The Collected Works of the Phantom Linguist

Can Primates Talk?
Humans seem to acquire language in a manner different from other types of behavioral learning. The onset of language learning is sudden (around the age of 2). We learn our native language without instruction, in fact, we can learn 2-3 languages at that age as easily as one. If we do not learn to speak between the ages of 2 and 6, it appears that we lose the ability to speak normally the rest of our lives. All this evidence suggests that we have a 'language organ' which other species do not possess, a segment of our brain which is triggered by a stage of development, much the same as walking is.

A simple way to disprove this Innateness Hypothesis, as linguists call it, is to demonstrate that other species have the capacity to speak but for some reason simply have not developed speech. A logical candidate for such a species is the chimpanzee, which shares 98.4% of the human genetic code. Chimpanzees cannot speak because, unlike humans, their vocal cords are located higher in their throats and cannot be controlled as well as human vocal cords.

It does not follow from their lack of speech, however, that chimpanzees are incapable of language, that is a human-like grammar. Perhaps they can acquire grammar and speak if they could only use grammar some way other than with a voice. The obvious alternative is sign language, since all primates have extremely dexterous hands and sign language is a language. You have probably already read about the regular chimpanzees Washoe and Nim Chimpsky, and the lowland gorilla Koko, all of whom learned to sign and interact very naturally with their trainers. All of these animals were taught to sign in order to get food, tickling, grooming, toys, and to get out of their cages. The question, then, is whether chimpanzee and gorilla signing is language; is it based on grammatical rules?

What is Language?
We have all managed to get around in a foreign country by 'talking with our hands'. This is possible because language is not the only symbolic means of communication. In order to prove that chimpanzees and gorillas are capable of language, and not simply a different kind of symbolic communication system, they must learn a sign language with the basic characteristics of human language. These include the following.

1. The language must be based on arbitrary symbols. The sounds of words do not normally contain any hint of their meaning. The sound of the word "dog" does not suggest dogs. In fact, the French word "chien" and Russian "sobaka" work just as well to express the same thing as "dog" though they express radically different sounds. So signs must not look like their meanings.

2. The arbitrary symbols must be used in strict order. In English we can say "I see the dog" but not "the see dog I". If we move words around in sentence, we have to change the morphology, the shapes of the words, as in "The dog was seen by me."

3. Recombining the same lexemes always results in a different meaning: a "houseboat" is quite a different object from a "boathouse", even though they comprise the same words (I mean, 'lexemes'). So chimps will have to be able to 'recombine' the same symbols to create different words, e.g. "top" versus "pot", and recombine words to create different phrases. In fact, chimps will have to come up with new, creative combinations to describe new, previously unencountered situations. So if Bill Gates were to introduce trains to get from one room to another in his house, we wouldn't think twice before calling it a 'housetrain', even though we have never heard the word before. Chimps have to be able to do the same.

4. The symbols will have to be relative and not absolute. Virtually no word in any human language has only one meaning. Rather, words have meanings only in context. Take a look at these examples.

The table was covered with junk food.
We'd better table this motion.

The chair was inviting.
She can chair a meeting harshly.

Chimps must be able to determine meaning on the basis of sentence context. Any dog or horse can learn to associate one sound with one meaning.

5. Chimps will have to talk Jabberwocky. Remember the 'Song of the Jabberwocky' from Alice in Wonderland mentioned in 'But There are no Such Things as Words?' That bit of free knowledge pointed out that sentences contain not words, but lexemes and morphemes. Lexemes are not so interesting because they are simple symbols: sound-meaning pairs. Morphemes are more interesting because they refer to the grammatical categories which define language and their meaning varies with context: "John was painting [verb] a painting [noun] while painting [participle] his room." In order to prove that pongids are using language and not a simply a semantic system concatenating ordinary symbols, there must be evidence of those arbitrary morphological categories which distinguish language from other types of communication systems and cognition (mental) processing.

It is possible to communicate using symbols (audible or visual) plus a semantic strategy for interpreting the meaning of symbol combinations. For example, if I were to say simply CAR HIT MAN, and you knew the meanings of these three words, you could figure out that some car probably hit or will hit some man-even without grammar. However, language is different. With the grammatical rules of language, not only can we specify tense, but word (lexeme) order actually plays a subordinate role. I could say the following in English, using the same lexeme order but with a wide variety of meanings:

The car hits the man.
The car is hit by the man.
Did the car hit the man?
The car that hit the man...

Notice the wide variation in the meaning of the sequence car hit man when joined by grammatical morphemes. Morphemes are critical to the understanding of the intent of the speaker in these sentences. So if we wish to prove that chimps are capable of LANGUAGE and not simply a symbolic communication system, we must demonstrate that they can learn and manipulate morphemes and syntax, as opposed to simple absolute symbols plus semantic strategies for interpreting them. Grammar is the soul of language and morphology is the expression of grammar.

So, Can Chimps Talk?
The evidence is mixed but not promising. Chimps and gorillas can easily learn up to 120 different arbitrary symbols if taught those signs using conventional reinforcement techniques. Trainers teach chimps these symbols by taking the chimp's hand and forming the symbol, then giving the chimp a food treat if he does it himself on the correct cue. It takes as many as 100 tries to get a chimpanzee to correctly form one sign. All symbols thus far have been absolute symbols whose meaning do not vary with context.

Kanzi, an unusually intelligent Bonobo chimp recently trained at Georgia State University, is remarkable in that he learned to use around 200 symbols on a portable electronic symbol board, a computer with buttons in the shape of absolute arbitrary symbols, rather than manual signs. More interestingly, Kanzi learned how to use this board while watching his foster mother, Matata, being taught by traditional reinforcement methods. So Kanzi did learn how to use arbitrary symbols without being taught, although he did observe direct reinforcement of each symbol during the process and the symbols were taught one at the time.

The evidence for the mastery of syntax is not so convincing. While chimpanzees can learn to order their symbols to get what they want, it is not clear that they have mastered syntax. The reason is that when they initiate communication, even adult chimps often abandon the order they have learned and phrases such as fight mad Austin (a famous utterance of Kanzi's friend Panbinisha), but the order doesn't seem to matter. Apparently Panbinisha was trying to express, "there was a fight at Austin's and someone (Austin?) was mad". It is interesting that even human children simply don't make such errors once language acquisition begins and certainly adult speakers would never utter such strings. Notice it lacks any morphemes ("-ing", "at", "I", "she", "-ed"), the hallmarks of grammar.

There have been occasional reports of chimps signing new combinations of words in an acceptable order in response to new situations. Washoe, for example, once was in a boat on a pond when she encountered her first duck. She signed 'water bird'. This could be a new compound noun or it could be two separate responses to the water and a bird. The problem is that we have no way of measuring the chimp's intent.

Perhaps the most troubling outcome of the chimp and gorilla research is the lack of 'Jabberwocky' evidence: none of the trained animals seem to assimilate grammatical morphemes. The best translation of a chimpanzee phrase corresponding to 'Give me the orange' is 'give Washoe/me orange' (where 'Washoe/me' is the hand pointing back at the signer). There is no evidence of tense, number, agreement, articles, or pronouns in primate signing. Certainly prefixes and suffixes do not exist. This is important for the symbols mastered thus far by chimps are absolute symbols that refer to real things in the real world, on a par with a system of highway signs. Grammatical morphemes refer exclusively to the categories and relations of grammar and represent the decisive proof that an organism is using language. Why can't chimps and gorillas use grammatical categories?

The maximum number of lexeme symbols (simple nouns, verbs, and adjectives) is interesting, too. The brightest chimpanzees master fewer than 200 of these symbols by adulthood. Human children know about 50 by the time they are 18 months old, when they begin learning nouns, verbs, and adjectives at the rate of about 5 per day. So the maximum achievement in symbol acquisition among other primates is more than matched by human children before the onset of language acquisition. In fact, the type of phrases spoken by chimpanzees and gorillas also resemble those spoken by children before they acquire their first language: "give orange" or "give Bobby orange" (without the pronoun me) are common expressions of children who have not yet learned to speak.

The results suggest that while chimpanzees and gorillas are far more intelligent than anyone had imagined during the first half of this century, they are not capable of human language. Rather, they have a primitive version of the semantic ability children use to begin learning language. Even though we are 98.4% genetically identical to chimpanzees, the difference is nonetheless qualitative, not quantitative. That is, the difference is not that we simply differ in the quantity of intelligence we have, but human beings seem to have a different kind of intelligence. The philosophical implications of this conclusion are interesting but the phantom linguist will leave that for a phantom philosopher.

Next: Mama Teached Me Talk >