By the end of the nine-month test period, both Kanzi and Alia had demonstrated a well-developed ability to comprehend all types (and subtypes) of sentences, with Kanzi scoring just a little ahead. Overall, Kanzi correctly answered 74 percent of the sentences, while Alia’s figure was 65 percent. Often, when errors were made they were semantic or through inattention, not a misunderstanding of the structure of the sentence. For instance, when I asked Kanzi to “Pour the milk on the cereal,” he poured the milk on the mushrooms. He performed the correct action according to the structure of the sentence, but with the wrong object. Similarly, Alia once put peaches into yogurt rather than into the tomatoes, as she had been asked.
On another occasion, Kanzi clearly mixed up words, and thought he was doing as I had asked, which was to “Put the paint in the potty.” He promptly picked up some clay (a similar play object to paint), and put it in the potty. I said, “What about the paint?” Kanzi put more clay in the potty. I said, “Thank you,” but “now put the paint in the potty.” Kanzi clearly thought me a little dumb, and so brought me the potty and placed it right in front of my face so that I could see that he had done what I was so persistently asking.
One difficulty we had not foreseen, but which proved illuminating, concerned requests to retrieve an object from another location, such as, “Go to the group room and get the ball.” If there was a ball where Kanzi was sitting when he heard the request, he often glanced at it, glanced toward the group room, perhaps touched the ball, and, 50 percent of the time, handed it to the experimenter. We thought of another way of asking the question, which linguistically is more complex: “Get the ball that’s in the group room.”
Structures such as “that’s in the” are known by linguists as embedded phrases and are thought to be uniquely reflective of human thought in their recursive structure. In such sentences, one part of the sentence refers back to another. In order to understand the meaning, one has to know that one word refers to a specific word that occurred earlier in the sentence, and that the second word, in some way, changes the meaning of the first. For example, in the sentence, “Get the ball that’s in the group room,” the words “group room” refer back to the word “ball” and specify a particular sort of ball. Since parents have not been observed actively explaining how “recursion” works to children, yet children understand such sentences, linguists assume that an innate grammatical device permits them to decode such embedded references.
For Kanzi, the embedded phrase helped clarify the request, not confuse him. When he heard such sentences he tended to set off in the direction of the required location, with rarely a glance at a decoy object in front of him. He scored 77 percent correct answers with such requests, compared with 52 percent for Alia.
Oddly, one of Kanzi’s greatest problems was with sentences that were grammatically the most simple, in which he was asked to do something with more than one object. For instance, if I said, “Give Sue the hat and the potato,” he would readily give me the hat, but not the potato. He would then give me the potato if I reminded him. Sometimes Kanzi forgot the first object, other times it was the second one. It seemed that as soon as he focused on one, the other was forgotten, suggesting that the problem was one of short-term memory. There was no inherent relationship between, in this case, the hat and the potato that helped him respond correctly. If I said, “Put the hat on the potato,” he had no difficulty remembering both objects, though he might simply place them side by side rather than putting the hat on the potato. Such errors revealed the way in which sentence structure aided comprehension, both for Kanzi and Alia.
Some things simply baffled Kanzi, however. For instance, he was completely at a loss with “Can you put the Coke can in the trash can?” Not only did the multiple use of “can” confuse him, but he could not come to terms with the concept of putting trash in special places. Bonobos leave trash where they make it. To put it in, say, a backpack to take back to the lab, where we would put it in another container, the trash can, was quite beyond bonobo comprehension—literally. The important thing to note about such an error is not that Kanzi could not understand the words or the structure of the sentence, but rather that he had no means of determining the set of things it was that made something deserve the label “trash.” The only common element lay in the fact that they were all things we did not want. Many of them were things that Kanzi was interested in, however. Similarly, things we were often interested in, such as computer diskettes, seemed little more than trash to Kanzi, as he had no use for them.
The test demonstrated what I had believed to be correct: namely, that Kanzi’s comprehension went far beyond single words and simple phrases. In fact, I realized that I had underestimated his abilities rather than overestimated them. In a formal description of the project, my colleagues and I wrote: “These data support the view that both Kanzi and Alia were sensitive to word order as well as to the semantic and syntactic cues that signaled when to ignore word order and when to attend to it… . The similarity between the two subjects is all the more remarkable in that, while able to comprehend sentences, neither subject was as yet a fluent speaker.”8
No one who sees Kanzi under these test conditions fails to be impressed, but I have found some people’s reactions curious. For instance, during one presentation at a meeting of the American Association for the Advancement of Science, in which I presented a videotape of the comprehension test just described, one person asked how it was that I could suggest Kanzi understood language, when he did precisely what I asked him to do. Since the main point of the presentation had been to demonstrate that Kanzi is able to do what he is asked to, I was at a loss as to how to answer this query. It seemed that, from the questioner’s point of view, comprehension had little, if anything, to do with language. From my perspective, comprehension was the essence of language, and was far more difficult to explain and to achieve than production. Comprehension demands an active intellectual process of listening to another party while trying to figure out, from a short burst of sounds, the other’s meaning and intent—both of which are always imperfectly conveyed. Production, by contrast, is simple. We know what we think and what we wish to mean. Speech production is simply a matter of mechanically transforming our thoughts into speech sounds. We don’t have to figure out “what it is we mean,” only how to say it. By contrast, when we listen to someone else, we not only have to determine what that other person is saying, but also what he or she means by what is said, without the insider’s knowledge that the speaker has.
Bonobos are more vocal than common chimpanzees, and Kanzi is no exception. In fact, when he was still quite young, I began to notice that he was exceptionally vocal, even for a bonobo. At times he even seemed to be trying to imitate some vowel sounds. For instance, when I gave him peanuts, I would say, “Kanzi, would you like peanuts?” And Kanzi would vocalize, “e-uh,” a two-step sound, like “peanut,” without the consonant. Similarly, with melons, he vocalized “eh-uhn,” again a two-step sound, like melon. I made the same sounds back to Kanzi, to encourage him. I talked about my observations with my colleagues, documented what I heard, and tried to get them to do as I did. It proved to be a difficult exercise, partly because some people had difficulty hearing the sounds, and we all sounded rather different when we made them. In short, we could make sounds that were within Kanzi’s range, but we really did not know how to construct a communication system that we could all readily understand at that time.
As I was carefully studying tapes of Kanzi and Sherman and Austin, I was impressed by the way Kanzi vocalized while communicating in other modes, such as using the keyboard and gestures. Symbols, gestures, and vocalizations seemed to be integrated as a communicative package. I decided I needed to look more carefully at the nature of his vocalizations.
With my colleague William Hopkins, I recorded Kanzi’s vocalizations and compared them with those of four bonobos at the Yerkes Regional Primate Research Center, Lorel, Laura, Linda, and Bosondjo, none of whom had been langu
age trained. Using spectrographic sound analysis (which is simply a way of transforming the wave form of sound into a visual picture), we were able to identify fourteen different sounds (strictly, groups of sounds) from the body of data we collected. Ten of them were common to all five chimps in the study, while four were unique to Kanzi. The four groups of vocalizations are “Ennn,” “ii-angh,” “WHAI,” and “Unnn.” Most bonobo sounds tend to slide from one syllable to the next, but some of Kanzi’s unique sounds displayed clear shifts, as in the “ii-angh group.
My impression had therefore been correct: Not only was Kanzi vocalizing more than other chimps, but he was also making novel sounds. And from our videotapes of the sound recording sessions, we could see that the novel vocalizations were nearly always a response to a question by his human companion, a response to a comment by a human companion, or a vocal request by Kanzi. What was he saying?
Although the distinction was not always sharp, we were able to discern some pattern in the use of the sounds. For instance, the “Ennn” group, which was the most frequently uttered sound, seemed to be part of a request, often accompanying a gesture. One example on tape was when Kanzi pointed to food in a nearby cooler, but his human companion failed to notice. Kanzi repeated the gesture, this time adding an urgent “Ennn.” Kanzi used the “ii-angh” group most often in response to questions, and particularly questions that ended with a two-syllable word, such as peanut. The “WHAI” vocalization was used most frequently after a question, such as, “Do you want to hide?” or “Do you still want to get your ball?” The last of the four, the “Unnn” group, followed a variety of human queries, such as, “I was going to put some Kool-Aid in your bowl, do you want some?”
Kanzi’s novel vocal repertoire challenges the widely held belief that nonhuman primate vocalization is hardwired and cannot be significantly modified. Not only are Kanzi’s sounds unique among bonobos, but they are also distinctly unlike bonobo sounds. The fact that he learned this repertoire in a language-rich environment while acquiring an extensive comprehension of spoken English prompts the speculation that he is trying to imitate human speech, or at least the inflection in such speech. As he has grown older, Kanzi has gained greater control over his vocal tract and appears to continue to attempt to imitate speechlike sounds.
While I was explaining the vocalization work to a reporter for a science magazine some while ago, the reporter said to me, “They’ll never believe you.” To which I replied, “Science is not about doing things people will believe. It must explore the phenomena that are out there, believable or not.” Had I been guided in my work only by what was thought “believable,” I would not have learned that Kanzi could acquire language spontaneously, as humans do, develop extensive comprehension, as humans do, and invent his own grammatical rules, as human ancestors once did. Who knows, maybe one day Kanzi will throw away his keyboard and say, “I’m fed up with Herbert Terrace claiming I don’t have language.” Personally, however, I don’t think Kanzi cares at all about that. I think Kanzi would say, “I’d like to meet a good-looking female bonobo, preferably one that has learned to speak.”
As soon as we had noticed that Kanzi, at two and a half years of age, had spontaneously acquired language skills, and continued to develop them, two obvious questions were raised. First, was Kanzi unique among bonobos? And second, were bonobos significantly more endowed linguistically than common chimps? My experience with Sherman and Austin seemed to indicate that common chimps were different. Eight years of language use around Sherman and Austin had not elicited any significant comprehension of spoken words.
We addressed the first question very early in the program, by raising Kanzi’s half sister Mulika in a language-rich, social environment, exposing her to the full range of lexigrams right from the beginning. She began to use lexigrams spontaneously by the age of one year, much earlier than Kanzi had. So we knew he wasn’t unique.
I formed the strong impression that bonobos were different from common chimpanzees, and stated so in several scientific papers. I should have waited until we conducted the obvious test, which was to raise a bonobo and a common chimp in the same language-rich environment. This we did with Kanzi’s second half sister, Panbanisha, and a common chimp named Parizee. At first it looked as if my initial conclusion had been correct, because Panbanisha began using lexigrams within a year, while Panzee did not. By the time she was eighteen months old, however, Panzee began using symbols, and went through a learning spurt. She never fully matched Panbanisha’s skills, though, either in production or comprehension.
The dual lesson we learned from the project with Kanzi, Mulika, Panbanisha, and Panzee, therefore, was that chimpanzees can acquire language skills spontaneously, through social exposure to a language-rich environment, as human children do. And, again like humans, early exposure is critical. Chimpanzees do travel down the language road given the appropriate rearing environment, but they travel more slowly than humans, and not as far. As Elizabeth Bates comments: “The Berlin Wall is down, and so is the wall that separates man from chimpanzee.”9
We must now determine where to go from here. Can we learn to live with these higher animals that are clearly no longer unfeeling, unthinking, stimulus-bound creatures of meat and bone? Can we meet them on their own terms? Can we even understand what their terms are? How shall we forge a new ethic that takes into account not only our fellow human beings and the fragile ecosystem of the planet, but the needs and wants of all manner of other sentient beings as well? Is it possible to structure a future in which not only Homo sapiens, but other conscious beings inherit the earth as well? We are still the creature that plans ahead the farthest, at least in any conscious sense. What kind of world should we create with our great planning skills and our newfound knowledge of the minds of apes?
7
Childside
I first encountered Bev* and Connie in 1981, when they were seventeen and nineteen years old, respectively. Both young women had been born with severe brain damage and were profoundly mentally retarded as a result. Despite remedial language programs, neither had learned to speak. Connie, whose motor-speech mechanism was impaired, often vocalized, but for the most part unintelligibly. Bev, who suffered cerebral palsy, was even more vocal, but was also largely unintelligible. Bev and Connie were part of a small experimental program at the Language Research Center in Atlanta, where I work with Kanzi, Sherman and Austin, and the other chimps. Before they began their visits to the center, the two institutionalized young women had been judged to have reached a level of cognitive development equivalent to that of a two-year-old. But to me, and to Mary Ann Romski, who was directing the program, there appeared to be an important difference between the two of them.
The research program was modeled after the center’s Animal Model Project, and therefore involved teaching speech-impaired humans to use lexigrams as referential symbols in a communicative way, just as we had taught chimpanzees to do. As I watched Mary Ann and our joint colleague Rose Sevcik work with these subjects, I became aware of subtle signs that Bev possessed a degree of comprehension that was absent in Connie. If, for instance, Mary Ann asked Bev, “Do you want a drink?,” Bev’s eyes might flick toward the drinking fountain. Or, if told, “It’s time to go,” Bev’s body might make a slight movement in response. Because of her cerebral palsy, voluntary movement was difficult for Bev, and she could not prevent herself from making involuntary limb and torso movements. Mary Ann and I talked about our separate observations of Bev’s apparent comprehension, and we agreed something different was going on with her compared with Connie. That difference was soon manifested in a different rate of learning lexigrams: Bev learned quickly, while Connie was slow.
I was not directly involved in Mary Ann’s research program, except as an enthusiastic spectator and occasional source of advice on how to proceed when teaching problems arose. To that point in time, my work with Sherman and Austin had served as a model and trouble-shooting device for the human program, particularly
in circumventing apparent learning barriers. Every one of the difficulties that Mary Ann faced in installing language in mentally retarded individuals, I had already experienced with Sherman and Austin. In every case, Mary Ann was able to overcome the problem using techniques I had used with the chimps. The rationale of the Language Research Center—to bring together language studies in humans and chimps—was vindicated by these successes. So, too, was it vindicated by my observations of Bev and Connie. Partly from seeing the effect of comprehension on Bev’s symbol acquisition skills, I became aware of the power of comprehension. I also became attuned to the subtle signs that betray comprehension. When, a few years later, Kanzi began to manifest such signs, I was sensitized to detect them. The center’s “childside” and “chimpside” were informing each other in important ways.
The Language Research Center had been established by the Georgia State University Foundation in 1981, and was a direct descendant of an innovative initiative that Duane Rumbaugh had promoted a decade earlier. He had joined the Yerkes Regional Primate Research Center in 1969, and became associate director and chief of behavior. Yerkes was (and still is) linked with Emory University. It represented the richest resource in the country for the study of primate behavior, including scientific personnel of many different disciplines. In the winter of 1970, the confluence of two factors led to Duane’s initiative. First, he became increasingly fascinated with reports from Allen and Beatrice Gardner of their language project with Washoe, and from David Premack of his project with Sarah. To Duane, the prospect of learning about the components of language acquisition through research with nonhuman primates represented a potential major advance in psycholinguistics. With an ape, it would be possible to manipulate the environment and observe the effects on language acquisition. Such manipulations were not possible with human children. The second factor was a publicly announced agenda by the National Institutes of Health (NIH) to fund research that could be applied directly to social and medical problems facing our society. (The NIH is the major funding agency for biomedical research in the United States.)
Kanzi: The Ape at the Brink of the Human Mind Page 21