Gorilla Talk

To the Editors:

As the target of many of its taunts, I am responding to Martin Gardner’s review of Nim and Speaking of Apes [NYR, March 20]. Gardner suggests that all research into two-way communication with animals is faddish and uses the two books reviewed to support his point. As an active researcher in the field, I would like to suggest that debunking can go both ways. Citing examples out of context and stripping them of supporting experimental documentation, the Sebeoks and Gardner are guilty of one of the oldest forms of journalistic deception. We are not, as the Sebeoks charge, basing our reports on “nothing more than a collection of anecdotes.” We are striving to outline the boundaries, to define the differences as well as the similarities, between human and nonhuman language use.

To say that the gorilla’s use of sign language is virtually identical to that of the human child is wrong; but to say that the gorilla’s use of sign language is uncreative, repetitious and forced or cued is equally wrong.

The first book reviewed is Nim by Herbert Terrace which states that “there is no evidence yet that apes understand any kind of syntax.” Terrace is evidently unaware of my publication of experimental evidence (controlled for cueing) of the comprehension of novel sign and spoken English sequences by a gorilla (AAAS Selected Symposium 16, 1978).

Gardner remarks that Terrace’s disenchantment with the chimpanzee Nim’s abilities arose when he studied his own extensive videotapes. These “extensive videotapes” are 3 1/2 hours of Nim’s signing sampled under artificial and high-pressure conditions which very likely contributed to the high levels of interruption and imitation he observed. Terrace’s extended conclusions on the language abilities of gorillas are based on fifty (50) seconds of footage produced for popular media consumption.

The second book Gardner reviews is Speaking of Apes edited by Donna Jean and Thomas Sebeok. Gardner seems to accept indiscriminately everything they say. I’d like to review several critical points.

The Sebeoks’ comments are opinions and conjecture, not factual statements. They have not examined my data and have neither research experience with apes nor expertise in American Sign Language.

Nonverbal cues are omnipresent in human communication as well as in ape-human communication. Speculating that perhaps “the best trainers are those most expressive in unconscious cueing,” the Sebeoks comment that “the apes’ teachers are anything but stone faced.” Facial movements and expressions are an integral part of sign language—they frequently function grammatically. A “stone-faced” teacher would not be a good model for sign language acquisition for either child or ape. Contrary to the Sebeoks’ assertion, it is easy to control for cues such as eye-pupil size and direction of gaze by wearing mirrored sunglasses. We have employed this and a variety of other controls. The result is that the gorillas’ spontaneous and appropriate signing continues unabated. Live observation and review of our videotapes indicate that in test situations requiring forced choices between objects or other materials, the gorillas are looking down at the materials, rather than searching our faces for cues, almost without exception. If we drop deliberate miscues by positioning, touching, leaning, or looking toward the wrong choice, the gorillas respond to the questions, not to the cues. We have restructured certain situations and our possible cues so as to deliberately mislead the gorillas. Instead of asking the usual “Where is your ear,” and so on, the experimenter asked, “Is this your ear?” pointing to her nose, and looking at the gorilla’s nose. In a recent test, the gorilla Michael responded in each case by correcting the questioner and not by following the cues.

The Sebeoks again reveal their ignorance of sign language structure in their arguments designed to dismiss sign modulations as mistakes. Errors in reproducing signs are not random for either human or gorilla users of sign language. Rather, a circumscribed set of parameters is systematically varied and many possible variations never occur. Neither Koko nor Michael routinely makes articulation errors in using signs such as drink. Errors are of a conceptual nature—eat or sip may be emitted where drink is appropriate, but the drink sign does not drift randomly about the signing space as the Sebeoks contend. Errors of articulation are made between signs whose locátion, configuration and motion are similar and are recorded as such. By moving the drink sign to her ear instead of to her mouth, Koko altered its articulation in a way that did not conform to any of the standard error patterns. Koko had never before this incident nor ever since placed the thumb of her fisted hand to her ear. The Sebeoks assume that because Koko had refused to use this sign on this particular day with this particular assistant, even when the teacher repeatedly demonstrated (“cued!”) the sign, that she had not mastered it; however drink was a sign Koko had then used reliably on a daily basis for several years. When Koko finally complied, a grin accompanied the distorted sign.

Koko’s humor, like that of the young child, is based on discrepant statements about overlearned relations. If the distortion is taken out of context, and the above constraints on articulation errors are disregarded, then the “drink in the ear” sign is perhaps best interpreted as an error. But given the contexts of the situation, of Koko’s behavior prior to and during the incident, the sign’s acquisition history and pattern of errors, the nature of infantile humor, and of the gorilla temperament (all of which the Sebeoks fail to take into account), it would be a mistake to categorize such a response as an error.

There are numerous (almost fifty) misstatements, nonsequiturs,misleading elliptical quotations and kindred erroneous remarks in the Sebeoks’ chapter. (Gardner’s review is based solely on this first chapter, not on any material by animal researchers.) One in particular evidently impressed a reporter from Time. He asked me half a dozen times if I had scored Koko’s inappropriate responses as errors in a blind test of her vocabulary. The Sebeoks state that the types of responses Koko gave to avoid the double blind test were not included in the four categories of errors I listed, so that “we may assume they are in fact not represented in the sixty percent score” she achieved. I could not believe my eyes reading this—those responses are indeed in the list of errors which is on the very same page as a quotation they include from that report!

Gardner asserts that “no teacher has bothered to record all the nonsense combinations produced by an ape….” I routinely record everything Koko signs in three formats: written notes, audiotaped samples and videotaped samples. Most of Koko’s signed communications are appropriate to the situations in which they occur. (Gardner calls these “lucky hits.”)

According to Gardner, one of my exceptional “claims” is that Koko has a capacity to rhyme the sounds of English words using signs. The full context of Koko’s “Flower pink fruit stink, fruit stink pink” rhyme was a dinnertime discussion of broccoli with two teachers, one of whom responded, “You’re rhyming, neat!” to which koko replied “Love meat sweet.” Following this incident, her ability to rhyme was tested. Koko successfully performed a task requiring her to produce signs whose English translations rhyme with the English translations of other signs. For example, Koko responded “do” to the word “blue” spoken by the experimenter and “wash” to “squash.” Note that these word pairs are not examples of Koko’s rhymes, as Gardner mistakenly assumes, but responses to test questions. She also demonstrated an ability to select from an array of objects those with rhyming English names or with names rhyming with an English word spoken by the experimenter.

Another of my “exceptional claims,” according to Gardner, is that Koko generates utterances which are innovative in a way paralleling metaphor. After documenting numerous instances of such novel descriptive phrases, we assessed the gorillas’ ability to appreciate metaphor using a test devised by Harvard psychologist Howard Gardner. Administered blind, the test involved the assignment by the gorillas of polar adjectives (such as loud-quiet and hard-soft) to pairs of colors. The level of performance of both gorillas (90 percent metaphoric matches) was on a par with that of seven-year-old human children (82 percent).

Gardner’s grand finale is quote from a book by Darwin published in 1871 presented as evidence that the Terrace and Sebeok pronouncements on apes fit neatly with evolutionary theory. Very few gorillas had been successfully maintained in captivity and no attempts had been made to assess their mental faculties at that point in time. Darwin’s statement was based on the mistaken notion that language is synonymous with vocal speech.

One cannot trace the evolution of language from an armchair in Indiana. By studying the cognitive capacities of man’s closest relative we may come a step closer to discovering what our ancestor’s language abilities were like five million years ago when humans and gorillas set out on separate evolutionary paths.

In summary:

Blind and double-blind tests have been administered, and the gorillas’ level of performance is significantly above chance.

The gorillas sign spontaneously and appropriately to themselves and to each other.

The gorillas sign to strangers.

The gorillas sign and respond to questions appropriately even when we try to mislead them with nonverbal cues.

The gorillas frequently inititate signed communication; the majority of their utterances are meaningful.

Francine Patterson

President, The Gorilla Foundation

Woodside, California