Kanzi: The Ape at the Brink of the Human Mind

by Sue Savage-Rumbaugh

  Mary Ann and her colleagues fostered naming and comprehension skills as the subjects learned each new lexigram by following the same procedures employed with Sherman and Austin five years earlier. Bev, Connie, and Ruth were like Sherman and Austin in that naming and comprehension skills needed to be fostered by specific training. At the time, we believed that this said something fundamental about the nature of language and how it must be put in place in ape and impaired-human brains. But as we were later to discover through our experience with Kanzi, it had at least as much to do with the teaching paradigm we used.

  Nevertheless, the study was important at the time, and demonstrated that “initiating instruction at the level of communicative request is a viable beginning for the establishment of symbolic communication in persons with mental retardation who have severe oral language impairments,” concluded Mary Ann, Rose, and James. “The four operational components of symbolic communication were observed concurrently after these individuals progressed into Phase 2 of the study”8

  Not only did labeling and comprehension skills emerge as the subjects’ capacity for symbolic communication emerged, but so too did other language-related behaviors, such as spontaneous initiation of communication and an improvement in comprehension of spoken words. In Bev’s case, improvement in her own speech also emerged. When Bev began the program she had been unable to comprehend the English words for any of the lexigrams she eventually was able to use. But as she learned to use the lexigrams, she also comprehended the spoken word for them, doubtless because Mary Ann and her colleagues also spoke the word relating to a particular lexigram when it was indicated. This comprehension seemed to drive Bev’s attempts to vocalize the lexigram words. Two words were clear, namely, “apple” and “Bev”; for the rest, she produced bisyllabic approximations where previously her utterances had typically been monosyllabic. Although Connie eventually comprehended five spoken versions of the lexigram set, she produced only a ma or ba vocalization when asked “What’s this?” Ruth was hindered by severe hearing impairment, and neither comprehended English words nor produced word approximations.

  One of the characteristics typical of severely mentally retarded individuals is that they tend to be responders, not initiators of communication. As their vocabularies expanded, each of the three subjects began increasingly to initiate communication with their instructors. Ruth was first, after she had learned four lexigrams. Her first spontaneous request was soda, to which she then waited for a response. Connie was the second of the three to begin spontaneous use of her lexigrams, after she had learned eleven of them. Bev began after learning twelve lexigrams, but did so much less frequently than Connie and Ruth. When Mary Ann and her colleagues saw the emergence of spontaneous communication, they decided to respond as parents do with normal children; they treated the utterances as intentional. They therefore adopted a “What you say is what you get” principle. The unfolding of these various communicative skills as the subjects increased their vocabularies was impressive, and it reveals the power of language as its components become assembled.

  Another manifestation of the power of language was an unexpected change in the subjects’ social demeanor. They became more sociable and positive and, instead of waiting for things to be done for them, actively took part in their world. They threw fewer temper tantrums and were able to attend to tasks much longer than previously. “They are happier,” observed Rose. “All of this we think is every bit as important as the communication skills. And basically it all can be traced to the word and sentence skills they acquired.”9 Stephen L. Watson, director of the Developmental Learning Center of Georgia Regional Hospital, said of Bev, Connie, and Ruth’s transformation: “This is not typical of institutionalized people, who are complacent because they grow accustomed to having their needs met before they ask.”

  Rose’s remark in attributing Bev, Connie, and Ruth’s increased engagement with their worlds to the power of language seemed plausible, given that Homo sapiens is a creature of language whose social fabric is very much the product of words. But her conclusion went beyond instinct, and included a careful study. The study addressed the possibility that the subject’s change in overall social demeanor might have been the result of the novel interaction to which they were exposed during the program, rather than the acquisition of language skills. Rose, Mary Ann, and Adele A. Abrahamsen compared various aspects of Bev, Connie, and Ruth’s behavior with those of individuals who had been exposed to daily contact with instructors but had not been taught to communicate symbolically.

  The individuals in the control group did show an increase in their sociability, as Bev, Connie, and Ruth had, but their intentional communication, rate of attentional shift, and attention complexity did not change. Significantly, neither did that of Max, who was exposed to symbols, but failed to learn. The improvement in Bev, Connie, and Ruth in these three measures can therefore be seen as a direct result of having acquired a degree of symbolic communication. “The acquisition of lexigrams occurred as part of a larger package of developmental changes in this study,” noted Rose and her colleagues, “which is suggestive that the process of acquiring the new symbols is embedded in a causal structure of changes involving several [cognitive] domains.”10

  This study presents a particularly clear insight into the connectedness of cognitive domains, and the power of language in humans in raising general intellectual achievement. I see a similar pattern of transformed behavior in language-trained apes, who appear to be more reflective, attentive, and sensitive to communication with humans. This is strong evidence, I believe, for the evolutionary continuity of the mental substrate between apes and humans.

  By the early 1980s, Duane’s vision of a decade earlier had already paid off. He, and later Mary Ann and her colleagues, had made entirely unexpected progress in teaching augmentative communication skills to severely mentally retarded children, by applying theoretical and practical knowledge gained from our ape-language studies. Individuals for whom all traditional methods of speech and other language training had failed, had learned to communicate for the first time in their lives, using Duane’s computer-based keyboard lexigram system. “The results with the Georgia Retardation Center and Developmental Learning Center projects had exceeded all expectations,” Duane now admits. “Quite simply, most people expected that the children would learn nothing at all, but they ended up learning a lot.”11 Not only had individuals in the program learned productive competence of a vocabulary of lexigrams, but most of them also generalized that competence to labeling and comprehension, and even to spontaneous communication. Most language intervention programs cannot make such claims.

  Despite the surprising efficacy of the Sherman and Austin teaching approach, it did conform to what was believed a necessary aspect of language learning in severely mentally retarded individuals: namely, that language skills had to be drilled into place through some kind of structured teaching. True, since the mid-to late 1970s, language intervention practitioners increasingly emphasized a more naturalistic setting for such instruction, but the acquisition of language itself was not naturalistic. Mary Ann and her colleagues worked thoughtfully to produce a naturalistic teaching environment at the Language Research Center, including lexigram games. Nevertheless, they considered themselves active teachers, putting requesting skills in place and watching as other skills emerged.

  When, in the spring of 1983, we began to realize that Kanzi had spontaneously acquired a small vocabulary of lexigrams and comprehension of a few spoken words, we were forced to rethink our hypothesis about the acquisition of symbolic communication in apes. Equally, rethinking would be necessary on the “childside.” In a description of the efficacy of the instructional approach with mentally retarded children, Mary Ann and I had said, “These subjects needed to have language learning broken down into small units, just as Sherman and Austin had.”12 True, these subjects had learned symbolic communication through having language broken down into small units, but perhaps they too
could acquire these skills spontaneously, without teaching, as Kanzi had?

  “The observations with Kanzi had an immediate effect on our thinking,” recalls Mary Ann. “I was anxious to be able to apply to mentally retarded individuals what Kanzi was telling us, and I was particularly keen to do it with younger children.”13 Most of the individuals in the Georgia Retardation Center and the Developmental Learning Center studies had been adolescents or young adults. Everything that linguists have learned about language acquisition in mentally normal individuals indicates the importance of early exposure—the existence of a critical period. Those who try to learn a second language after childhood are aware of the struggle it can be, compared with language acquisition in childhood. Mentally retarded individuals who are trying to learn a communicative system after childhood are therefore doubly disadvantaged, by age and by their cognitive impairment.

  In the spring of 1983, a serendipitous event presented an opportunity for beginning to apply a Kanzi-like paradigm in mentally retarded school children. A group of local schoolteachers who were completing a graduate course at Georgia State University paid a visit to the Language Research Center, to observe the program with Bev, Connie, and Ruth. They were enthusiastic about what they saw, and intrigued by the description of Kanzi and his spontaneous acquisition of symbol use and comprehension. They said: “Have you thought about applying some of this out of the laboratory, to children in school, for instance?” There is often a suspicion of experimental programs in schools, not least because researchers all too frequently treat the exercise as a way of collecting data, and don’t have the long-term interests of the schools at heart. Making an approach to engage in some kind of experimental program therefore has to be done with great diplomacy, but it was something Mary Ann and her colleagues planned to do at some point. The invitation by the teachers who visited that day, therefore, was more than welcome. “It was music to our ears,” recalls Rose.14

  Through further discussions with the teachers and their principals, and a scramble to find funds from various sources, a summer program was set up at the Language Research Center for a small number of students from the Clayton County Public School system. The school system benefited because it could offer a summer program for which it otherwise had no funding, and the Language Research Center benefited, because it was taking its first significant step from research in the laboratory to application in schools. The successful summer program revealed that it was possible to make significant progress with children at this severe level of retardation in an unstructured setting that zeroed in on comprehension rather than production.

  After this pilot program, Mary Ann wrote a grant proposal that fully incorporated the Kanzi paradigm along with the idea of studying young children in Clayton County schools. As typically happens in such cases, a grant review committee came for a site visit, to see what the Language Research Center offered, and to hear the arguments in favor of the proposed work. This was in 1984.

  “It was early in the morning,” remembers Rose. “We were in the group room, sitting around a table with the review committee. One person on the committee, a very prominent individual in mental retardation research, was saying very kind things about our work, about how innovative it was, and how important for potentially changing the lives of mentally retarded individuals. He was being just too kind, and you knew a giant BUT was coming down the pike. Sure enough, at the end he said, ‘But, there is nothing in my experience nor in the mental retardation literature that would indicate such an approach has any chance of working.’ We spent the rest of the day telling him and his colleagues about the success with the chimps, and about the history of how such successes had always translated to mentally retarded individuals.”15

  Rose and Mary Ann must have made a persuasive case, because the grant was awarded.

  The new study began in 1985, with thirteen mentally retarded male students having a mean age of twelve years, four months. None of the children had more than ten intelligible spoken words, and all had failed to acquire communicative skills by other teaching methods. Mary Ann’s grant was for a two-year study, which not only would explore the efficacy of truly naturalistic learning but would also compare the effects of school and home environments. For the first year, therefore, half the children were exposed to lexigram use at home, and half at school. During the second year, they used the system both at home and at school.

  Mary Ann called the program the System for Acquiring Language (SAL), and it had five components.

  The first was the computerized keyboard, which by this time was smaller and more powerful than previous versions, and now included a speech synthesizer. Portability was a crucial factor, because if the system were to be a feasible part of an individual’s life, it must not be cumbersome and obtrusive. Initially, the children had a Words+ Personal Voice II System, via a Unicorn touch-sensitive expanded keyboard, attached to a Votrax word synthesizer. Later a SuperWolf was substituted for the Personal Voice II. Technology has been central to the language studies from the beginning, first at Yerkes and then at the Language Research Center, and the program has benefited from the microcomputer revolution that has occurred in the past decade. Compared with what is available these days in terms of sophistication and computing power, the SuperWolf is rather primitive. But it is effective, robust, and inexpensive. No doubt great improvements are possible.

  The second component was the vocabulary of lexigrams, the assembly of words that were selected in the beginning. Because teachers and parents would be interacting with the children, the written English equivalent was printed on each of the keys, to facilitate two-way communication.

  Third, unlike the previous program, in which individuals were taught request skills, no direct teaching was to take place. The children were encouraged but not required to use the keyboard in the communication opportunities that occurred in their daily events.

  The fourth component involved the children’s adult partners and their active role in communication. After instruction in use of the system, the adults employed the keyboard in ways we had done with Kanzi. That is, they discussed what was going to happen, made comments, or asked questions, all using normal spoken sentences, but hitting the keyboard as well at appropriate times. For instance, a parent might say, “Johnny, let’s go outside and ride your bike,” where “outside” and “bike” appear as lexigrams on the keyboard.

  Last, Mary Ann and her colleagues needed a way to monitor the children’s progress; they achieved this through a Teacher/Parent questionnaire.

  At the end of the two-year program, Mary Ann and her colleagues had amassed more than 31,000 communicative events, collected as audiotaped interactions, which were then transcribed. All the children eventually acquired a vocabulary of symbols, though the size range was large. The lexigrams in the children’s vocabularies were those they had effectively chosen to learn, not those their teachers had wished them to learn, just as Kanzi had extracted from his language environment those words that were most salient to his life. All the children learned to use the symbols in clear communicative ways, often accompanied by naturalistic gestures. “In general, the SAL permitted the youths to convey specific information that their partners could respond to, thus promoting the initiation as well as the continuation of conversations and the addition of new information,” remarked Mary Ann and Rose, in a report of the program.16

  As had been observed in an earlier study of mentally retarded individuals using the Sherman and Austin instructional regime, there emerged two classes of learning patterns in this program, beginning and advanced. The four youths who displayed the beginning learning pattern acquired production and comprehension skills only slowly, and had a small vocabulary at the end, between twenty and thirty lexigrams. The advanced learners acquired large vocabularies (some with more than two hundred lexigrams) and developed comprehension and production simultaneously and rapidly. Individuals in this group also learned to use combinations of lexigrams and other symbolic skills, inclu
ding the recognition of printed English words and categorization.

  Mary Ann and Rose were able to identify the underlying cause of the difference in performance between these two groups by comparing results of certain language and cognitive tests that had been administered prior to the study. “The salient factor that distinguished the two groups was the speech comprehension skills they demonstrated at the onset of the study,” they noted.

  We suggest, then, that individuals who comprehended speech prior to the onset of the study readily extracted the critical visual information from the environment, paired it with their spoken language knowledge, processed it, and produced symbolic communications. Individuals with limited comprehension abilities apparently were confronted with a different task. They had to segment the visual component of the signal, develop a set of visually based symbol experiences, process the visual information, and then first comprehend and later produce symbolic communications.17

  The importance of comprehension in the process of language acquisition has emerged repeatedly, both in ape and in human studies.

  Early in the Clayton County Public Schools program, teachers and parents asked Rose to include in the lexigram vocabulary words that help mediate normal social interaction, such as please, thank you, I’m finished, help, yes, no, and goodbye. The children incorporated these words into their productive vocabulary surprisingly quickly. These social-regulative words were used frequently in the context of social interaction, but there was no overall increase in the use of the SAL as a result. People who interacted with the children very much appreciated their ability to communicate in a way that had the semblance of being “more like a normal child.” This is important, because one of the aims of giving mentally retarded children the ability to communicate is that they should be able to become more a part of the outside world. Strangers’ perceptions of such children are significant in the children’s acceptance in that outside world.