same semantic expression is simply baseless. And this absence of common ground is not something that happens to be so; it’s something that has to be so.

  The deepest reason why a holistic protolanguage is impossible is the profound fallacy that underlies Alison Wray’s proposal. Once again, it’s the fallacy that language and animal communication are fundamentally the same kind of thing. Animals were struggling toward language, but the poor schmucks just weren’t smart enough to make it—we, on the other hand, were. This belief, tacitly and probably quite unconsciously held by many who see themselves as fighting anthropomorphic biases, is in fact hopelessly anthropomorphic itself. It should therefore be quickly abandoned by them once they realize its true nature.

  A holistic protolanguage, even if it could exist, wouldn’t really be a protolanguage at all. It would be in some ways like a hyperinflated ACS, a series of reactions to specific situations; it would, however, be anomalous as an ACS, insofar as the situations it reacted to would seldom be fitness-enhancing ones. It would in fact be a hybrid, neither one thing nor the other, yet still not a viable intermediate stage between ACSs and language.

  A holistic protolanguage assumes that the parts that make up an ACS are like the parts that make up language, only finite instead of infinite and instinctive rather than learned. Make it so those parts become open-ended and learnable and you’ve taken a giant step toward language—so goes the holistic claim. But that cannot be, because ACS units and language units are totally different in form, in function, and in anything else you can think of.

  Moreover, unlike the kind of compositional protolanguage I had proposed, a holistic protolanguage would not be able to do even one of the most basic things language can do. You could not use it to ask questions or make negative statements. You could not hold a conversation in it. You could not use it to provide any kind of novel information. All you could do with it is what you do with an ACS—manipulate people.

  Wray’s objection to my kind of protolanguage is that it would have been initially crude, highly ambiguous, and not much use in manipulating people. Fine, right on the money—manipulating people wasn’t its job. For that, we already had, and we still have, a perfectly good ACS. We have screams, tears, laughter, rage faces and play faces, presentations of finger and/or rump, a plethora of body language to show how we feel and what we want others to do about it.

  An ACS is one thing, language another. If there’s any way to get from an ACS to language, it can’t consist of just blowing up the ACS until it almost bursts, then hoping the pressure will make it morph into something completely different. If the transition was ever to be accomplished, it could only come through introducing into an ACS some alien factor, something like the piece of grit that, when inserted into a humble oyster, produces a magnificent pearl.

  BABIES TO THE RESCUE?

  The singing-ape hypothesis is not the only appealing idea that tries to solve the language-origins problem by turning the meaningless into the meaningful. A variant on it has recently been proposed by Dean Falk of Florida State University, one that involves mothers and babies. Again it’s an ingenious and seemingly plausible idea, but again it founders when we start asking how meaning crept in.

  The best thing about Falk’s proposal is that, unlike the singing ape, it’s based on a real evolutionary development unique to the human line. (Remember that starting from such a development was listed in chapter 1 as one of the minimal conditions any valid theory of language evolution has to meet.) This development occurred when brains began to enlarge.

  Normally, mammals are born with their brains fully developed—everyone who’s seen a calf or a lamb born must have marveled at how soon after birth they’re able to do things like walking that take human infants a year or more to acquire. But a fully developed human brain just wouldn’t go through the birth canal. Evolution therefore favored mothers who brought their babies to term when brain enlargement was far from complete. The downside to this: babies are helpless for several months after birth, and need a mother’s fairly constant attention for several months more. Which presents a problem if Mommy simultaneously has to forage for her (and the baby’s) living. If you doubt this, try picking berries (or worse, digging up roots) while you’re holding a squirming baby.

  So how would she control baby? She’d have to put it down. Falk pours scorn on the notion that baby slings were an early invention; she says if you believe that, go into any woodland and try making one out of natural materials (and that’s without even taking into account that in the savannas there weren’t any woods to go into). There might later, when hunting of larger animals began, have been slings made out of animal hides, but the problem of the terrestrial toddler arose long before that. On the ground, back in the Pliocene, there was limitless trouble a crawler or a toddler could get into. The only thing that would work, according to Falk, would be some kind of vocal communication.

  Now no one could dispute, in this context, the value of sounds of reassurance, hushing sounds to keep baby quiet when there were predators around, shouts of warning when the little fellow was about to put a poisonous berry into its mouth, and so on. But why did these sounds have to develop into meaningful words? Wouldn’t meaningless sounds that were pleasing (for reassurance) or alarming (for warnings) and simply stayed that way have done the job equally well?

  Falk has never addressed this issue head-on. Instead, when she published her theory in the journal Behavioral and Brain Sciences, she sidestepped the problem by saying, “As is true for human babies towards the end of the first year, prosodic (and gestural) markings by mothers would have helped early hominin infants to identify the meanings of certain utterances within their vocal streams . . . Over time, words would have emerged in hominids from the prelinguistic melody and become conventionalized” (my italics).

  Behavioral and Brain Sciences is rare among scientific journals in that every article appearing in it is followed by commentaries from a couple of dozen or so scholars from the various fields that the article involves, and includes a response to those commentaries by the article’s author. In my commentary on Falk’s article, I pointed to the passage I just quoted, and remarked that it provided virtually all Falk had to say about the transition from meaningless infant-controlling noises to a meaningful word-based protolanguage. But how could infants “identify the meanings of certain utterances” unless the meanings of those utterances were known to the mothers who uttered them? And how could the mothers have learned those meanings? Well, from their mothers. And how would their mothers have learned them?

  See where I’m going? Right—infinite regress. If we take Falk literally, language has to have been there since the beginning of time. If we don’t take her literally, what could she mean? That “words would have emerged”? How? Why? They just popped out one day?

  The kind of utterances Falk is talking about have all the properties of ACS calls and none of the properties of words. They directly affect evolutionary fitness. They’re not symbolic. They show no signs of displacement. They’re meaningless outside of the contexts in which they’re uttered. There’s no way they could have brought us a tad nearer language.

  In her response to commentary, Falk responded to a couple of other points I made, but not to the one about the transition to language. There’s no way “words would have emerged” without some particular, highly specific set of circumstances that forced words to emerge—not from the singing of apes, not from the cooing of mothers, not from grooming, not from the decomposition of holistic utterances, not from any of the dozens more proposals that have been p
ut forward over the years.

  Wait, you say. What about the attempts that have been made, over the past four or five decades, to teach language to apes? Why waste time wondering how language began if in fact apes already have language—or at least are able to acquire it?

  Their so-called language abilities deserve a chapter to themselves.

  4

  CHATTING APES?

  ERASING THE RUBICON

  On Saturday, August 24, 1661, Samuel Pepys, diarist, top-level bureaucrat, and inveterate womanizer, was taken to see “the strange creature that Captain Holmes hath brought with him from Guiny.” Pepys was told it was a “great baboon,” though it was more probably a chimpanzee or gorilla, species at that date still unidentified by Europeans. And Pepys, with no background whatsoever in ethology, ecology, primatology, psychology, or linguistics, had an insight that it would take science three centuries to catch up with: “I do believe that it already understands much English, and I am of the mind it might be taught to speak or make signs” (my italics).

  About speaking he was wrong, of course. In the late 1940s, Keith and Cathy Hayes of the Yerkes National Primate Research Center, after years of trying, got their chimpanzee Viki to produce just four words: “mama,” “papa,” “cup,” and “up.” (I should make it clear from the very beginning that when I use the term “words” in connection with animal learning, I’m not in any way claiming these were words in the full human sense—“protowords” would be better, but since that’s awkward and ugly I’ll go on calling them words and ask you to remember this caveat.) An ape’s vocal control and physiology just aren’t capable of speech, although its hands, as Pepys knew (he was perfectly familiar with the sign language of the deaf) were well able to cope with the mechanical side of language. Another decade had to go by before a second couple, Allen and Beatrice Gardner of the University of Nevada, tried what should have been, but clearly wasn’t, the obvious.

  I certainly don’t want in any way to detract from the Gardners’ groundbreaking achievement, but there’s a point I’ve never seen made that needs to be made. Their experiment was low-tech—none the worse for that, but the point is, it could have been carried out at almost any time in the past, and certainly by the second half of the seventeenth century. After all, that was the heyday, the first flush, of science and the empirical method. The year before Pepys’s insight, the Royal Society of London (“for the improvement of natural knowledge”) had been founded. While Pepys was watching his ape, Robert Boyle’s book on how the human body uses oxygen was on sale in neighborhood bookstores. Two years later the first reflecting telescope was invented. A few years after that, van Leeuwenhoek discovered bacteria and the first measurement of light speed was made.

  A century rolled by. Now we’re in the Enlightenment, so-called, the era in which anything and everything could be questioned—including, naturally, the origin of language. And questioned it was: in 1769, the Royal Prussian Academy of Sciences offered a prize for the best essay on it. In the essays poured, all thirty-four of them, all right off the tops of their composers’ heads. It did not occur to any of those scholars to see what, if anything, other species could learn. Or to any of the French philosophes, Condillac, Diderot, Maupertuis, Rousseau, who also dabbled in the topic—except one, de La Mettrie, author of the notorious L’Homme Machine, who thought that a trained teacher of the deaf might be able to teach sign language to an ape. But he didn’t actually get one to try it, nor did anyone else.

  Then came the Romantics; “wolf-children” were “in,” and Jean Itard spent years trying to teach language to the Wild Boy of Aveyron. It’s ironic to think what he might have accomplished if he’d devoted the same kind of energy to teaching sign language to an ape. And mind-boggling to think how different the behavioral sciences might have become, if someone had taken Pepys’s hint three and a half centuries ago.

  So why wasn’t it done?

  The answer can probably be found in the words of Max Müller, professor of classical philology at Oxford, who responded to Darwin’s claim of the primate ancestry of humans by pontificating that “the one great barrier between the brute and man is Language [his italics]. Man speaks, and no brute has ever uttered a word. Language is the Rubicon, and no brute will dare to cross it.” Müller was, of course, a confirmed Christian, a Lutheran outraged at the suggestion that, by teaching about Hindu religion, he had somehow undermined Christianity. His words reflect an all-but-universal mind-set that saw humans as distinct from the rest of creation—one that effectively prevented anyone (save a few like the atheist de La Mettrie, who thought there was no such thing as the soul) from even thinking that Müller’s Rubicon could be crossed.

  So it wasn’t until the scientific mind had been saturated by a century of Darwinism that it became possible for the Gardners to think the unthinkable, and for one “brute,” the chimpanzee Washoe, to dare Müller’s Crossing.

  It was both a strength and a weakness of the “ape-language” experiments that so many who took part in them were driven by an agenda. Absent that agenda of not merely crossing but erasing the Rubicon—showing that no line could truly be drawn between apes and humans—the attempt might never have gotten off the ground. At the same time, the fact that the agenda was as much an article of faith as the religious conviction it aimed to overthrow drove advocates to rash and excessive claims. And this in turn led to an escalating series of controversies that, for many years, prevented any calm and objective assessment of exactly what those experiments revealed.

  “ALMOST HUMAN” OR ONE-TRICK PONY?

  On both sides, claims were carried to extremes. To some critics, signing apes (or their successors who were taught with some form of lexigrams, arbitrary and holistic representations of words that the animal could point to or touch) were, wittingly or unwittingly, conspirators in some kind of fraud. They were compared to Clever Hans, a horse in early-twentieth-century Germany who could answer mathematical and other questions posed by his trainer. So long as the answer could be given in terms of a number, Hans would provide it; he would stamp his hoof until he reached the correct number, then stop.

  Oskar Pfungst, the psychologist who investigated Hans, found that the miraculous horse was picking up subliminal cues from his trainer’s body language. As Hans approached the critical number of stamps, the trainer quite unconsciously grew tenser, then abruptly relaxed the moment that number was reached. Replace the trainer, or have him ask a question whose answer he didn’t know, and Hans would stumble, his performance falling below chance. The totally involuntary nature of these clues is shown by the fact that on occasion Pfungst himself caught himself inadvertently giving away the answer.

  A few of the earliest experiments may have been vulnerable to Clever Hans charges, but the Gardners and their successors soon developed techniques that ruled out unconscious cueing. What they did not, unfortunately, rule out were excessive claims by the experimenters that brought negative reactions not just from jealously turf-guarding linguists but from many impartial observers who would have accepted more modest assessments. Here are some examples of this Rubiconfording hubris:

  “Washoe learned a natural human language.”

  “Apes appear to be very similar to 2 to 3 year old human children learning to speak.”

  “Koko has learned to use American Sign Language—the very same sign language used by the deaf.”

  These claims are simply untrue. First of all, “natural human language” contains at least two major ingredients that even the most sophisticated ape has never employed, let alone mastered. One is grammatical structure, or syntax, the complex set of rules and principles that determines whether a string of words constitutes an acceptable sentence in a given language or is simply a string of words (the odds that it will be the second of these, for any randomly chosen string, are tens of thousands or even millions to one, by the way). The other is grammatical items, all the “at”s and “did”s and “for”s and -eds and -ings and -ses that serve as signals of tha
t grammatical structure, enabling us to parse and understand it rapidly and automatically, without a smidgen of conscious thought going into what is, let’s face it, a frighteningly complex process.

  These are not extra doodads indulged in by spoken languages but dispensed with by the deaf. Sign languages have just as rigorous a structure as spoken languages and just as many grammatical items to signpost that structure. Many of these grammatical items involve subtle body movements and facial expressions that the hearing observer, fascinated by flying fingers, simply misses.

  Indeed, the claim that apes learned American Sign Language is absurd. They were taught a handful of signs equivalent to common referential words, and that was it. Some observers have described it, pretty accurately, as “pidgin sign.” The apes never acquired the structure of ASL. Researchers would have been better off admitting that; then they would have been able to see something the apes did do that was, as you’ll see, significant and surprising.

  So even to claim that apes were “very similar” to two-to-three-year-old children is misleading. To begin with, very shortly after age two the average child transitions from short, unstructured word strings to complete, albeit still short, sentences. Soon after that, many children still several months short of their third birthday begin to produce several different types of complex sentence, using a range of prepositions, auxiliary verbs, determiners, and other grammatical items that no ape has ever produced.