The AFU and Urban Legend Archive Language primates and language Select a topic - - - - - - - - Home Searches AFU FAQ - - - - - - - - AFU Animals Books Celebrities Classic Collegiate Death Disney Drugs Food GIF Language Legal Medical Misc Movies Politics Products Religion Science Sex Songs TV - - - - - - - - Other sites

From: cindy@nvg.unit.no (Cindy Kandolf)
Newsgroups: alt.folklore.urban
Subject: Primates and Language
Date: 13 Jan 1995 17:36:24 GMT

A short time ago, a question came up on this newsgroup about primates other than humans having been taught sign language. My response was that such experiments did exist but that in the opinion of many who have examined the experiments the claims are based on misinterpretation and wishful thinking. I received a few e-mail messages asking me to present this side of the argument, and my full response follows.

Warning: it is rather long.

-Cindy Kandolf, certified language mechanic, mamma flodnak cindy@nvg.unit.no
Trondheim, Norway
We're looking for a little, bewildered girl.

Some abbreviations, shortcuts, and other things that should probably be cleared up:

FR = Fromkin, Victoria, and Rodman, Robert. 1988. _An Introduction to Language_ (Fourth Edition). New York: Holt, Rinehart and Winston, Inc.

SP = Pinker, Steven. 1994. _The Language Instinct_. New York: William Morrow and Company, Inc.

NHP = non-human primates. There is no _way_ i'm going to type that phrase out a thousand times!

ASL = American Sign Language, the sign language most commonly used in the United States.

Clever Hans effect - Clever Hans was a horse who could, his trainer claimed, solve arithmetic problems (among other amazing skills). It turned out that the trainer was unknowingly giving him clues about what to do, for instance when to stop and start stamping his hoof (his main method of communication). The "Clever Hans effect" is a general term for the subject picking up clues from someone involved in the experiment.

The meanings of the signs made by chimps are written in ALL CAPS.

Article follows:

Before i begin explaining why, in my opinion and in the opinion of many cognitive scientists, the NHP language studies have not proven much, it is probably best to discuss what makes human language different than non-human animal communication systems. Other than human language, animal communication systems fall into three categories. One category is a finite number of calls - such as what we rather rudely call the screeches and hoots chimps use to communicate. One call might mean danger, another could be used to say that the danger is past, and yet another could mean "Follow me". The second is an analog system, such as the dance of a bee, in which each of a number of characteristics can be varied along a continuous scale to communicate some information. The overall direction of a bee's dance, for instance, says which way the food is; the vigorousness of the bee's wiggling tells how much food is there. And the third system is a theme (or a relatively small number of themes) with variations, like the songs of many birds.

Human language has none of these characteristics. It is infinite - there is no theoretical limit to the number of words that can exist in all the languages of the world. It is also digital - there is no continuous scale to vary upon, and small changes (changing one sound in one word, for instance) change the meaning of the whole utterence. And it is compositional - the meaning of each utterence is different than the meaning of similar utterences, and predictable from the components and from the rules governing the components.

Clever readers will recognize these differences in the discussion that follows.

In the early days of NHP language studies, two chimps were "adopted" by human families, who attempted to teach them spoken language. The first, Gua, was raised along side the couple's own baby son, and given no special training. She learned to recognize certain words and simple commands, but never spoke a word - not surprisingly, since human speech equipment is quite different than the corresponding parts of a chimp's mouth. In the second study, a chimp named Viki was coached to speak, to the extent that her "parents" actually molded her lips and tongue into the proper shape. Her trainers claimed that she could say four words, "mama", "papa", "cup" and "up". Some others who heard her speak claim that it was hard to distinguish them.

Obviously, these two chimps were operating at a great disadvantage. Chimps have completely different mouths than humans, and cannot manipulate them in the same way that humans can. However, chimps have similar hands and can use them in much the same way. So experiments in which the chimps could let their hands do the talking for them seemed to be an excellent way of levelling out the playing field. These studies fall into two groups: symbol-language studies and sign-language studies.

The two best-known symbol-language studies are the Sarah and the LANA (sometimes also known as the Yerkish) studies. Sarah was a chimp who was trained to respond to arbitrary symbols, each with a magnet on the back, displayed on a metal board. Unlike in other studies, Sarah was not allowed to spontaneously try to communicate; she had to wait for her trainers to start the interaction. In this case, what was being studied was not whether chimps could be taught language, but chimpanzee cognition - that is, how chimps think. The LANA study used "lexigrams" - arbitrary geometic shapes representing words - on a computer keyboard to teach four chimps an artificial language called Yerkish (after the Yerkes Regional Primate Research Center, where the experiments were conducted). This experiment was a mix: the researchers were most interested in the ability of chimps to use symbolic communication, but they were unquestionably trying to teach them language (a specific type of symbolic communication).

The sign-language studies, however, unquestionably have captured the public imagination more than the sight of chimps pressing buttons. I will therefore concentrate on these studies, making reference to the symbol-language when appropriate.

The first of the well-known studies involved a chimp by the name of Washoe. Washoe was raised, according to the couple that led the research (the Gardners), much as a child in a deaf community would be - surrounded by ASL in the way that Gua and Viki were surrounded by spoken English. At the age of four, she could allegedly make 85 signs, and combine them into two-word "sentences". For comparison, most human children can do the same when they are about two years old.

Probably the most famous of the NHP studies is that centering around Koko the gorilla. Koko's primary trainer, Francine Patterson, certainly makes the most far-reaching claims about the experiment. She says that Koko knows several hundred signs, composes sentences of varying length, and shows a fondness for jokes, puns, and metaphors.

However, both of these experiments were criticized for being too uncontrolled. A research team led by psychologist H. S. Terrace, composed of people who firmly believed the previous studies showed language ability in NHPs, set up a more controlled experiment with a chimp they named Nim Chimpsky. (He was given that name specifically because Noam Chomsky had stated that he did not believe NHPs were capable of learning language, and the research team hoped to prove him wrong.) The study included a program of careful record-keeping, and lots and lots of videotaping. The results were not what they had expected.

The beginning of the experiment was encouraging - Nim produced his first ASL sign after just four months. However, at the end of four years, Nim had learned 125 signs by the most optimistic count (Laura Pettito, another member of the team, suggested that Nim may have learned as few as 25 genuine signs). And while he put them together into "sentences" with a varying degree of signs in each, the researchers were disappointed to find that the content did not get more complex as the number of signs increased. In the words of Terrace: "His three sign combinations do not ... provide new information. .... Nim's most frequent two- and three-sign combinations [were] PLAY ME and PLAY ME NIM. Adding NIM to PLAY ME is simply redundant." [quoted in FR, p 393] The Terrace team began looking at what data was available from the other experiments, and found no indication that other chimps had done better than Nim.

What were the reasons for such different interpretations of the experiments? Obviously i am, in this article, presenting the view of a linguist who does not believe the NHP experiments showed that chimps have an ability to learn language, so i am presenting one side. The other side has had sufficient exposure in the popular press, on TV, etc. that i assume the reader is familiar with it.

To begin with, the claim that the NHPs had learned "ASL" stems from a lack of understanding about ASL. Most hearing people have had little contact with ASL and tend to assume that "sign language" is a series of rather crude, relatively obvious gestures. This is understandable, given that few hearing people have very much contact with sign language. However, to me at least it is amazing that people who claim to be teaching it would have this view. ASL, like all signed languages, is as rich and complex as a spoken language, and almost all of its signs are completely arbitrary. The vocabulary counts were as a rule made by people who were not fluent in ASL. The Washoe team did have one native signer, but this person's count apparently did not influence the final count:

        "Every time the chimp made a sign, we were supposed to
        write it down in the log. ... They were always complaining
        because my log didn't show enough signs.  All the hearing
        people turned in logs with long lists of signs.  They 
        always saw more signs than I did. ... The hearing people
        were logging every movement the chimp made as a sign.  
        Every time the chimp put his finger in his mouth, they'd 
        say 'Oh, he's making the sign for DRINK,' ... When the 
        chimp scratched himself, they'd record it as the sign
        for SCRATCH. ... When [the chimps] want something, they
        reach.  Sometimes [the trainers] would say, 'Oh, amazing,
        look at that, it's exactly like the ASL sign for GIVE!'
        It wasn't."  [quoted in SP, p. 337-8.]

        Jane Goodall commented, when watching Nim at work, that she

recognized all of Nim's "signs" as gestures chimpanzees make in the wild. [see SP, p. 338]

Another problem with the chimps' (and Koko's) language use is its lack of grammar. Here we are not talking about grammar in the sense of the long lists of rules children memorize at school, but rather the set of rules about how to put together meaningful sentences and how to determine the meaning of the sentences others make, something every human child learns without any direct teaching. As mentioned before, ASL is not like a game of charades, but a language that is as complex as a spoken language. The grammar of ASL is highly inflective, with inflections being indicated by changes in the manner of articulation (e.g. moving the hands in a different way, perhaps repeating a motion) and in facial expression and head position. None of the NHPs showed any sign of understanding this. (To be fair, i have seen no evidence that the trainers knew it either, so the problem may be one step further back!) The animals also didn't seem to understand much about syntax - recall the PLAY ME vs. PLAY ME NIM example. One of Nim's longer utterances was "GIVE ORANGE ME GIVE EAT ORANGE ME EAT ORANGE GIVE ME EAT ORANGE GIVE ME YOU" - this is "word salad", not a sentence following the rules of ASL syntax.

One of the characteristics of the Nim experiment that Terrace and his team found most disappointing was his lack of progress beyond the two-word stage. They found no evidence of further progress in other experiments, either. And none of the NHPs had, even in the most optimistic counts, managed to learn more than a few hundred signs. Compare this to a human child, who after four years of being exposed to language, can manage to produce sentences of varying length, sometimes including for example relative clauses, and has a vocabulary of thousands of words. (Experiments suggest that the average American high school graduate probably has a vocabulary of 45-60,000 words; see SP p. 149-51.) The NHPs may have come as far as the two-word stage, which human children reach on average at about age 2, but no farther; for human children the two-word stage is the launching pad for a long period of explosive growth.

The sentences that the NHPs could produce were not only very short, but they were monotonous. Many times the chimps simply imitated the sentence or a part of the sentence the trainer had just made. Sentences that were not pure imitation usually relied on stock "formulas" which the animals had been taught and drilled on many times. As already mentioned, Nim's sentences were quite repetitive - and Terrace et al found this to be true of all of the "signing" primates. Not only were the patterns simple, but as a rule the sentences were communicating simple wants - they wanted a favorite food, someone to tickle them, or a playmate. This shows that our fellow primates like the social life, yes, but it does not show that they use language like humans do. Even quite young children use language for far more things, including the most amazing non-sequitors and random observations (if you don't believe me, try hanging around a two- ot three-year-old for an afternoon). Furthermore, the NHPs studied seldom started a conversation spontaneously, as humans do from the moment they learn to talk.

The NHPs also broke many conventions that human children learn very early. For instance, ASL and in fact all signed languages have a conventional signing space, in front of the face and chest and going a short distance out to each side of the body. Not only is staying within that space a convention that ASL speakers all follow (similar to how it is considered polite to look at the person you are speaking to whenever possible), but some signs are distinguished from other signs only by their place of articulation within the signing space. For instance, if [FR] is to be believed, an otherwise identical sign (hand configuration and type of movement is the same) means SUMMER if made across the forehead, UGLY if made across the middle of the face, and DRY if made across the chin. The NHPs never seemed to understand this; they sign off to the side, under something, or in other ways that make it hard for their partner to see what they are doing. They also do not know that they should take turns (human children may interrupt sometimes, but as a rule they do wait until their partner is finished), but sign simultaneously with their human trainers.

The "signs" the NHPs make often stand for not just one object or one class of objects, the way humans use them. Each sign can mean not just the object being signed but things associated with that sign. If Nim or another chimp signed BALL, for example, he might mean "ball", or "Bring me my ball", or "the place where my ball is kept", or "I want to play with my ball", or even "It's time to play". Sometimes human children use a word like "ball" to mean all round things, but as soon as they learn to put two words together to make a primitive sentence, they stop using a single word to convey messages like "I want to play with my ball".

There is no documented case of a chimp teaching another chimp to sign or pass on any other communication system they were trained to use. Although some chimps who have been trained to sign have had offspring during the experiment, the baby chimps have never learned sign language without human intervention. Humans pass on language to their children as a matter of course.

Other than the Nim experiment, the NHP signing studies have not been particularly scientific. This was in fact why the Nim study was set up in the first place - to counter criticism of the poor methodology of the earlier studies. Record-keeping was not extensive, and others who wanted to examine the studies found it hard to obtain any raw data to use. In fact, the Gardners (Washoe's trainers) threatened to sue Terrace for using frames from one of their films in an article criticizing the studies. (Yes, he was using them without permission, but this doesn't sound like good scientific cooperation to _me_...) Questions have also been raised about the possibility of the Clever Hans effect sneaking into the experiment - trainers who after all are truly want to teach their subjects to communicate with them giving clues about how to respond. The Clever Hans effect is more strongly suspected in the symbol experiments than in the sign language experiments, however.

To me, the most striking difference of all between human and chimpanzee language use is this final one: the chimps needed training. As Fromkin and Rodman write: "In the case of the chimpanzees, each new 'rule' or sentence form was introduced in a deliberate, highly constrained way. ...[W]hen parents speak to children they do not confine themselves to a few words in a particular order for months, rewarding the child with a chocolate bar or a banana each time the child correctly responds to a command. Nor do they wait until the child has masteres one rule of grammar before going on to a different structure. Young children need no special training." [FR p. 395]

So there is the evidence against. I thought it would be fitting to finish up with a summary of what is happening now. The team that worked with Nim, as mentioned, have changed their opinion entirely: they are no longer working with Nim and are concentrating on debunking the other sign-language studies. Sarah's team, which didn't try to teach her language, is now working on a new experiment to study chimps' concepts of causation. The LANA (Yerkish) team admits that, while they learned much about how chimps reason from their earlier study, the chimps' language behavior was disappointing. They are now working with a bonobo or "pygmy chimp" named Kanzi, and claim that pygmy chimps are smarter than common chimps so they expect Kanzi's results to be better. The Gardners and others are working with a new generation of chimps in addition to Washoe, last i heard. Patterson is making ever more far-reaching claims about Koko's ability. In a rather diplomatic statement, Pinker writes that both the Garders and Patterson "have distanced themselves from the community of scientific discourse for decades." [SP p. 341] The Gardners, as i have mentioned, were not happy when Terrace used frames of their film in his work; they as well as Patterson seem to prefer popular-science type television shows and mass-market magazines to psychology or linguistics journals when it comes to publicizing their work.

In closing i have a point to stress. Saying that other primates are not able to use language is not saying that we should be proud of our superiority. I firmly believe we should respect and protect our fellow primates, but not because they are so smart or because they are just like us. We should respect them and all living creatures because of who they are, parts of an interdependent system that we fragile, naked humans need. Because we do not understand that system we need to care for all parts of it, not just that which we call "intelligent" (read: like us) or which reminds us of our own positive qualities. I do not think it is being conceited to claim that we have a behavior (language) that is specific to us, any more than the elephant would be conceited for claiming the uniqueness of his trunk or the bat for pointing out that he alone among mammals can fly. The other animals do not, as far as we know, attach any importance to our language: they do not miss it. Its importance comes only from the value we attach to it.

        "What an irony it is that the supposed attempt to being
        Homo sapiens down a few notches in the natural order has
        taken the form of us humans hectoring another species 
        into emulating our instinctive form of communication, or
        some artificial form we have invented, as if that were
        the measure of biological worth.  The chimpanzees' 
        resistance is no shame on them; a human would surely do
        no better if trained to hoot and shriek like a chimp, a
        symmetrical project that makes about as much scientific
        sense.  In fact, the idea that that some species needs
        our intervention before its members can display a 
        useful skill, like some bird that could not fly until
        given a human education, is far from humble!" [SP, p. 342]