by Bob Kopp
1 June 02000
Language is, without a doubt, one of our species' most important abilities. Many philosophers have claimed that it is our ability to speak that separates humans from other animals. One of the first questions that comes to the mind of any child or scientist is 'why?', so it is only natural that humans wish to know why they can speak and why they can ask 'why?'.
This paper examines the evolution of human language through comparative studies of Homo spaiens' great ape cousins. These studies can provide clues about when the structures necessary for language first appeared. With this information, it is easier to form hypotheses about the selective pressures that lead to the evolution of these structures.
The question of when and how language arose in human beings is a controversial one. Some researchers have argued that natural selection, while successful in many other domains, cannot explain the evolution of language. Among those who accept natural selection, there is disagreement about when the ability to speak evolved; some believe that the basic structures necessary for language were present in a primitive form in our common ancestor with other great apes, while others believe that these structures are uniquely human.
Among the most prominent advocates of this view are paleontologist Stephen Jay Gould and linguist Noam Chomsky, founder of the modern discipline of linguistics. Gould proposes that language is an example of what he calls a spandrel, or an exaption: a trait which, having evolved for one reason, is later used for another. In architecture, a spandrel is the area at the top of an arch. It was developed purely for structural reasons, but during the European Middle Ages became highly ornamented. The use of spandrels as ornamentation is Gould's archetypal example of an exaption (Dennett, 1995).
Gould argues that humans evolved through natural selection not a capacity for language, but a general-purpose learning device. The use of this learning device - the human brain - for language is, in Gould's view, an exaption. Chomsky, who is famous for his view that the human brain contains the basics for language within it, nevertheless agrees with Gould that this capacity did not evolve through natural selection. He has not proposed any theory of how language did develop, but suggests that the answer may be found in the study of physical laws (Pinker & Bloom, 1990).
Most evolutionary biologists agree with Gould that exaptions play an important role in evolution, but would argue that natural selection would have to modify an exaption before it can be used for a complex task. A bird may use its wings to shield its eyes without natural selection modifying its wings, but natural selection is needed to transform the feet of a land bird into flippers suitable for an aquatic bird. The belief that any complex biological system is explainable by modification by natural selection is a primary tenet of the neo-Darwinian synthesis, the view held by most modern evolutionary biologists. Advocates of the neo-Darwinian view believe that the theory of natural selection can explain human language as well as it can explain any other complex, biological trait (Pinker & Bloom, 1990).
There is by no means, however, a consensus among neo-Darwinians as to when and how language evolved. Pinker (1994) is a strong advocate of the view that the ability to speak and the structures necessary for language are unique to the genus Homo. Expecting other great apes to possess some form of these capabilities, he argues, is as absurd as expecting the guinea pig-like hyrax, the elephant's closest living relative, to possess a primitive form of the elephant's trunk.
Other researchers, such as Parker and Gibson (1979), support the view that primitive forms of the structures that allow humans to speak may well have been present in the common ancestor of the great apes. These researchers highlight the tool-using capabilities of chimpanzees, which resemble those of human children. It is not, in their view, inconceivable that other traits regarded as characteristically human, including language, first appeared in pre-human apes.
Despite Pinker, the only way to resolve this issue is by looking for clues in the genus Pan, which consists of the common chimpanzee (Pan troglodytes) and the bonobo chimpanzee (Pan paniscus) - although, as Pinker points out, researchers should not assume that they will find clues in these species. These two species are humanity's closest living relatives. Molecular evidence places the common ancestor of chimpanzees and humans some time between eight million and five million years ago (Waddell & Penny, 1996). The traits humans share with chimpanzees were most likely also present in our common ancestor.
Thus, if researchers were to find the primitive building blocks of language in chimpanzees, it would strongly suggest that these building blocks were also present in our common ancestor. If, on the other hand, researchers were not to find primitive protolinguistic structures in chimpanzees, then these structures most likely developed in humanity's more recent ancestors. Researchers have looked for these structures in chimpanzees in two main ways: by examining chimpanzees' neurophysiology and by studying their communication skills.
Two regions of the human brain are known to play important roles in language production and processing: Broca's area and Wernicke's area. These areas were identified in the nineteenth century through the study of brain damage patients. Patients with damage to Broca's area exhibit an inability to produce grammatical sentences, although they can use single words properly. Patients with damage to Wernicke's area, in contrast, produce well-formed but meaningless sentences. At least 96% of patients exhibiting the symptoms of Broca's aphasia or Wernicke's aphasia have damage in the left hemisphere of their neocortex. The parts of the brain believed to compose these areas, which are located in the left hemisphere, are larger than the corresponding structures in the right hemisphere (Geschwind, 1965).
This asymmetry is not, however, unique to human beings. Gannon et al. (1998) discovered a parallel asymmetry in a homologous region of the chimpanzee brain homologous to Wernicke's area. In 94% of the common chimpanzees they examined, the left planum temporale - which in humans is part of Wernicke's area - was significantly enlarged compared to the right planum temporale. This suggests that the neural asymmetry believed to be related to language use in humans was already present in the common ancestor of chimpanzees and humans.
Whether this ape used the left hemisphere of the neocortex for communications, however, is not clear, as the question of which areas of the brain great apes use for communication has not been extensively studied. This issue has, however, been examined in other primates. Research conducted with monkeys and gibbons has shown that the neocortex does not play a driving role in their gesture-call systems. These appear instead to be rooted in the midbrain, as electrical stimulation of specific regions of the midbrain lead to the production of specific calls in these species. Electrical stimulation of the neocortex, however, produces no such response. It is possible that in the species the neocortex does plays a role in inhibiting vocalizations, as the neocortex is known to have a major role in controlling social behavior. In addition, stimulating the human neocortex interrupts speech (Deacon, 1989).
The converse of the question of the role the neocortex plays in non-human primate communication is the question of the role the midbrain plays in human communication. Although the development of language may have pushed the human gesture-call system into a secondary role, humans nevertheless retain such a system. Laughter and sobbing are examples of Homo sapiens' calls, while facial signaling is another form of communication driven by the midbrain. Burling (1993) suggests that human language and the human gesture-call system can be distinguished on the basis of discreteness. Language consists of discrete words, while human calls fall along a continuum that stretches from laughter to giggling to snorts to cries to sobs. Those signals that are part of the human gesture-call system are also more difficult to consciously control than language. Using this criteria, Burling contends that the intonation of words is also part of the human gesture-call system.
Deacon (1989) notes that the human gesture-call system is more limited in the range of vocalizations it can produce than are those of other primates. This leads him to propose that, over the course of evolution of Homo erectus and Homo habilis from about 2 million years ago to 750,000 years ago, the control of vocalizations shifted forward from the midbrain to the neocortex and thus expanded an emerging language facility at the expense of the gesture-call system.
Studies that have attempted to teach limited forms of human language to other great apes, however, suggest that the language facility may have developed earlier. Many of these highly mediagenic studies are well-known to the general public: the story of Koko, a gorilla who was taught a small subset of American Sign Language (ASL), is the subject of a children's book. Nim Chimpsky, a chimpanzee to whom workers attempted to teach ASL, is also fairly well-known. But perhaps the most promising great ape language studies were those conducted by Savage-Rambaugh that centered first on a common chimpanzee called Lana and later on a bonobo chimpanzee called Kanzi.
Rather than attempting to teach her chimpanzees ASL, Savage-Rambaugh attempted to teach them to use a keyboard with large symbols printed on it. This approach proved moderately successful in the 1970's, when Lana, along with two other common chimpanzees, were taught to use the symbols in stock phrases. Savage-Rambaugh therefore expanded her project to include a female bonobo named Matata, who at the time the project started had a six-month old child named Kanzi (Savage-Rambaugh et al., 1998).
The common chimpanzees had learned to use about fifty symbols, but after two years of instruction Matata had difficulty with a twelve symbol keyboard. Savage-Rambaugh had from time to time examined Kanzi's language skills as well, but Kanzi was not particularly cooperative with his mother around. When Kanzi was two and a half years old, however, Matata was taken away one day for breeding. To Savage-Rambaugh's surprise, Kanzi exhibited a surprising level of skill with the keyboard and was able produced 120 different phrases using the twelve symbols. Kanzi also become more cooperative in other respects; for instance, he no longer protested using the toilet. With the surprise discovery that Kanzi had acquired the ability to use the keyboard simply by being around his mother during her lesson, Kanzi became the new focus of Savage-Rambaugh's study.
Based upon her studies of language development in human children, Greenfield (1991) has identified several stages of symbol combination in human protolanguage. In the first stage, a child makes simple, one-word utterances. Around fifteen months, children begin to compose two-word utterances, such as "more cracker." In the next phase, children begin to make simple three-word utterances with regular protosyntactic structures, before moving on to three-word utterances that include noun phrases, such as "want more cracker".
Kanzi passed through similar stages in his use of the keyboard. Before his separation from his mother, he made only one-word requests for specific food items. On the day of his mother's departure, however, he combined symbols to make phrases in which he requested multiple items, such as "raisin peanut." Later he began to make simple three-word utterances, such as "chase hide you" to communicate multiple requests. Following a protosyntax, Kanzi consistently composed these phrases from two sequential actions followed by an agent.
By the time he was three and a half years old, Kanzi had, in a fashion similar to human children, begun using composite noun phrases. For instance, he composed the phrase "balloon water hide" after he and his caretakers had been hiding balloons filled with Kool-Aid. Although he did not often use such utterance, when he did they followed this protosyntactic form. Unlike human children, however, Kanzi's linguistic skills did not continue to develop beyond this point; while his vocabulary did grow, the complexity of his utterances did not.
Savage-Rambaugh (1998) reports that, in addition to composing sentences using the symbols on a keyboard, Kanzi also learned to understand sentences spoken by a human in English. When Kanzi was nine years old, Savage-Rambaugh performed a test to compare Kanzi's comprehension abilities to that of a two and a half year old human child named Alia. This test measured only the ability to comprehend requests, framed either as imperatives or questions, as the only standard Savage-Rambaugh had available was the subject's compliance with the request. More abstract sentences provided no ready scale for measuring comprehension. In order to prevent herself from subconsciously cuing the subject, Savage-Rambaugh sat behind a one-way mirror when she issued her requests. In addition, she avoided making requests similar to those she made during Kanzi's normal routine.
Savage-Rambaugh found that Kanzi correctly carried out 72 percent of the requests she made. Alia performed similarly and correctly carried out 66 percent of the requests. Both subjects were able to understand grammatically complex sentences such as "Show me the ball that's on TV" and "Get the ball that's in the cereal." These sentences require the subject to be able to construct the proper relationship among multiple objects. Kanzi had a 77 percent success rate on such sentences, while Alia had a 52 percent success rate.
Kanzi did have problems resolving some forms of lexical ambiguity. Savage-Rambaugh notes particularly that Kanzi could not properly decipher the use of the word 'can' when it had multiple meanings in a sentence or when it was used as a modifier. For instance, he could not properly understand the sentence "Can you use the can opener to open a can of Coke?", although he could understand "Can you put the chicken in the potty?"
Kanzi also had difficulty with sentences that required him to retain two unlinked objects in memory. He performed correctly on only 33 percent of sentences with the form "Show someone X and Y.' When Kanzi failed to retrieve one of the named objects, however, Savage-Rumbaugh needed only to mention the name of the forgotten object in order for him to remember it; she did not need to phrase a complete request.
Kanzi's ability to communicate with a protolanguage need not invalidate Deacon's theory that control of vocalizations has shifted forward from the midbrain to the neocortex during human evolution. No chimpanzee, after all, has been successfully taught to speak verbally. The regions of the neocortex associated with language in human beings may have been used for primitive symbolic processing before they were used for speech.
Greenfield (1991) suggests that some of the same regions in the brain used in language processing are also used in object combination and tool use, which has been observed in wild chimpanzees. Thus, as Burling (1993) proposes, language may have evolved as a tool for thought and later been adapted as a tool for communications. If one accepts Savage-Rumbaugh's studies, then the common ancestor of chimpanzees and humans most likely possessed symbolic processing capabilities that were adapted for vocal communications in Homo sapiens' more immediate ancestors.
One must, however, be cautious with primate language studies. It is important to note, when examining such studies, that no non-human great ape exhibits any of the signs of language use in the wild. In humans, the early development of protolanguage is accompanied by the development of a 'gestural complex' that allows an child to indicate an object by pointing and signal his requests. This behavior has been observed in great apes who have been taught language, but has not been observed in the wild (Parker & Gibson, 1979). Thus, if Savage-Rumbaugh is correct in concluding that chimpanzees have the capacity to use simple language, then this capacity is probably not employed outside of captivity.
More severe, systematic flaws may also be associated with great ape language studies. Whenever the only evidence available for a hypothesis comes from a small number of case studies, one must be cautious in reaching conclusions. Particularly when dealing with primates, researchers can become emotionally attached to their subjects. It is possible that this emotional attachment can lead them to conclusions a more objective observer would not make. Seidenberg and Petitto (1987) suggest that Kanzi may use the keyboard symbols as instruments to obtain food or other objects without understanding that they correspond to classes of objects; Savage-Rumbaugh's affection for Kanzi may have led her to argue that Kanzi is, in fact, employing a protolanguage.
In addition, if chimpanzees like Kanzi can communicate symbolically, workers need to confirm that the protolanguage they use is homologous to the protolanguage used by human children. This could best be done by using modern brain imaging techniques to determine what regions of the brain a chimpanzee like Kanzi uses when he or she communicates with a keyboard or processes human speech. If this activity takes place in the left hemisphere of the neocortex, it would strongly support the hypothesis that chimpanzees do have a capacity for human-like protolanguage. The failure of brain scans to reveal this would not necessarily imply chimpanzees lack this ability. Little research has been done on which regions of the brain human children use when speaking protolanguage; this area also should be examined.
Burling, R. (1993). Primate calls, human language and nonverbal communications. Current Anthropology, 34, 25-53.
Deacon, T.W. (1989). The neural circuitry underlying primate calls and human language. Human evolution, 4, 367-401.
Dennett, D. (1995). Darwin's dangerous idea. New York: Simon & Schuster.
Gannon, P., Holloway, R., Broadfield, D., & Braun, A. Asymmetry of chimpanzee planum temporale: Humanlike pattern of Wernicke's brain language area homolog. Science, 279, 220-222.
Geschwind, N. (1965). The organization of language and the brain. Science, 170, 940-944.
Greenfield, P. (1991). Language, tools and brain: The ontogeny and phylogeny of hierarchically organization sequential behavior. Behavioral and Brain Sciences, 14, 531-595.
Parker, S.T., & Gibson, K.R. (1979). A developmental model for the evolution of language and intelligence in early hominids. Behavioral and Brain Sciences, 2, 367-408.
Pinker, S. (1994). The language instinct. New York: William Morrow and Company.
Pinker, S., & Bloom, P. (1990). Natural language and natural selection. Behavioral and Brain Sciences, 13, 707-784.
Savage-Rumbaugh, S., Shanker, S., & Taylor, T. (1998). Apes, language and the human mind. New York: Oxford University Press.
Seidenberg, M. & Petitto, T. (1987). Communication, symbolic communication and language: Comments on Savage-Rumbaugh, McDonald, Sevick, Hopkins, and Rupert (1986). Journal of Experimental Psychology: General, 116, 279-287.
Waddell, P., & Penny, D. (1996). Evolutionary trees of apes and humans from DNA sequences. In A. Lock & C. Peters (Eds.), Handbook of human symbolic evolution (pp. 53-73). New York: Oxford University Press.
Copyright © 2000 by Robert Kopp III. All rights reserved.
| Bob Kopp <r-kopp@uchicago.edu> | Last Updated: 4 December 02000 |