In response to:
Gorilla Talk from the October 9, 1980 issue
To the Editors:
In his reply to my letter [NYR, October 9], Herbert Terrace [NYR, December 4] accuses me of not coming to grips with the facts which led him to “reverse [his] original interpretation of Nim’s multi-sign sequences as sentences.” I am familiar with the facts in Terrace’s study and they speak to somewhat different conclusions than those he has drawn. Moreover, there are indications that Terrace’s experiment was an example of what the Sebeoks call the self-fulfilling prophecy.
Co-principal investigator on his project was Thomas Bever, who before the study began, had stated in his book with Fodor and Garret, The Psychology of Language, “What have the chimpanzee studies shown us about language? Literally, nothing.” Bever et al. contend that even if the studies had been fully successful, they would not have been relevant to the question of the species-specificity of language-learning capacities in humans.
The semantic and syntactic analyses carried out by Terrace, Petitto, Sanders, and Bever (Science 1979, 206, pp. 891-902) providing evidence of structurally constrained regularities in the signed utterances of a chimpanzee support and extend earlier work because both their evidence of nonhuman discourse patterns, and their claim that such evidence invalidates that important result, are questionable.
Terrace asserts that because the signing apes sometimes interrupt, imitate, and respond to the utterances of others, they are not using language. Each of these patterns also occurs in child speech. Katherine Nelson, in addressing Terrace’s conclusions (Science News, 1980, 117, pp. 298-301), stated that children do interrupt. “To suggest that the young child is a restrained, polite and sophisticated conversationalist in contrast to the ape is sheer nonsense.” She also commented that imitation “seems to play a major role in the acquisition of lexical items and for some children also in establishing new grammatical rules.”
In a review of Terrace’s book in Nature (1980, 287, pp. 369-370), linguist Jean Berko Gleason comments that his conclusion that Nim’s three-sign combinations do not provide any more information than his two-sign combinations “is based on a rather selective view of the data.” She performed an “armchair analysis” and found that “there are a number of instances listed where a three-sign combination seems to build on earlier two-sign combinations in a way that very much resembles children’s language.”
Terrace is also selective in the excerpts from my thesis he chooses to make. He fails to provide context surrounding the utterances by Koko which he presents as examples of unrelated strings of signs. It would be easy to review transcripts’ of the speech of young children and excerpt similar seemingly non-sensical utterances to make a similar point. When he quotes me as saying that the majority of Koko’s utterances were elicited, he is strictly correct; however, an analysis of the samples contained in my thesis reveals that an average of 40 percent of Koko’s utterances were spontaneous and 60 percent elicited—close to an even conversational give and take. Terrace asks for a rigorous description of the training used to establish in the gorillas linguistic skills such as perception of spoken English, rhyming, and appreciation of metaphor. In each of these cases the answer is that it was much the same as that the average child receives: None, other than a general exposure to spoken language and the simple informal examples which occur in everyday speech. It is important to point out that we are in large measure attempting to assess the gorillas’ capabilities given their immersion in a bimodal bilingual environment. Drill and training are minimal. One primary goal of my work is to determine what the gorillas will do on their own with the signs they have learned.
Although Terrace says he is confident that the same patterns of discourse that were reported in his article in Science hold true for Koko after viewing the film Koko, a Talking Gorilla several times, he repeatedly fails to report even the title of the film accurately. He suggests that I perform a discourse analysis of Koko’s signing in that documentary. Terrace is evidently not cognizant of all the methodological problems which make such an analysis scientifically unsound: All filmed footage of Koko has been shot non-continuously, that is, the camera men did on-the-spot editing and, in the case of the film Koko, a Talking Gorilla, further commercial editing was done by the producer with the result that some parts are no longer veridical with respect to timing, sequencing, and details of content. In addition, there are many segments in which the teacher is not in frame with the gorilla and therefore the records of the interactions are incomplete. Finally, the performances documented by these films cannot be considered normative because of the conditions under which they were made: The presence of equipment, lights, and a camera crew can alter dramatically the behavior of both human and ape. The only visual record which could be used to yield analyses with scientific validity would be continuously run videotapes made by project personnel who are fully adapted to both the apes and the recording sessions.
The Sebeoks [NYR, December 4] appear to be either unwilling or unable to entertain ideas other than their own; therefore, any response to their rather shrill and offensive letter would be an exercise in futility.
Debate is healthy and vital to the advancement of science. The present spate of letter writing, however, is becoming counterproductive and I feel that my time will be much better spent conversing with the gorillas.
The Gorilla Foundation
To the Editors:
The promising studies on communicating with the apes have become bogged down in controversies. The problems are reflected in the Patterson, Terrace, and Sebeok letters to The New York Review, but are rooted in much more fundamental issues than are discussed in these letters. It may surprise most readers that there is no agreement among scientists as to how comparisons of behavior should be made, and this is the cause of the confusions.
The issue may be made clear by an example. When different languages are compared, the linguist assumes (correctly, in my opinion) that the differences are in the languages, not in the brains of the people speaking the languages. But when the communications of apes are considered, no such assumption is justified.
Gesture (vision in and hand movements out) is normal for both ape and human. In training an ape, the human teacher is using a human brain to increase the ape’s repertoire of gestures, but the learning is in accord with the ape’s biology.
Apes cannot learn to speak. Even after great effort—including bringing up apes with children—apes do not speak. Children easily learn to speak because of their biology. A large part of the cortex of the human brain is concerned with speech and we know that insults to the cortex affect speech in a variety of ways. In monkeys, however, even after extensive removal of the cortex, all the normal sounds aare still produced. They may be elicited by electrical stimulation of primitive parts of the brain. The sounds made by non-human primates are parts of a biologically determined system of communication that is minimally modified by learning.
The critical point is the interrelation of the brain and learning. Human beings easily learn to speak; apes cannot speak at all, and it is the biology of the brain that shows why this is the case.
From the point of view of comparison or evolution it is misleading to use the same word, “language,” for two very different kinds of behavior—speech and gesture. If gestures are called “words,” and sequences of gestures “syntax,” the vocabulary constantly confuses the comparisons and reduces the apparent differences between the species.
Speech differs from the sounds used in the communications of nonhuman animals because short, meaningless sounds (phonemes, alphabet); are combined into meaningful combinations (morphemes, words). It is this sound code which makes human languages open and able to easily add new words. The gestures taught to apes symbolize words, but no ape learns a code with which it makes words, and so on ape learns the most important aspect of human language. This is why confusing gestural communication with speech is so misleading.
While American Sign Language consists primarily of signs for words, there also are signs for letters, and, if necessary, a word may be spelled out. That is very slow and it is easy to see why gestures for words are much more useful for rapid communication. But deaf humans read, so, both through the alphabet and reading, human sign language is enriched in ways that ape signing is not.
Further, there is an enormous difference in motivation between humans and apes. Children want to babble and later talk. Deaf humans make great efforts to communicate, easily learning hundreds of signs. Even a minimal use of signs by apes requires the time, help, and motivation of human beings. Human speech is a complex behavior which is new from an evolutionary point of view and is not shared by any other primate.
The fundamental issue is not Clever Hans; it is ignoring the biological differences between apes and human beings. Those who compare language behaviors must consider the brain or confusion will inevitably result.
Department of Anthropology
University of California, Berkeley
Martin Gardner replies:
Dr. Francine Patterson is surely right in saying that her letter exchanges in the NYR have become counterproductive. Indeed, she and her critics seem to argue across a chasm so wide that she finds it more worthwhile to talk to her gorillas. Those who care to read more about the controversy are urged to look up Patterson’s exchange with Herbert Terrace in Science, Vol. 211, January 2, 1981, pp. 86-88, and Garry Hanauer’s article about Patterson’s work in the November 1980 issue of Penthouse. There are excellent photos of Koko in the nude.
Professor Washburn’s letter strikes me as a mix of obviosities and irrelevancies. Who doesn’t know that apes and humans have different brains, that humans learn to speak but apes don’t, and so on? But can an ape acquire a dim awareness of syntax, or does it merely arrange its gestures in strings that either imitate a trainer, or are taught by conscious or unconscious reactions by a trainer when the ape accidentally hits on a meaningful sequence?
Patterson is convinced that Koko not only creates true “sentences,” but also invents clever metaphors, makes up word rhymes and gestural puns, and speculates on where she goes when she dies. Her critics believe that Patterson is deceiving herself. Everybody agrees that apes and humans are biologically different. The significant question is whether the difference is so large that apes cannot grasp simple syntax. This is not a trivial scientific question, and I am puzzled that Washburn nowhere comes to grips with it