In the 1970s a young academic named Herb Terrace heard about the Washoe work. He was excited by the results and wanted to replicate them, so he obtained a chimpanzee and called him Nim Chimpsky. Terrace followed the Gardners’ work closely, although he had many more people interact with Nim than had ever interacted with Washoe. Initially, it looked as if he had successfully taught Nim to use words and some syntax. But when he did a frame-by-frame video analysis, he realized that what Nim was doing was less symbolic than imitative: Nim wasn’t using language independently but instead responding to cues that Terrace or other caretakers were giving him. At the same time Terrace also did a video analysis of Washoe and Koko and concluded they, too, were being inadvertently cued by their handlers and neither thinking nor communicating. He published the results of his investigation in the journal Science in 1979.
The damage from Terrace’s findings was immediate and devastating. His article was picked up by the press, and a popular and scientific consensus quickly developed that the apes weren’t doing anything their caretakers hadn’t cued them to do. Funding for animal language research very rapidly dried up. The Gardners were effectively shut down, and one of their graduate students, Roger Fouts, took over and was for a long time only able to maintain but not expand the Washoe project.
From that point on, said Lyn, it became very hard to get any animal language data published. After the Nim Chimpsky publication, Lou Herman started his studies with the dolphins Akeakamai and Phoenix, using an artificial language and focusing on comprehension (his funding for the project was secured before 1979). The fact that he was concerned with comprehension, rather than production of language, was probably what saved his work, said Lyn. People found it easier to consider the possibility of animals’ understanding versus producing language. Still, Herman didn’t publish his first paper until 1984. As soon as the paper came out, he was criticized intensely for using linguistic terms like “sentence,” and “noun,” and “verb” to describe what the dolphins were doing. That response was unjustified, said Lyn. In fact, she said, Herman has the best data on syntax for any animal, anywhere. Akeakamai and Phoenix have mastered a complex grammatical system. If Herman gives the dolphins nongrammatical sentences, they will either refuse them or make grammatical sentences out of them.
A year after Terrace’s Science article was published, Martin Gardner reviewed a number of books about animal language training in the New York Review of Books. He began by tracing a direct line from crackpot claims that dolphins communicated through ESP to ape language research. His first pass at evaluating Penny Patterson’s work with Koko and the attention it received had more to do with Patterson herself than with her science. “It is not hard to understand why Penny—young, pretty, with long blond hair—has received such enormous publicity,” he wrote. “What could be more dramatic than color photographs of Beauty and the Beast, heads together, raptly chattering to one another?”
It is hard to understand how comments like Gardner’s become part of the debate: the same would never have happened had the scholar in question been, say, Chomsky, who has likely never had his physical appearance assessed in reference to his work and its public appeal or been called “Noam” in similar circumstances.
Apes might have a “feeble talent” for putting together signs in meaningful ways, but it was more likely, Gardner concluded, that ape language research amounted to little more than an unconscious collusion between a cooperative animal and a hopeful human. As he wrote: “There is no solid evidence that an ape has ever invented a composite sign by understanding its parts. In the course of several years an ape will put together signs in thousands of random ways. It would be surprising if it did not frequently hit on happy combinations that would elicit an immediate Clever Hans response.” (Clever Hans was a famous horse who could allegedly perform mathematical computation. He would indicate the answer to a problem by pawing at the ground the correct number of times. A 1907 study showed that Hans’s owner gave him subtle and unconscious cues when to stop pawing at the ground.)4
Terrace did make some important contributions, explained Lyn, by pointing out that there had been no scientific controls in the studies assessing the apes’ syntactic ability. Mostly, the claims for syntax were based on naturalistic observations and had not been rigorously tested. But because Terrace found instances of cuing, the scientific community and the public decided that all of the behavior was cued. There were, in fact, numerous examples of solid, double-blind experiments, such as one where Washoe was placed alone in a room. A camera was trained on her, and pictures were flashed up on a screen before her. The chimpanzee made the signs for every object in the pictures, and because she was by herself, cuing was impossible.
Luckily for Savage-Rumbaugh, her funding had been renewed for five years just before the Terrace article appeared. She spent those years producing valuable findings. For example, Kanzi and Panbanisha have spent time with other apes in different experimental situations. For a while, they were raised with another bonobo, Tamuli. But while Kanzi and Panbanisha were exposed to language from the time they were just a few weeks old, Tamuli’s exposure began much later in life. She was initially reared by her mother, but at three and a half years of age she was allowed to accompany Kanzi and Panbanisha in their daily activities, like taking trips to the forest. Kanzi and Panbanisha’s human caretakers also spoke to Tamuli while pointing at the picture keyboard and describing their daily activities.
Tamuli never developed language skills comparable to those of the other apes. In this respect she is like human children who, for whatever reason (for example, undiagnosed deafness or abuse), are not exposed to language at an early age. There is a crucial learning period for humans when they must be exposed to language. Even if they are neurologically normal, they will never fully acquire language if its foundations are not laid in this early period of brain development. Genie, the most famous of these cases, was kept locked in a room, denied normal human communication, and never taught language, and by the time she was rescued, she was unable to acquire much more than basic language skills. Her experience shows that if you are denied language, you don’t spontaneously produce it. Since Genie’s case was studied, it has become fairly well established that language is not innate in the same way as, say, our instinct to breathe or cry. Tamuli’s experience suggests that apes have a similar window of opportunity.
Even though Tamuli could not relate at the sophisticated level that Kanzi and Panbanisha did, she at least seemed to understand that the keyboard was intended for communication. While the apes often tried to use it to relay messages, Tamuli’s usage was a bit like that of a young child banging away on a piano or a keyboard. She made no sense.
One thing that Tamuli lacked was the ability to recognize that her interlocutors had separate minds and that communication with them could alter their perceptions. Kanzi and Panbanisha, on the other hand, seemed to have acquired a theory of mind along with language. When Panbanisha saw her trainer remove candy from a box, replace it with a bug, and then give the box—supposedly still with candy—to Kanzi, she called her “bad.” The chimp demonstrated that she could understand what was going on in her trainer’s mind independent of the reality and apply language to the situation. She herself scared Kanzi when she used language to tell him there was a snake nearby, when in fact there was no snake. Panbanisha used language to manipulate the contents of Kanzi’s mind, just as her trainer had manipulated the contents of the box.
The ape experiments indicate how memory is a vital component of language use, even at this rudimentary level. A chimpanzee, Panpanzee, who was raised with Kanzi and Panbanisha, would sometimes make language mistakes that demonstrated a limited memory, such as when she was asked to put a sweet potato in the microwave. Typically in an experiment like this, an object, like the potato, would be made readily available for Panpanzee, placed right before her eyes. But in this example, the chimpanzee had to retrieve one from the refrigerator in order to complete the request. Th
is she did. But instead of putting the vegetable in the microwave, she took it to the sink and proceeded to wash it. Somewhere between the retrieval process and the end task, the request became scrambled for the chimp. In similar situations, Panpanzee’s incorrect response suggested she was falling back on her knowledge of routines, rather than correctly remembering a novel request (something people occasionally find themselves doing as well).
Sometimes language mistakes can be as useful as correct responses. Eliciting errors in human speech is one of the main methods that psycholinguists use to expose the mental strategies that underpin language use. Spoonerisms, for example, aren’t just sound swaps: “pea tot” (teapot), “dood gog” (good dog), and “band hag” (handbag) suggest that speech is not entirely spontaneous. If a speaker accidentally begins a word with the first sound of the next word, he must be planning what he is about to say, even if he is not aware of it.
Lyn analyzed eleven years’ worth of Kanzi’s and Panbanisha’s language error data and found that when the apes accidentally pressed one keyboard picture instead of another, or when they misunderstood a spoken word, their errors usually revealed an underlying connection between the intended word and the mistaken one. Just like humans, the apes made category substitutions, like mistaking colors, such as red for black. They made word association errors, confusing the names of locations with items that were found in those locations. And they made phonological (sound) errors, like using a word because it rhymed with the intended word.
Lyn and colleagues found that Panbanisha and Panpanzee have more symbol ordering rules in common with each other than with their caretaker. It’s possible, even probable, that the last common ancestor between these apes and humans had the ability to understand meaning-based ordering strategies. Lyn also found that these apes have a gesture-last rule: they always touch the lexigram, and then gesture in the real world.
Bonobos acquire language up to the level of human children. For example, they can understand sentences that contain one verb and a three-noun phrase (“Will you carry the M&M’s to the middle test room?”), but they have trouble with conjoined sentences that require two separate actions (“Bring me the ball and the orange”). They do not speak English words, though they attempt to do so. Their short-term memory seems to be only half the capacity of human children’s, so they are not as good at imitating a series of utterances without a lot of repetition.5 The more complicated syntax gets, the more trouble they have with it.
The ape language research led Savage-Rumbaugh and her colleagues to conclude that language consists of “a large number of component parts and interacting functions.”6 Even though their work has not had the impact of Chomsky’s, most researchers in language evolution would today think about language in these terms.
What’s most striking about the older criticism of ape language research is its basic attitude, which is more motivated to discredit than evaluate. In much ALR commentary, there is a strong sense that the critics have already made up their minds before arguing or offering reasons why ape language couldn’t work. There are claims of falsifying data and even of people being out to get each other. In the 1980s the debate was rarely conducted without tones of disdain and contempt.
Even now, scholars who work with animal language are often characterized as daft idealists or outright frauds, believing that beneath the fur or behind the beak are creatures with souls. Yet if you speak to these researchers, you won’t find anyone downplaying the enormous differences between humans and other animals, despite the fact that they happen to be interested in the commonalities.
One legacy of the Terrace paper has been an ongoing difficulty getting funding for this kind of work. Researchers often have to go outside the typical funding bodies of academia to keep their studies going, turning to special interest groups and private individuals. The promotional literature for the Koko research mentions visits from Sting and Robin Williams, for example, a gambit that gives animal language research a weird profile. Such marketing gives the impression that it is not solid, straightforward science.
Still, the basic tenor of the commentary has begun to shift over the years. Critics used to dismiss the research by saying, “All that the animals have is a few words, and they don’t have any syntax whatsoever.” Now the fact that apes can acquire words is treated as an interesting phenomenon.
Frans de Waal, a professor of primate behavior at Emory University and the author of The Ape and the Sushi Master, says:
I think the trend is clearly towards poking holes in the wall that exists between us and animals, and increasingly people embrace the comparison, so to speak. In the 1970s, when I had to give a lecture on chimpanzees, some people would say, “How can you use the term ‘reconciliation’?” They would have strong objections. Or let’s say it was about sex differences, and they would say, “How can you compare chimpanzees and humans?” Because obviously, we are cultural beings and we can change our behavior.
When I give lectures on these topics today, that never happens anymore. It’s because there’s a gene on the cover of Time or Newsweek almost every week, a gene for this or a gene for that, so people are getting very used to the idea that genes add something to behavior. So the climate is totally different, and there’s a much greater openness to seeing us as animals, as Darwin always wanted and as many other people wanted.
I was recently invited to give a talk for business ethicists. Now, business ethicists are basically philosophers who teach at business schools. Even there, there is an enormous openness for these comparisons, whereas I’m sure twenty years ago they would not want to even touch a monkey. So I think the trend is clearly towards more comparisons. More comparisons doesn’t necessarily mean that we fully accept the similarities. Usually they’ll want to keep something like, “This is typically human” or “This is unique to humans”—they want to keep this to some degree.
One of the most important contributions of ape language research is its challenge to the traditional idea that other animals have a fixed mental bag of tricks, and humans are different because we have language and that makes us mentally flexible. If that were the case, Kanzi would have been unable to learn the language skills he has. Clearly, these apes who have the rudiments of language can also be flexible and creative with their communication.
Ape language research, and Kanzi in particular, opened one fascinating window into the problems of language evolution. Steven Pinker and Paul Bloom opened another in 1990 when they published a paper in which they sidestepped the question of how much animal language training can teach us about language evolution and instead argued directly that not only could language evolution be studied but it should be studied. The two scholars—one a rising academic star, the other a graduate student with a brilliant idea—inflamed hearts and minds because their proposal was clever, innovative, and engaging. And even though they weren’t the first to propose that language evolution was a valid topic of inquiry, their paper ignited a small blaze that quickly grew and spread.
3. Steven Pinker and Paul Bloom
In 1989 Paul Bloom, a twenty-five-year-old graduate student in the psychology department at MIT, was doing research in child language development. He was interested in word learning in young children, which had nothing to do with evolution, but he was increasingly bothered by the general agreement that language could not have evolved.
“Two things happened at once,” he recalled in an interview.
One was that Leda Cosmides, who is now at the University of California, Santa Barbara, came to give a talk at MIT. She’s a prominent evolutionary psychologist, and she started talking about the mind and language from an adaptive point of view. When we met later, I said,
“This is ridiculous!” I responded to her with the Stephen Jay Gould line, which I had totally been persuaded by years before. There was no reason to favor an adaptionist account of language (as opposed to the view that it was an evolutionary accident).
She was very civil and intelligent, and s
he said, “No, no, you’re mistaken.” And she convinced me that it made sense to apply an evolutionary analysis to mental life. Some things may be artifacts of biology, but there are good reasons to believe that something as rich and as complicated as language could have evolved by natural selection.
It was one of the rare case where an academic changes his mind. After thinking about it for a while, I realized that it made sense.
And then, at the same time, Massimo Piatelli-Palmarini, a colleague and friend of mine in the Department of Linguistics and Philosophy, published an article in Cognition on the evolution of cognition and language. His article presented in this very sharp, cogent fashion the Chomskyan view on evolution—basically he said that there was very little interesting to make of the connection between natural selection and cognition and that language has features that simply cannot be explained in terms of adaptation. I strongly disagreed with it.
The First Word: The Search for the Origins of Language Page 6