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

by Bickerton, Derek

  Ah, say those who believe apes have, or can acquire, language. You’re talking about production. We’re talking about comprehension. Apes can comprehend language at the level of at least a two-year-old, and we can prove it. And in fact, they claim, comprehension is much harder than production. After all, in comprehension you have to figure out what the other person means. Whereas in production, you know what you mean, you just have to put it into words.

  To a linguist, this is a bizarre position, the very reverse of the truth. The idea that we decide to say something and then dress it in words is one of those ideas, like the sun going around the earth, that seem obvious and irrefutable to the naive, untrained mind, but bear no relation to what actually happens in the real world. Even if the naive view were right, you’d still need the nuts-and-bolts, highly specific and detailed knowledge of how to put sentences together, in whatever language you were trying to speak, and how to do this smoothly and swiftly so your audience didn’t get bored and wander away before you’d finished. In comprehension, on the other hand, you don’t need to know how to put sentences together. If you know what enough of the words mean, and you know where you are and what’s happening, and you can apply your common sense and practical knowledge of the world, you don’t need syntax to figure out what the other person means.

  If for instance someone says, “Go to the refrigerator and get an orange,” you don’t have to know that this consists of two coordinate clauses, that “to” introduces a locative phrase, or that “orange” is the direct object of “get.” These are all things you’d have to know, in some sense, at some quite unconscious level—not the names for them, but what those names represent—if you were going to produce the sentence. To understand it, all you need are the meanings of four words: “go,” “refrigerator,” “get,” and “orange.” “Go” tells you about moving to another location—“refrigerator”—and “get” means you have to obtain something—“an orange.”

  So when Sue Savage-Rumbaugh pitted her bonobo, Kanzi, against a female toddler, Alia, on a series of commands like the refrigerator sentence (but varying the structure and content, naturally), Kanzi scored correctly 72 percent of the time against Alia’s 66 percent. But did this really show that they were at a similar level of development? Kanzi was age eight and had had years of experience in hearing and executing instructions like these. Whether Alia had any experience at all is unclear from published accounts, but since testing started when she was barely eighteen months old, she could hardly have had more than a few weeks, at most. Add to that the fact, reluctantly admitted by the experimenters, that Alia’s MLU (mean length of utterance, measured by the number of meaningful units—words and affixes—in a sentence) went from 1.91 to 3.19 over the six months of testing, while Kanzi’s remained stubbornly stuck on 1.5, and you can see this experiment, impressive though it might seem, hardly shows parity between ape and child.

  But statistics and formal measures don’t really get to the heart of the matter. The real difference lies in content—not how apes and children communicate, but what they communicate about.

  Ape conversation is ego-centered. All that any ape, including Kanzi, the Einstein of apes, ever talks about are things like where they want to go, what they want (or want you) to do, or what they’d like to eat. General topics are out. Objective information about the environment or events in it is never exchanged. And this, after all, is exactly what you’d expect from an animal without a natural language but with a fully functioning ACS. The things apes talk about, their own wants, needs, and desires, and the manipulative way in which these are expressed are, as we saw in earlier chapters, just the things that an ACS deals with, that an ACS is specially designed to deal with.

  Contrast with this the behavior of Seth, the child whose brief flirtation with serial verbs I described in Bastard Tongues. When Seth was around the same age as Alia at the start of her tests—eighteen months—his father, one of my students, made a recording of a conversation he was having with a couple of friends, which Seth, still in the one-word stage, insisted on interrupting, even though no one was taking the slightest notice of him. Unfortunately I no longer have the transcript, but my memory of it is vivid enough to give you the flavor of a brief extract:

  ADULT: Blah blah blah blah blah.

  SETH: Telephone.

  ADULT: Blah blah blah. Blah blah blah blah.

  SETH: Fan.

  ADULT: Blah blah blah blah blah blah blah.

  SETH: Doggie.

  Seth was systematically naming all the objects in the room. Obviously this was manipulative and self-serving—he wanted to join in the grown-ups’ conversation. What’s striking is the way he chose to do this: by showing the grown-ups all the things he knew and could recognize and had names for.

  And that, as we saw in the last chapter, captures one of the most essential differences between the uses of language and the uses of an ACS. In using an ACS, manipulation is uppermost and information, if any, is incidental; in using language, information is inescapable—the mere fact of using language automatically transfers factual information from one individual to another. Seth surely was trying to break into the conversation. How he stated that need, though, was not through some ego-centered demand, as an ape would have done, but by informing everyone of the names of things—thereby showing that he now knew the human code, and was therefore entitled to join in the conversation.

  SO WHAT CAN APES REALLY DO?

  If the ape-language researchers hadn’t been seduced by the excitement of dramatic claims and high-profile media impact, they might have been quicker to realize the full significance of what the apes were actually doing.

  Most or all of the apes seem to have done at least three things, hardly mentioned in the literature, that were highly significant for any understanding of how language could have begun. I’ll deal with these in ascending order of importance.

  The first significant thing apes did was to distinguish between words and proper names.

  To us, the distinction is self-evident. If I introduce you to a round-faced, short-bearded, spectacle-wearing individual named Rudolph, you don’t start calling every round-faced, short-bearded, spectacle-wearing individual Rudolph. Similarly, if I show you a new kind of fruit and tell you it’s a cherimoya, you don’t, on seeing another such fruit in the market, ask the stallholder, “What do you call that one?” But why should the distinction between things like “Rudolph” and things like “cherimoya” be self-evident to an alingual ape?

  Surprisingly enough, it apparently was. I say “apparently,” because I don’t remember seeing anywhere in the ape-language literature any specific discussion of this point. However, in a fairly extensive reading of that literature, I’ve seen no mention of any case where an ape called one of its trainers by another trainer’s name, or where, on having been taught a sign for, say, “banana,” the ape failed to apply it, or showed puzzlement when a banana different from the original training banana appeared. In other words, they seem intuitively to have grasped the difference between words for individuals and words for categories.

  I don’t know why this is so—again, no one studied it—but I would surmise that it comes from being a social species. You meet, and have to deal with, other members of your social group on an individual basis. You behave toward each of them in a different way. But you behave to every banana in the same way. In other words, the distinction is one that all social animals, simply by being social, get for free.

  The second significant thing they did was to spontaneously put signs together.

  As we’ll see in a moment, it took them a long time to grasp what signs were about, even though they were being trained intensively. But they began to put signs together, to make messages, without any explicit training at all. They got some modeling, because their trainers would address them with sequences of signs, but not a drop of explicit instruction.

  Granted, the combinations they made were seldom of more than two signs, which accounts for Kanzi�
�s 1.5 MLU. But, lacking anything in the way of syntax, what would you expect? Granted, some of these messages were simply “X and Y”—two quite disconnected signs that happened to be given in sequence. But enough of them took the form “X[Y]”—true predications, things like “Roger tickle,” “trailer go,” “no balloon”—for these not to have been mere accidents.

  Combining things, you’ll remember, is something that animals using an ACS simply cannot do. There’s no precedent for it in the animal world. How did they do it? I’ll postpone that until we come to where our ancestors faced the same problem.

  The third and most important of the significant things apes did is something I can only describe as “getting it.”

  The idea that an arbitrary symbol—be it a spoken word, a manual sign, or, as was increasingly the case, a pictorial symbol on a screen that the ape had to touch—can stand in for something in the real world is as plain as a pikestaff to you and me, but not to any member of another species. The apes took a long time to grasp it. Washoe took three months to acquire her first sign. Lana, an ape trained by Duane Rumbaugh, then at the Yerkes National Primate Center, took 1,600 trials to learn symbols for “banana slices” and “M&Ms.” After that, however, they took off. Washoe was soon down to ten trials or less. Lana “succeeded in naming ball on its first presentation.” It was as if a light-bulb had suddenly gone on in their heads: “So that’s what these dumb humans are trying to get me to do!”

  Well, a lightbulb going on in their heads is probably pretty close to what actually happened.

  When we’re talking about behavior, we tend to focus on what’s going on in the external world, rather than on what must be happening inside the head of the behaver. If we think of anything in there, we usually speak of the behaver’s “mind,” and we probably think of that as some kind of cerebral screen, on which, as on the wall of Plato’s cave, outside events are shadowed. Due perhaps to a hangover from centuries of dualism, we tend to ignore or downplay purely physical events in the head. But through the rest of this book, we must increasingly bear in mind the never-ending interaction between the external-physical and the internal-physical. What happens in the outside world triggers electrochemical events in the brain—sends messages racing down axons, enzymes leaping across synapses—but it doesn’t only do that. It changes the way the brain is wired. And the long delay between the initial presentation of signs to apes and their first grasp of meaning follows directly from the following axiom:

  “Neurons that fire together, wire together.”

  This is Hebb’s Rule, a pithier version, known to all first-year neurology students, of a more nuanced but canonical statement by Donald Hebb, one of the pioneers of cognitive science. The brain’s plasticity is hard to exaggerate. It can’t do major architectural changes, but it can remodel most of its many rooms while it keeps the house running without the slightest glitch. Exactly what happened in Washoe’s and Lana’s brains, what hitherto weak or nonexistent links were forged or strengthened, we simply don’t know. Nobody ever asked. Only years of neurological research could answer that question, and the ape researchers weren’t even neurologists.

  But we know in essence what must have happened.

  The presentation of novel signs to apes, coupled with the presentation of physical objects, caused certain neurons to fire simultaneously that had never fired simultaneously before (this is true of any new experience). These neurons were those in the visual cortex that directly responded to the first sign and those (probably in either the motor or visual cortex) that represented the sign or written symbol that the apes were learning. (An interesting line of research that I don’t think has yet been followed would be to confirm what one might expect: that apes stored representations of manual signs in the motor cortex but those of pictorial symbols in the visual cortex.)

  It took repeated exposures to the signs, hundreds and thousands of them, for those same neurons to keep on firing together, each firing spreading and strengthening new connections until they were strong and far-reaching enough for enlightenment to dawn. That accounts for the long initial delay between first presentation and “learning.” But once the first few signs had been learned, and hence the first few links had been established between the brain areas involved, a pattern was set up that, when new signs were taught, could be quickly repeated for each new sign. Only some development of this kind could account for the hundred-times-faster speed with which, after the first handful, new signs and meanings were connected.

  Two things only could have triggered the growth of the neural network that made it possible to connect arbitrary signals with things in the outside world. One of them, the one that worked for apes, was a deliberate act of intervention by another species: us. The other, the one that worked for our remote ancestors, was factor X—the factor this book is looking for.

  But if at least a few of the many features of language can be taught to apes—if, in other words, they can be taught some kind of protolanguage, maybe the kind our ancestors developed—how is it that they have never used that capacity for themselves, in the wild?

  BACK TO THE WILD

  First, we should deal with the claim “Well, they do use it in the wild—we just haven’t been smart enough to understand how they do it.”

  This claim had a lot more going for it in the early years of primate ethology. But today, apes of several species have been studied in their natural surroundings, as well as in some zoos and research centers where those surroundings were replicated to the extent they could be. They have been studied by many acute and highly motivated observers for nearly half a century. The behaviors of these species have been described and discussed and analyzed over and over again. Yet not one researcher has ever come up with any behavior that seemed remotely languagelike. And with every year that passes, the likelihood that one ever will gets smaller and smaller.

  You can’t prove a negative, but if you’re ever going to get anywhere in science, you just have to ignore possibilities, no matter how tempting, for which there is absolutely zero evidence.

  So let’s look at the opposite argument: “If they really have these capacities, how come they never use them in the wild?” (Implied: if they don’t use them there, they can’t have them—they must be merely artifacts of the experiments.)

  Remarks like this show a serious misunderstanding of how biology works. You can make such remarks only if you believe that every potentiality existing in the genes must be expressed in terms of behavior. A moment’s thought shows that such a belief would automatically rule out any kind of innovation. Animals would not only have to play exclusively with the deck nature dealt them—they would be compelled, like helpless automatons, to play every card in that deck. Only a mutation, a new card in their genetic deck, could bring about innovation.

  There are, of course, simple creatures, programmed down to the wire, of which the foregoing might be true. Once you’re past nematodes, it’s a different ball game. More complex animals can learn from experience, and they couldn’t do this if genetic determinism was total and absolute. When environments change, some members of a species often survive, and they can do this only by doing things that their genetic equipment allows, but that they had never done before, because they’d never had to do such things. Around every animal there’s an envelope of potentiality, of things they’re not specifically programmed to do but that they can do somehow, if they have to in order to stay alive.

  In the next chapter I’ll deal at much greater length with the complex relationships between genes, behavior, and environment. For now, we need only note that Washoe and Lana and Kanzi and all the other trained apes underwent a radical change in their environment, a change that, like many other environmental changes, came about through a new kind of contact with a different species—in this case, us. And like many animals before them, they successfully adapted to that change, by producing the kind of behavior—protolinguistic behavior—that seemed to be required in order to survive and prosper (read,
get banana slices and M&Ms).

  Which of course is very far from the behaviorist creed—gospel in the middle of the last century—that you could condition any animal to perform any kind of behavior. Eörs Szathmáry, John Maynard Smith’s ex-student now with the Hungarian think tank Collegium Budapest, made a nice distinction in a recent paper between changes that are variation-limited and changes that are selection-limited. To say that some potential change is variation-limited means that the kind and/or degree of genetic variation necessary to bring that change about simply isn’t present in the species concerned, and isn’t likely to develop within any reasonable time frame. No matter how strong a selective pressure exists, a variation-limited change simply can’t happen.

  To say that a change is selection-limited, however, means that the genetic bits and pieces sufficient to bring that change about are all present (or can be very easily attained) and all that is required is a strong-enough selective pressure.

  For most species, changes that might lead in the direction of language are variation-limited; the stuff that would build those changes just isn’t there. For probably all four varieties of great ape—chimpanzee, bonobo, gorilla, orangutan, and hence necessarily for our own ancestors—such changes are selection-limited, awaiting only the right kind of pressure to bring them to birth.

  Which immediately raises the question, are the great apes unique? Or are there other species for which the first steps toward language are only selection-limited?

  WHO’S READY FOR PROTOLANGUAGE?

  Sea lions wouldn’t be on many people’s list of the most intelligent species. But from the late sixties on, Ron Schusterman, working out of California State and later the University of California, Santa Cruz, showed that while sea lions weren’t adapted for any productive skills, they could master most if not all of the receptive skills that apes could. Simultaneously, Lou Herman at my own institution, the University of Hawaii, was performing similar experiments with dolphins. And, perhaps most amazing of all, Irene Pepperberg, first at the University of Arizona and now at Brandeis, trained an African gray parrot, Alex, to do all the things that sea lions, dolphins, and apes had been trained to do.