THE INFLUENCE OF THE HOME-RAISING EXPERIMENTS

With the beginning of the twentieth century and the application of scientific principles to psychological developments and discoveries, an innovative type of animal-training experiment developed. Using an evolutionarily close relative of man, the chimpanzee, the experimenters attempted to duplicate the environment of a human household, the conditions of which are most favorable to language acquisition in man. Because the requirements for human language were obviously met in such a setting, i.e., adequate social atmosphere, sufficient periods in which to practice babbling, and an appropriate model for vocal imitation, it was assumed that the chimpanzee would acquire vocal language. Lightner Witmer (1909) summarized the rationale for this type of experiment:

While my tests of Peter give no positive assurance that he can acquire language, on the other hand they yield no proof that he cannot. If Peter had a human face and were brought to me as a backward child and this child responded to my tests as credibly as Peter did, I should unhesitatingly say that I could teach him to speak, to write and to read within a year's time.Peter has not a human form, and what limitations his ape's brain may disclose after a persistent effort to educate him, it is impossible to foretell. His behavior, however, is sufficiently intelligent to make this educational experiment well worth the expenditure of time and effort.

In general, the rationale for the home-raising experiments that followed were based on the above assumptions. If the necessary languageeliciting environment were provided, perhaps the puzzle of nonhuman primate language could be solved.

As pointed out by Kellogg (1968), keeping nonhuman primates as pets in the home is certainly not a novel idea and can be traced back several centuries. Such practices frequently occur today, but no instances of intellectual language use by such pets have been reported (Yerkes, 1925; Kellogg, 1968). A major problem is that the great majority of those who have nonhuman primates in the home are not familiar with language training and are ill equipped to observe and record the animals' reactions. However, as Kellogg (1968) observes, "It is quite another story for trained and qualified psychobiologists to observe and measure the reactions of a home-raised pongid amid controlled experimental home surroundings." We will describe several home-raising experiments, all of which were designed to determine the extent to which a nonhuman primate could acquire a vocal language capability.

Peter

The first chimpanzee to be mentioned was named Peter and was observed by L. Witmer (1909). Although this study was not a home-raising experiment, the rationale and desire to undertake such a task with an ape was inspired by this and similar reports found in the literature during this period.

Peter was a 4to 6-year-old chimpanzee owned by a man named McArdles, who had trained him for 2½ years. Employed by Keiths Theatre in Philadelphia as an example of "a monkey who made a man of himself," the entertaining, humanlike chimpanzee aroused the curosity of Witmer, who made arrangements to test Peter to determine the extent of his intelligence. The tests were conducted in the fall of 1909 at the Psychological Clinic in Philadelphia. Most of them involved motor coordination and simple reasoning tasks—opening a box to obtain articles inside, unlocking locks with keys, and driving nails into a board with a hammer—all of which Peter was able to do with relative ease. It was observed that Peter showed only imitative writing movements and possessed no special writing ability. He was able to articulate the word "mama," however; it is noted that he did so with considerable effort and with apparent unwillingness. The articulation of the sound "m" was said to be perfect, but the second "ma" in "mama" sounded more like "ah" and was inaudible more often than not. Witmer noted that the sound was a hoarse whisper rather than an articulated word and that Peter always tried to speak with the inspired rather than the expired breath. Peter was trained to articulate the sound "p" in only a few minutes, leading Witmer to conclude:

This experiment was enough to convince me that Peter can be taught to articulate a number of consonantal sounds and probably to voice correctly some of the vowels. ... If a child without language were brought to me and on the first trial had learned to articulate the sound "p" as readily as Peter did, I should express the opinion that he could be taught most of the elements of articulate language within six months' time.

Witmer also observed that although Peter was unable to speak, he was nonetheless able to understand spoken words. Using the analogy of Helen Keller, who first comprehended the use of symbols in the place of objects, Witmer proposed that in a similar manner Peter could be made to comprehend symbols as representing objects; and with further training to articulate these symbols, he would be able to communicate. Recognizing that early language training would be crucial in the event that vocal language could be taught to the chimpanzee, Witmer predicted that "within a few years, chimpanzees will be taken early in life and subjected for purposes of scientific investigation to a course of procedure more closely resembling that which is accorded the human child." Clearly a precursor for things to come, Witmer's prediction has been attempted on several occasions.

Joni

Joni (Kohts, 1935) was a male chimpanzee raised and observed by N. Kohts and her family from 1½ to 4 years of age. The period of observation was from 1913 to 1916, and the daily events pertaining to Joni's behavior were recorded and later compared with those of Kohts's own human child, Roody, during the period 1925 to 1929. Kohts's manuscript was prepared in the early 1930s, fifteen years after the chimpanzee had been observed and tested. The comparison of the developmental sequences of the home-raised chimpanzee and of the child probably reflects the influence of Yerkes. Only a small part of the report is devoted to language capacity in the chimpanzee, and this is in direct comparison to that of the human child. No special attempts were made to train Joni to use articulate language, the author's purpose being to record the language capability emitted in the home environment without any special training. Kohts reported that Joni was able to produce at least twenty-five sounds elicited by various stimuli within his environment, but they were clearly his own natural sounds to express emotions and desires. Our own observations of chimpanzees are to a very great extent in agreement with Kohts's, in that vocalizations in chimpanzees seem to be elicited by the environment. (Fouts, 1973).

During the period of observation the chimpanzee never attempted to imitate the human voice or to express himself by other than his own characteristic utterances. The lack of vocal language communication was interpreted by Kohts to mean that the mental capability of the chimpanzee was qualitatively different from that of the human child. This was possibly in reaction to Yerkes, who had published his book Almost Human in 1927. Kohts responded by saying: "Not only is it impossible to say that he is 'almost human'; we must go even further and state quite definitely that he is 'by no means human.' " This conclusion was based mostly on language capability, for Kohts was able to observe many instances of human-chimpanzee commonality in play behavior, emotional expressions, conditioned reflexes, and "a few" intellectual processes, including curiosity, recognition, identification, and "sounds of an undifferentiated nature." It was concluded that the more biologically important the function the greater the capability of the ape to approach or surpass that of man. Conversely, the higher the intellectual process involved (Kohts considered language the highest) the more dominant man became over the apes.

Kohts's observation that there is a qualitative difference between humans and chimpanzees stems from her emphasis on vocal language capability rather than language capability regardless of mode. To view intellectual processes solely from the position of vocal language ability, in our opinion, limits the possible avenues for two-way communication with nonhuman primates. We conclude from reading Kohts's account that the study was undertaken with the expectation of establishing two-way vocal communication with the chimpanzee within a matter of months. When this expectation was not met but was later achieved with Roody, Kohts's interpretation was that the chimpanzee was qualitatively different from humans in mental capabilities. A valuable lesson learned here is that physical similarity does not necessarily mean total similarity. The chimpanzee may look and act like man but its mode of communication is not necessarily the same. Kohts's conclusion that a qualitative difference in intelligence exists between humans and chimpanzees because of differing communication modes is a common, prejudicial misjudgment.

Gua

The experiment involving the infant chimpanzee Gua evolved as a direct result of the influence of Witmer (1909), and Yerkes (1925), and Yerkes and Yerkes (1929), who had expressed opinions concerning the feasibility of raising an infant chimpanzee in the home. Gua was 7½ months old when she was obtained by W. N. Kellogg and L. A. Kellogg (1967) from the Yerkes Experimental Station in Orange Park, Florida. Gua lived in the Kelloggs' home for nine months and was afforded the same surroundings and treatments as their similarly aged son, Donald. During this time the Kelloggs were able to distinguish four naturally occurring sounds made by Gua. All of them seemed either to be elicited by the external environment or to be the result of the emotional state of the chimpanzee. The sounds—barking, food bark, screech or scream, and the "oo-oo" cry—are similar to those reported by other observers of chimpanzees (Yerkes, 1925; Yerkes and Yerkes, 1929; Kellogg, 1968). In attempting to teach the two syllable word "pa-pa," the Kelloggs noted that Gua showed considerable curiosity in the facial movements although she never tried to imitate the sounds. Upon being encouraged to do so by manipulation of the lips, Gua made occasional lip reactions but did not attempt to produce the sound. The Kelloggs proposed that if the chimpanzee ever progressed to actual articulation of human sounds it would be under training circumstances similar to those experienced by Gua. At the same time the Kelloggs were attempting to teach their son to articulate spoken words, also without success. Although Donald was clearly making gurgling and babbling noises, Gua was not able to make such sounds.

Viki

One of the most successful attempts to train a chimpanzee to speak was conducted at the Yerkes Laboratories of Primate Biology in Florida by Keith Hayes and Catherine Hayes (1952). They obtained a female chimpanzee only a few days after birth and tried to provide a background of experience resembling that of a human infant as closely as possible. The chimpanzee, Viki, lived in the Hayeses' home for six years and learned to produce four words, recognizable as "mama," "papa," "cup," and "up" (Hayes and Hayes, 1952). In agreement with an earlier report (Witmer, 1909), the Hayeses found the chimpanzee's vocal expression hoarse and seemingly difficult for the animal to produce. Viki's language training, which extended over several years, consisted of manipulating her mouth and lips and subsequently rewarding approximations to desired sounds. As training progressed, fewer manipulations were required to obtain the desired syllables. In this manner Viki was able to produce the words "mama" and "papa" on demand, although her difficulty in speaking was apparent. The words "cup" and "up" were more readily added to her vocabulary as they closely resemble sounds naturally produced by a chimpanzee. In his review of home-rearing experiments, Kellogg (1968) indicates that this project represents the acme of chimpanzee vocal achievement in human sounds. But it must be noted that even after these words were learned they were often inaudible and used incorrectly to identify objects.

Other Experiments

It is generally recognized that of all the animals, the chimpanzee most closely resembles man in physiology, intelligence, and imitative ability (Yerkes and Yerkes, 1929). Attempts at home-raising experimental animals other than chimpanzees have been reported, but none have produced any notable differences in vocal language capability from that of the chimpanzee. Furness (1916) attempted to train a young orangutan in a home-type experiment but after extended training was able to report only limited success. The orangutan was able to say "papa" and "cup," but these words were hoarse and were produced with considerable difficulty. These findings are consistent with later ones concerning vocalizations in chimpanzees (Hayes and Hayes, 1952). For the orangutan to produce the sound "papa," Furness found it necessary to manipulate the animal's lips with his fingers. "Cup" was pronounced with relatively greater ease, probably because it is a more naturally occurring sound for such primates, as Hayes and Hayes (1952) noted.

Toto, a gorilla, was raised in the home of A. M. Hoyt (1941) in Havana, Cuba. Notable comparisons between Toto and other home-raised pongids were recorded, but if the gorilla produced any humanlike sounds they were not reported by Hoyt.

Conclusions about Home-Raising Experiments

The relative lack of success in home-raising experiments designed to teach vocal speech to nonhuman primates has led a number of people to conclude that there exist qualitative differences between man and apes that constitute a definition of man. According to this line of reasoning, Noam Chomsky (1968) and other linguists imply that man is unique because he uses language, which provides a linguistic reservoir from which he can structure thought. By rearranging the linguistic symbols he can alter his thought; he can use the symbols in relation to tense; he can abstract thought; and, ultimately, by modifying his concepts he can produce novel combinations of symbols, which he can relate to other humans by the use of language. According to Chomsky this capability makes man unique, the only animal capable of creative thinking.

Chomsky's argument is to a great extent based on a traditional point of view. His assumption is that if an event cannot be observed it does not exist. We consider this to be reasoning based on "negative evidence." In the present instance it is erroneous to assume that language per se does not exist because vocal language is not exhibited to any great extent by chimpanzees. If all the researchers had assumed that language ability did not exist in chimpanzees, scientific attempts to investigate communication with nonhuman primates might well have ended with the home-raising experiments. Instead, the rationale for such experiments was changed because it was felt that attempts to communicate vocally were leading nowhere. Kellogg and Kellogg (1967) summed up, "We feel safe in predicting ... it is unlikely any anthropoid ape will ever be taught to say more than a half-dozen words, if indeed it should accomplish this remarkable feat."

The results of the home-raising experiments indicated two possibilities. The first was that vocal communication with lower primates on a significant level is impossible because it is beyond the capability of animals other than humans. The Hayes and Hayes (1951) study, which involved the most highly controlled and systematic attempt to teach vocal communication to a chimpanzee, has often been cited by those who take this position. The negative-evidence rule applies here: Those who hold this position see no communication; therefore the capacity for communication does not exist.

The second possibility (based on anatomical evidence recently confirmed by Lieberman, 1968) is that speech as commonly used by humans is not a suitable medium of communication for chimpanzees. Similarly, Gardner and Gardner (1971) have reported that the reason chimpanzees do not learn to speak is behavioral as well as anatomical. Some specific portions of an animal's behavioral repertoire are highly, and perhaps completely, resistant to modification; attempts to teach vocal language to chimpanzees have apparently failed because vocalization is resistant to modification (Gardner and Gardner, 1971).

Although attempts at vocal training were largely unsuccessful, chimpanzees use their limbs, particularly their hands, in a highly efficient and well-coordinated way (Witmer, 1909; Kohts, 1935; Hayes and Hayes, 1952; Riesen and Kinder, 1952). Having noted that a speciesspecific characteristic of using hand gestures to communicate either greeting or threat had been observed in both wild (van Lawick-Goodall, 1968) and captive chimpanzees (van Hooff, 1971), Gardner and Gardner (1971) proposed a method of communication using a form of gestural language. They selected the American Sign Language (Ameslan), which does not totally satisfy the linguistic criterion of language capability (Gardner and Gardner, 1971; Chomsky, 1968), but nonetheless provided an excellent means of determining linguistic abilities of chimpanzees. Ameslan was chosen because it is actually used as a language by a group of human beings (Gardner and Gardner, 1971). Widely employed by the deaf, it is composed of gestures, mainly executed by the fingers, hands, and arms, and it has specific movements and places where the signs begin and end in relation to the signer's body. The signs are analogous to words in a spoken language (Stokoe, Casterline, and Croneberg, 1965).

The results obtained by Gardner and Gardner with their chimpanzee, Washoe, indicated that their choice of language medium was a good one. By using the chimpanzee's natural ability to use gestures and her imitative and intellectual capacities, they were successful in teaching Washoe to use Ameslan. The Gardners provided evidence that a true two-way, nonvocal communication channel between man and chimpanzee can be established. Recently, using Ameslan, experimenters have begun to explore the intellectual capacities of chimpanzees (Mellgren, Fouts, and Lemmon, 1973; Fouts, Chown, and Goodin, 1973; Fouts, Mellgren, and Lemmon, 1973). This method of communication has begun to reveal conceptual processes in chimpanzees that were heretofore impossible to determine, and what has been accomplished shows that the nature of animal intelligence may at last be studied to an extent never before attempted.

Project Washoe

Project Washoe was the first successful attempt to teach a nonhuman primate in human language. The project was begun at the University of Nevada in Reno by R. A. Gardner and B. T. Gardner (1969, 1971) in June 1966 and was terminated in October 1970, by which time Washoe had acquired a vocabulary of over 130 signs. Although the number of signs is relatively small, Washoe was able to use her vocabulary very well. She could readily produce spontaneous combinations of signs that demonstrated her syntactic ability, and her combinations were contextually correct. She was also able to transfer her signs and combinations to novel situations with ease and with a high degree of reliability.

Washoe, an infant female chimpanzee estimated to be 8-14 months of age in June 1966, was obtained by the Gardners from a trader in the United States. It is assumed that she was born in the wild and was raised for several months by her natural mother before being captured. Washoe was raised in the Gardners' backyard, an area of 5,000 square feet. She lived in a completely self-contained house trailer (8x24 ft), which provided for her toilet, kitchen, and sleeping needs. Throughout the project her researchers used only Ameslan to communicate with Washoe, and, when in her presence, they used only Ameslan to communicate with one another.

McCarthy (1954) has indicated that the barren social and physical environments once common to institutions for the mentally retarded are not favorable conditions for the development of language in humans. Gardner and Gardner (1971) extended this hypothesis to the chimpanzee and claimed that the same could be said for the raising of chimpanzees in cages. Since one would expect less linguistic capability from chimpanzees raised under these conditions, Washoe was provided with an environment that was kept as interesting as possible. Her teaching program was made part of her environment and was maintained throughout her daily routine.

A member of the research team was with Washoe during all her waking hours. At the change of shift two researchers overlapped for an hour. She was often afforded additional companionship when visitors were present and during frequent outings in the nearby community. She played and climbed trees and playground equipment in the Gardners' backyard. The function of the researchers on the team was to keep Washoe totally immersed in Ameslan. Since their principal job was to be a conversing companion and a model for Washoe, they used Ameslan in normal daily routines, games, and general activities. They chatted with her in Ameslan while cooking meals, cleaning, brushing her teeth, playing with her in her sandbox, and correcting her lapses in toilet training.

METHODS OF ACQUISITION

Manual Babbling

A number of methods of acquiring the signs were examined. Manual babbling, which is considered analogous to vocal babbling in human infants, was infrequently observed early in the experiment, but as the project progressed, the occurrence of babbling increased. The increase continued until the end of the second year, after which time manual babbling was rarely observed. This decline is attributed to Washoe's progress in acquiring a vocabulary during this period, and it was concluded that the acceleration of signing may have replaced the babbling, just as babbling in humans decreases as vocal speech develops. Gardner and Gardner (1971) reported that the only sign attributable to this method of acquisition was the funny sign, which is produced by touching the nose with the index finger.

Shaping

Shaping procedures similar to those used in operant conditioning were used in teaching Washoe new signs. She was rewarded whenever she made an approximation to a sign in order to encourage her repeating it. Successively closer approximations were then rewarded, and in this manner several signs were acquired. For example, originally Washoe would bang on doors with her fists when she wanted the door to be opened. This natural response was shaped using rewards for successive approximations to the correct sign. The sign for open consists of placing the two open palms against the object to be opened and then moving them up and apart. Washoe quickly acquired the open sign and soon generalized it spontaneously from doors to other objects such as books, briefcases, boxes, and drawers. Although shaping procedure was used a great deal during the first year to introduce new signs, it soon became apparent to the Gardners that this technique was not as efficient as other methods.

Guidance

The Gardners considered guidance the most effective method of teaching a new sign to Washoe. It consists of physically molding the hands and arms in the appropriate position for the sign, usually in the presence of an object or action that represents the sign. An example is given by Gardner and Gardner (1971):

The sixth sign that Washoe acquired was the sign tickle. It is made by holding one hand open with fingers together, palm down, and drawing the extended index finger of the other hand across the back of the first hand. We introduced this sign by holding Washoe's hand in ours, forming her hands with ours, putting her hands through the required movement, and then tickling her.

Molding

A more recent investigation into the optimal training method (Fouts, 1972) utilized a design intended to test three procedures: molding, which involved physically guiding Washoe's hands into the correct position and movement for the sign; imitation, which consisted of the experimenter's making the sign and Washoe's being required to imitate his example; and free style, which was a combination of molding and imitation. Under the conditions of the experiment it was found that the molding procedure produced the most rapid acquisition of signs, followed by free style (which is not significantly different from molding), and lastly by imitation. Although imitation showed the poorest performance in the experiment, it nonetheless resulted in the acquisition of signs.

Observational Learning

Gardner and Gardner have reported that during the project Washoe apparently learned to comprehend Ameslan as practiced by her researchers although no overt efforts were made to teach her the signs. Since it was the researchers' practice to converse in Ameslan in Washoe's presence, the Gardners called this process of learning "observational learning." It is analogous to Fouts's (1972) use of imitation, although in the Fouts experiment a deliberate effort was made to train the chimpanzee to produce a particular sign. For example, Washoe learned the sign for sweet (which is produced by touching the lower lip or the tongue with the extended index and second fingers of one hand while the remaining fingers are pressed into the palm) merely by observing her trainer. Flower, which Washoe at first signed incorrectly, was later corrected by observing how the researchers made the sign. Gardner and Gardner (1971) indicate that because Washoe was totally immersed in an environment of sign language, she often acquired signs after several months' exposure to them without any concerted effort on the part of the researchers to teach them to her. It would appear that she spontaneously acquired the signs.

RECORDING WASHOE'S SIGNS

The various signs given by Washoe were recorded each day. A major portion of the recording procedure was concerned with whether the signs were spontaneous or prompted. Spontaneous signs were noted when Washoe made a correct signing response in answer to a question or made one entirely on her own, such as open. Prompted signs were those for which a correct response required assistance from one of the members of the research team. Additionally, information concerning the correctness of the form of the response was recorded. The criterion for reliability of responding was whether Washoe could produce the response spontaneously and correctly on fifteen consecutive days. In addition to keeping a signing record, the researchers kept a daily diary of the uses Washoe made of various signs and combinations and the contexts in which she used them. Later in the project a daily tape recording was added to the recording procedures. One of the researchers observed the team in a training session and verbally recorded the training procedures and Washoe's responses using a whisper microphone.

The first thirty-six months of the project yielded notable results in Washoe's ability to acquire signs. By the end of this period she was using 85 signs reliably. By the time of this writing (June 1974), she was using over 160 signs reliably. A cumbersome recording process is required in a project such as this; therefore it should be stressed that the reported size of Washoe's vocabulary is limited more by the researchers' ability to handle the recording and testing of the vocabulary than by Washoe's ability to acquire signs.

Testing Procedures

Because the approach in this project was at variance with any previous attempts, new ways of testing had to be designed in order to quantify the accumulated data accurately. Several methods were tried before a test was found that satisfied the requirements of both the researchers and the chimpanzee. A double-blind procedure (a procedure in which the observer evaluating the behavior of the subject does not know what treatment the subject has received until after the evaluation is completed) was used to control for the possibility that the experimenter might cue the subject as to the correct answer.

The first test used flash cards. Washoe was shown pictures on large cards and then questioned about the cards. This test had some drawbacks, the main one being that the test was experimenter-paced and so required an excessive amount of discipline. The box test was then devised. Washoe was to identify three-dimensional objects placed in a box by the researcher. Although this test was a great improvement over the flash-card technique, it was logistically difficult to conduct. It was replaced by the slide test, which was similar to the box test except that 35 mm color transparencies were used as exemplars. The slide test was efficient and easy to administer, and it differed from the other two tests in that it could be paced by Washoe.

The Gardners report that in the initial slide test Washoe correctly identified 53 items out of a possible 99. This performance was considerably above the chance level, which was 3 correct responses out of a possible 99. Although the correct responses were obviously encouraging to the researchers, Washoe's errors produced equally interesting data in that the errors fit into meaningful conceptual categories. Conceptual categories are such things as animals, foods, or grooming articles. Instances of responding to conceptual categories occurred when Washoe signed dog in response to a picture of a cat, brush for a picture of a comb, and food for a picture of meat. Also it was reported that when pictures of three-dimensional replicas of objects (e.g., toy cats, toy dogs) were used in the box test, the baby sign occurred frequently among Washoe's errors. When the slides were of real items, the baby error did not occur; however, the baby error continued to occur when a picture of a replica was used. For example, when a slide of a doll resembling a cat was shown, Washoe made four errors out of ten trials, and all four were the baby sign. However, when a photograph of a real cat was shown, she correctly identified and signed seven out of eight trials, and the single error was not the baby sign.

One important criterion of language use is that words must be used with others to form phrases, or more appropriately, combinations. Washoe first signed the phrase gimme sweet and come open during the tenth month of the project. At this time she was approximately 18 to 24 months old (Gardner and Gardner, 1971). It is interesting to note that similar instances of language use appear in humans at approximately the same age. Fouts (1973) has recently found that the chimpanzee is capable of forming combinations at a much earlier age. The discrepancy between the new findings and those reported by Gardner and Gardner (1971) is most likely due to the comparatively late start in training Washoe to use Ameslan (8-14 months), whereas in Fouts's research the teaching of Ameslan was begun at a much earlier age. In a more recent project by Gardner and Gardner (pers. comm.), two young chimpanzees acquired their first signs at age 3 months. Schlesinger and Meadows (1972) have found a correspondingly early acquisition of signs in humans, in which deaf children exhibited the use of signs at 5 months of age. The two-month difference is probably not due to qualitative differences in intelligence but to comparatively faster motor-coordination development in chimpanzees (Kellogg and Kellogg, 1967).

The segmentation of combinations by Washoe was done in much the same manner as executed by human signers. Washoe would keep her hands raised in the signing space until she had completed the combination. She would terminate the combination by making contact with some object or surface, comparable to the human signers' hands in repose.

The contextual relevance in which Washoe used her signs was found to be very good. She would correctly sign intended destinations with phrases such as go in, go out, or in down bed. When playing with members of the research team she did not generalize but referred to each one by name, signing Roger you tickle, you Greg peekaboo (Gardner and Gardner, 1971). Washoe signed at a locked door on thirteen separate occasions, and each time her contextual relevance was appropriate and correct: gimme key, more key, gimme key more, open key, key open, open more, more open, key in, open key please, open gimme key, in open help, help key in, and open key help hurry (1971:167).

The Gardners analyzed Washoe's two-sign combinations according to a method proposed by Brown (1970) for use with children. It was found that her earlier combinations were comparable to the earliest two-word combinations of children in terms of expressed meanings and semantic classes. Longer combinations were often formed by Washoe by adding appeal signs, such as please and come, to shorter combinations. Between April 1967 and June 1969, 245 different combinations of three or more signs were recorded in the researchers' diary. About half were formed by the addition of an appeal sign. The remaining ones were introduced by new information and relationships among signs, such as pronouns or proper names.

In analyzing Washoe's combinations the Gardners found evidence that possibly indicates that Washoe had specific preferences for word order. The combination you me was preferred in over 90 percent of the samples taken, with me you used in the remaining 10 percent. Earlier instances of this combination showed a preference for you-me-action, but later this changed to a preference for you-action-me order. The Gardners were reluctant to accept this order as an indication of syntax in Washoe's manual language. They point out that it may merely be her imitation of the preferred order of the members of the research team. From a behavioral viewpoint, however, there appears to be little difference between Washoe's preferences in word order and language behavior in human children in learning syntax.

Fouts (pers. obs.) has also noted the use of sign order to express meaning in signing chimpanzees. One of his chimpanzees, Lucy, has a definite preference for the sign order tickle Lucy when asking someone to tickle her; and when the order is reversed to Lucy tickle, she correspondingly tickles her companion.

Washoe and the Oklahoma Chimpanzees

Project Washoe was terminated in October 1970, when several members of the research team were receiving their degrees and leaving the project. Washoe was then brought to the Institute of Primate Studies in Norman, Oklahoma, directed by W. B. Lemmon. The Institute has many chimpanzees in its main colony, and since Washoe's arrival, it has been directing research with chimpanzees reared in private homes around the area. In both the colony proper and the private homes, the animals have been taught to use Ameslan.

One of the first experiments using Ameslan conducted in Oklahoma (Fouts, 1973) was designed to determine the relative ease or difficulty of acquiring signs by four young chimpanzees, two males and two females. An interesting finding was that just as in humans, chimpanzees have different rates of acquiring signs. By means of a molding procedure (Fouts, 1972), a total of ten signs were taught to the chimpanzees in daily thirty-minute training sessions. The acquisition rate for each sign was compared on the basis of the number of minutes of training necessary to reach five consecutive unprompted responses. After the chimpanzees had acquired the ten signs, they were tested on nine of them (all nouns), with the sign for more (an adjective) excluded from the test. Testing was conducted using the double-blind box test procedure, similar to the test described by Gardner and Gardner (1971).

The results indicated that some of the signs were consistently easy or difficult for the chimpanzees to acquire. The mean times for acquiring the signs ranged from 9.75 minutes to 316 minutes. The mean times to reach criteria for each chimpanzee across signs were: 54.3, 79.7, 136.4, and 159.1 minutes. These differences may be partially due to the individual chimpanzees' behavior in the training sessions.

All the chimpanzees performed above the chance level during testing. The correct responses in the double-blind box were: 26.4%, 58.3%, 57.7% and 90.3% correct. The low score of 26.4% obtained by one of the chimpanzees may have been a result of the difference between acquisition and testing. This particular chimpanzee seemed to require much praise and positive feedback from the experimenter for her correct responses when acquiring the signs. However, in the double-blind testing situation the observers who recorded her scores were unable to give her any positive feedback, and, as a result, her performance would begin to deteriorate noticeably after the initial trials in the test were completed. Another important finding was that Washoe was not the only chimpanzee that had the capacity to use Ameslan.

Mellgren, Fouts, and Lemmon (1973) examined the conceptual ability of a chimpanzee in regard to the category of items by studying the relationship between generic and specific signs in Ameslan. The subject in this experiment was Lucy, a seven-year-old chimpanzee that had been raised in species-isolation (by humans without ever seeing another chimpanzee) in a human home since she was two days of age. She had been taught Ameslan for two years and had a vocabulary of seventy-five signs. Our objective in the experiment was to determine if a new sign would become generic or specific relative to a category of items. Lucy had previously learned five food-related signs: food, fruit, and drink, which she used in a generic manner; and candy and banana, which she used specifically. The sign we chose to teach her was berry, and the category of items consisted of twenty-four different fruits and vegetables, ranging from a quarter of a watermelon and a grapefruit to small berries and berrylike items such as blueberries, cherry tomatoes, and radishes. The exemplar for the berry sign was a cherry.

The twenty-four different fruits and vegetables were presented to Lucy in a vocabulary drill, and her responses were recorded. As each item was presented she was asked what that? She was allowed to handle the various foods and eat them if she wished. The order of presentation was varied from day to day. The food-related items were presented along with at least two other items that were not in the fruit and vegetable category but were items for which she had a sign in her vocabulary. For example, she would be asked to identify a shoe, a string, and then a fruit or vegetable. Following her response to the food items she was questioned about such things as a book or a doll. For the first four days of training data were collected to obtain a systematic baseline of her responding to the twenty-four items to determine her usual response to them. On the fifth day she was taught the berry sign, using the cherry as the exemplar, and for the second four days the berry sign remained entirely specific to cherries. On the ninth day she was taught the berry sign again, but this time blueberries were used as the exemplar. On the ninth and tenth days she called the blueberries berry, but on the eleventh and twelfth days she switched back to what she had previously called them; but throughout these four days she persisted in calling cherries the sign she had originally been taught for them, berry. By her responses it was apparent that she preferred to use the berry sign in a specific sense.

Lucy's conceptualizations of fruits and vegetables were also examined. She showed a preference for labeling the fruit items with the fruit sign in 85% of the trials and for using the food sign in 15% of the trials. A dichotomy of responding to the two categories was indicated because she preferred to refer to vegetables as food 65% of the time and as fruit 35% of the time.

In a very revealing finding, Lucy created novel combinations to describe her perception of the stimulus item. She preferred to call a watermelon candy drink or drink fruit, whereas the experimenters referred to it with entirely different signs, which Lucy did not have in her vocabulary (water and melon). Another, more striking example occurred with radishes. For the first three days of the experiment Lucy labeled them fruit food or drink. On the fourth day she bit into a radish, spat it out, and called it cry hurt food. She continued to use cry and hurt to describe the radish for the next eight days. Sixty-five percent of the smell signs were used to describe the four citrus fruits by labeling them smell fruit, probably referring to the odor released when one bites into the skin of citrus fruits. The spontaneous generation of novel combinations not only demonstrated Lucy's ability to form new combinations but also indicated her ability to use her existing vocabulary of signs to map various concepts she had about the categories of the fruits and vegetables she was presented with.

A good understanding of vocal English appears to exist in a number of the home-reared chimpanzees near the Institute. Some of them were exposed to vocal language before being taught Ameslan. To determine the relationship between their English vocabularies and their Ameslan vocabularies, a study was undertaken by Fouts, Chown, and Goodwin (1973) using Ally, a young male chimpanzee that was being home-reared. A pretraining test was administered to determine his understanding of ten vocal English words. He was given such vocal commands as "Give me the spoon," "Pick up the spoon," "Find the spoon." Ally had to obey the command by choosing the correct item from a group containing several other objects. Only after correctly obeying the command five consecutive times was he considered to have an understanding of the vocal English word. Training was begun by dividing the ten signs into two lists of five signs each. One experimenter attempted to teach Ally a sign using only the vocal English word as the exemplar. Following this a second experimenter, who did not know which words had been taught or if any had been acquired, would test Ally on all five, using the objects corresponding to the vocal English words. For the second list of five words the experimenters exchanged roles. Ally was able to transfer the sign he was taught to use for the vocal English word to the object representing that word. This finding is very similar to the acquisition of a second language in humans. A second possible implication is that the learning occurred via cross-modal means.

The initial research in gestural language ability in chimpanzees indicates that chimpanzees can produce novel combinations of signs in their existing vocabularies. Humans have this capacity also but in addition they are able to understand novel combinations produced by someone else. A recently completed experiment (Chown, Fouts, and Goodin, 1974) seems to indicate that chimpanzees are able to understand novel combinations when they are used as a command. In the first phase of the experiment Ally was taught to pick out one of five objects in a box and place it in one of three places. For example, a command might be Put baby in purse. When Ally was sufficiently adept at this, the second phase of the experiment began. New items that had not been used in training were placed in the box and a new place to put the object was added. A screen between the experimenter and Ally prevented the experimenter from giving helping cues. With five items to choose from and three places to put them, chance would produce one correct response in fifteen trials. But Ally's performance was far above the chance level. During one test session he responded correctly to 22 of 36 commands.

CHIMPANZEE-TO-CHIMPANZEE COMMUNICATION USING AMESLAN

Because a major criterion of language is that it be used by members of the same species, we have been interested in intraspecific communication using Ameslan in chimpanzees. Fouts, Mellgren, and Lemmon (1973) explored various conditions under which such communication might be observed. The experimental subjects were Booee and Bruno, two young male chimpanzees who had already acquired a vocabulary of 36 signs each. The experimenters intended to keep the chimpanzees' vocabularies small so that they could examine the animals' acquisition of new signs when Washoe was introduced into the signing dyad, but their vocabulary has already increased to over 40 signs. Activities such as tickling, play, mutual comforting, and mutual sharing are most conducive to communication in Ameslan between the chimpanzees. Booee and Bruno, however, seem to prefer their own natural communication over Ameslan, perhaps because of their relatively greater exposure to other chimpanzees. Washoe and Ally are quite different, and prefer to use Ameslan when communicating with humans or other chimpanzees. When Booee and Bruno reach a preset criterion of reliability in chimpanzee-to-chimpanzee communication, Washoe will be introduced into the dyad. In April 1974 Ally was introduced to Bruno, and a good deal of communication from Ally to Bruno has been observed and recorded, mostly signs referring to food or play.

Fig. 1. Bruno signing hat.

Fig. 2. Bruno signing look.

Fig. 3. Bruno demonstrates his ability to sign drink with either hand.

Fig. 4. Booee signing baby.

Manny, a young chimpanzee in the colony, has acquired from Washoe the come hug sign, which is used correctly when the chimpanzees greet one another or are engaged in mutual comforting. Kiko, a three-year-old chimpanzee acquired the food and drink signs from Booee and Bruno and displayed them correctly, but he was stricken with pneumonia and died. Another mature chimpanzee, housed in close proximity to Washoe, has often shown the food, drink, and fruit signs, but when offered a drink or a piece of fruit he has failed to show the signs again. Apparently he has failed to make the important connecting link between the sign and the object it represents.

One thing that makes research with chimpanzees so interesting is their ability to use Ameslan in situations other than experiments and at times when it is not expected. Both Lucy and Washoe have spontaneously invented signs for objects that were not in their vocabulary. Washoe invented a sign for bib by making the outline of a bib on her chest (Gardner and Gardner, 1971), and Lucy invented a sign for leash by making a hook with her index finger on her neck. Washoe was observed making a new combination, gimme rock berry. When the experimenter approached to question her about the seemingly incorrect sign, he found that Washoe was pointing to a box of brazil nuts on the other side of the room. On another occasion Washoe referred to a rhesus monkey (with whom she had earlier had a fight) as a dirty monkey. Until this time she had used the sign dirty to refer to feces or soiled items, usually as a noun. She now uses it regularly as an adjective to describe people who refuse her requests.

Sarah and Communication via Plastic Objects

Premack (1970, 1971a, 1971b) and Premack and Premack (1972) were the first to devise an artificial system for two-way communication between two species by using variously colored and shaped pieces of plastic to represent words. Sarah, the subject used in their research, is a wild-born female chimpanzee estimated to be six years of age when the project was begun. Since the Premacks' artificial language is visual and written they decided to use the chimpanzee because of the similarities between its visual system and man's.

Pieces of plastic that vary in size, shape, texture, and color are used to represent words. The plastic pieces are backed with metal and can be arranged in lines on a magnetized board. In this manner, the pieces can be displaced in space but not in time on the board, thus avoiding the necessity of relying on memory (which is an integral part of human languages, gestural or vocal). The use of pieces of plastic allows the human experimenter to determine which pieces of plastic will be available to the subject at any time. Also, the design of the experiment controls for the individual difficulty of any problem, but at the expense of the spontaneity of usage typically found in human language. It should also be pointed out that when an experiment is so overcontrolled, it limits the possible findings to those preconceived by the human experimenter rather than explores the full mental capacities of the subject.

The Premacks' approach emphasizes the functional aspects of language by breaking into behavioral constituents and providing environmental contingencies for each constituent they selected for training Sarah. They briefly summarize their work as follows: "We have been teaching Sarah to read and write with various shaped and colored pieces of plastic, each representing a word; Sarah has a vocabulary of about 130 terms that she uses with a reliability of between 75 and 80 percent" (Premack and Premack, 1972:92).

The Premacks appear to work from the premise that the relational and logical functions of language are derived from operant procedures, and therefore they use training procedures based on operant methodology. The procedure is of the same type as that used by psychologists to train pigeons or rats to peck keys or press bars. They reduce the constituents to very simple steps and then use standard operant techniques to train the subjects. For example, training may be started by placing some fruit on a board and allowing Sarah to eat it. Next, Sarah is required to place a piece of plastic representing the same fruit on a magnetized board before being allowed to eat the fruit.

As training continues in this manner, new pieces of plastic are added simultaneously with new aspects of the situation, e.g., a different kind of fruit or a new person. For example, a new piece of plastic may be added when Sarah's reward is changed from bananas to apples; or she may be required to place a piece of plastic representing the new trainer's name (e.g., Mary or Jim) on the board. In the next step a piece of plastic representing the word give is introduced, and Sarah is required to place it in between the trainer's name and the name of the fruit. Finally, a piece of plastic representing Sarah's name is introduced, and it has to be placed at the end of the sequence. Sarah appears to have little difficulty making these conditional discriminations. She may also be induced to respond to a sequence like "Sarah give apple Mary" by offering her a piece of chocolate if she gives up her apple. By using a more preferred item (chocolate) they are able to induce Sarah to place the pieces representing "Mary give apple Jim."

Premack and Premack report that Sarah is able to use and understand the negative article, the interrogative "wh" (who, what, why), the concept of name, dimensional classes, prepositions, hierarchically organized sentences, and the conditional.

Sarah has managed to learn a code, a simple language that nevertheless included some of the characteristic features of natural language. Each step of the training program was made as simple as possible. The objective was to reduce complex notions to a series of simple and highly learnable steps . . . compared with a two-year-old child Sarah holds her own in language ability. [Premack and Premack, 1972:99]

Lana and the Computer

Rumbaugh, Gill, and von Glasersfeld (1973) devised a computer-controlled training situation that objectively examined some of the language capacities in a chimpanzee. They used Lana, a 2½-year-old female chimpanzee, as their subject. After six months of training Lana was able to read projected word characters, complete an incomplete sentence based on its meaning and serial order, and reject an incomplete sentence that was grammatically incorrect.

Rumbaugh et al. are using a PDP-8 computer with two consoles containing twenty-five keys each. On each key is a lexigram in "Yerkish," an artificial language developed by the experimenters. The symbols are white geometric figures, created from nine stimulus elements, used singly or in combination. The keys, on which the symbols are displayed, have colored backgrounds made up of three colors used singly or in combination. When the key is available for use by Lana it is softly backlit. When Lana presses a key, it becomes brightly lit. When a key is not available for use, it has no backlighting. When Lana depresses a key, a facsimile of the lexigram appears in serial order on one of seven projectors above the console. The computer also dispenses appropriate incentives to Lana when she depresses the keys in the correct serial order in accordance with the grammar of Yerkish. She may ask for such things as food, liquids, music, movies, toys, to have a window opened, to have a trainer come in, and so on, when they are available (that is, when the appropriate keys are softly backlit). There is also a console available only to the human experimenters so that the computer can mediate conversations between the experimenters and Lana.

Lana's training was begun by requiring her to press a single key in order to receive an incentive. Next, she was required to begin each request with a please and end it with a period. The depression of the period key instructed the computer to evaluate the phrase for correctness of serial order. If it was correct a tone sounded and Lana would receive what she had asked for; if not the computer would erase the projector display and reset the keys on the console. Later, Lana was required to depress holophrases (e.g., machine gives M & M) in between a please and a period. Following this, she was taught to depress each key represented in the original holophrase (e.g., Please/Machine/give/M & M/period). Then the keys were randomized on the console and she had to select and press them in the correct serial order.

One very interesting finding was that Lana soon learned to attend to the lexigrams on the projectors without training. She would erase sentences in which she had made an error, by pressing the period key, rather than finish them. On the basis of Lana's spontaneously learning to attend to the projected lexigrams and their order, Rumbaugh et al. were able to examine her ability to read sentence beginnings, to discriminate between valid and invalid beginnings, and to complete sentences. In their first experiment Lana was presented with one valid sentence beginning (Please machine give) and six invalid beginnings. She could either erase them or complete them. If she completed them she had to choose from correct lexigrams (e.g. Juice, M & M, or piece of banana) and incorrect lexigrams (make, machine, music, Tim, movie, Lana). Music and movies were incorrect since the computer was programmed to accept these with make rather than with give. Lana's performance on the various aspects of this test ranged from 88% correct (please Xgive, an invalid beginning) to 100% correct (please machine give, the valid beginning). The second experiment was the same except that make was substituted for give in the sentences with the valid beginning. Lana's performance was 86% or more correct in this experiment. The third experiment used only valid beginnings with varying numbers of words in them: e.g.: (1) please, (2) please machine, (3) please machine give, (4) please machine give piece, (5) please machine give piece of, and so on. Lana ranged from 70% to 100% correct on the various beginnings.

Rumbaugh et al. concluded that Lana accurately read and perceived the serial order in Yerkish and was able to discriminate between valid and invalid beginnings of incomplete sentences in order to receive an incentive.

Comparison of the Methods

Using artificial languages or Ameslan avoids the problem of vocal communication. In one project using an artificial language (Rumbaugh et al., 1973) the human element has been removed by using a computer as an intermediary, and because a computer is used the experimenters are able to keep an exact record of the chimpanzees' communication. However, this refinement is also expensive in that the chimpanzee is forced into a strict and rigid paradigm that allows only that behavior to appear that will fit into the experimental situation. For example, the computer is not programmed to accept novel or innovative uses of the language. Although this method has managed to find and confirm conclusively such behavior as responsiveness to word order (syntax), in exploring the mental capacities of the chimpanzee, the artificial language approaches are limited to examining only behavior conceived of by the experimenters, and not responses created by the chimpanzees.

It is our contention that when conducting initial experimental research, such as we are doing with chimpanzee language acquisition, the situation must be structured only to the extent that control of the experiment remains in the hands of the experimenter. This point was made by Köhler (1921):

Lack of ambiguity in the experimental setup in the sense of an either-or has, to be sure, unfavorable as well as favorable consequences. The decisive explanations for the understanding of apes frequently arise from quite unforeseen kinds of behavior, for example, use of tools by the animals in ways very different from human beings. If we arrange all conditions in such a way that, so far as possible, the ape can only show the kinds of behavior in which we are interested in advance, or else nothing essential at all, then it will become less likely that the animal does the unexpected and thus teaches the observer something.

We feel the gestural language approach more closely fits the idea expressed by Köhler. Using this method we can simultaneously conduct highly controlled experiments and allow the chimpanzee to show, spontaneously, many of its capabilities. This is possible because the chimpanzee, not the experimenter, has the control of its language use. The chimpanzee can from within itself make statements or do what it wishes without having to rely on an arbitrary symbol to do so. It is not bound to a computer program or by the limits of the experimenter in making vocabulary available.

For example, in the study done with Lucy on the twenty-four different fruit and vegetables (Mellgren et al., 1973), had we limited her possible responses to the five food-related signs in her vocabulary, we would not have discovered her conceptualizations of the items. Nor would Washoe be able to insult people by calling them dirty if she had not been allowed to change a noun into an adjective. Similarly, Washoe would not be able to refer to brazil nuts as rock berry.

Each approach has its advantages, and it is up to the scientists to decide whether they wish to examine only those things they are capable of conceiving of, or if they are willing to accept some help from the chimpanzee in examining the animal's mental capacities.

Conclusions

The language skills of the chimpanzee are similar to those displayed by humans, although many definitions of language have attempted to exclude the chimpanzee from the realm of language as used by humans. An often-quoted, popular paper, Bronowski and Bellugi (1970) lists five characteristics of language: delay between stimulus and utterance, separation of affect from content, prolongation of reference, internalization, and reconstitution. Bronowski and Bellugi contended that Washoe probably met the first four, but failed to demonstrate the last. They define the structure activity of reconstitution as consisting "of two linked procedures—namely, a procedure of analysis, by which messages are not treated as inviolate wholes but are broken down into smaller parts, and a procedure of synthesis by which the parts are rearranged to form other messages" (1970:670).

Bronowski and Bellugi conclude: "What the example of Washoe shows in a profound way is that it is the process of total reconstitution which is the evolutionary hallmark of the human mind, and for which so far we have no evidence in the mind of the nonhuman primate, even when he is given the vocabulary ready made" (1970:673). We disagree with their contention of "no evidence." Most certainly the empirical evidence presented earlier in this chapter demonstrates that chimpanzees have the capacity for reconstitution, particularly Lucy's reference to a radish as cry hurt food and Washoe's calling a brazil nut rock berry. We contend that these gestural utterances more than meet the most restricting definitions of reconstitution and represent a remarkable intellectual and linguistic accomplishment, given the limited vocabulary we have allowed the chimpanzees to learn. And if Bronowski and Bellugi's contention that reconstitution is the "evolutionary hallmark of the human mind" is correct, then we must assume that the capacity for language was in the repertoire of the species before the great apes split off from hominoid evolution. Another alternative may be that the basis of language is not unique to language per se, but may actually be the basis of other behavior of which language is just one product.

References

Amman, J. C., 1692. Surdus Loquens. London: Sampson Low, Marsten Low and Seale.

Amman, J. C., 1700. Dissertatio de loquela. Reprint. Amsterdam: North-Holland Publishing Company, 1965.

Bronowski, M., and Bellugi, U., 1970. Language, name, and concept. Science, 168:669-73.

Brown, R. W., 1970. The first words of child and chimpanzee. In: Psycholinguistics: Selected Papers, R. W. Brown, ed. New York: The Free Press, pp.208-31.

Chomsky, N., 1968. Language and Mind. New York: Harcourt, Brace, and World.

Chown, W. B.; Fouts, R. S.; Goodin, L. T.; 1974. Productive competence in a chimpanzee's comprehension of commands. Masters thesis, University of Oklahoma.

Fouts, R., 1972. The use of guidance in teaching sign language to a chimpanzee. Journal of Comparative and Physiological Psychology, 80:515-22.

Fouts, R., 1973. Acquisition and testing of gestural signs in four young chimpanzees. Science, 180:97880.

Fouts, R.; Chown, W.; and Goodin, L.; 1973. The use of vocal English to teach American Sign Language (ASL) to a chimpanzee: Translation from English to ASL. Paper presented at the Southwestern Psychological Association Meeting, Dallas, Texas.

Fouts, R.; Mellgren, R.; and Lemmon, W.; 1973. American Sign Language in the chimpanzee: chimpanzee-to-chimpanzee communication. Paper presented at the Midwestern Psychological Association Meeting, Chicago, Illinois.

Furness, W. H., 1916. Observations on the mentality of chimpanzees and orangutans. Proc. Amer. Phil. Soc., 55:281.

Gardner, B. T., and Gardner, R. A., 1971. Two-way communication with an infant chimpanzee. In: Behavior of Non-Human Primates, A. M. Schrier and F. Stollnitz, eds. New York: Academic Press.

Gardner, R. A., and Gardner, B. T., 1969. Teaching sign language to a chimpanzee. Science, 165:644-72.

Hayes, K., and Hayes, C., 1951. The intellectual development of a home-raised chimpanzee. Proc. Amer. Phil. Soc., 95:105-109.

Hayes, K., and Hayes, C., 1952. Imitation in a homeraised chimpanzee. Journal of Comparative and Physiological Psychology, 45:450-59.

Hoyt, A., 1941. Toto and I: A Gorilla in the Family. New York: Lippincott.

Kellogg, W. N., 1968. Communication and language in the home-raised chimpanzee. Science, 162:42327.

Kellogg, W. N., and Kellogg, L. A., 1967. The Ape and the Child: A Study of Environmental Influence on Early Behavior. New York: Hafner.

Köhler, W., 1921. Methods of psychological research with apes. In: The Selected Papers of Wolfgang Köhler, Mary Henle, ed. New York: Liveright, 1971, pp. 197-223.

Kohts, N., 1935. Infant ape and human child (instincts, emotions, play and habits). Scientific Memoirs of the Museum Darwinian (Moscow).

La Mettrie, Julien Offray de, 1948. L'Homme machine. Reprint. Chicago: Opencourt, 1912.

Lenneberg, E. H., 1968. The natural history of language. In: The Genesis of Language, a Psycholinguistic Approach, F. Smith and A. Miller, eds. Cambridge: MIT Press, pp.215-52.

Lieberman, P., 1968. Primate vocalizations and human linguistic ability. Journal of the Acoustical Society of America, 44:1574-84.

McCarthy, D., 1954. Language development in children. In: Manual of Child Psychology, L. Carmichael, ed. New York: Wiley.

Mellgren, R.; Fouts, R.; and Lemmon, W.; 1973. American Sign Language in the chimpanzee: semantic and conceptual functions of signs. Paper presented at the Midwestern Psychological Association Meeting, Chicago, Illinois.

Pepys, S. The Diary of Samuel Pepys, vol. 3, R. Latham and M. Williams, eds., 1912. Reprint. Berkeley: University of California Press, 1970.

Premack, A. J., and Premack, D., 1972. Teaching language to an ape. Scientific American, 227:92-99.

Premack, D., 1970. A functional analysis of language. Journal of the Experimental Analysis of Behavior, 14:10725.

Premack, D., 1971a. Language in chimpanzee? Science, 172:808-22.

Premack, D., 1971b. On the assessment of language competence and the chimpanzee. In: Behavior of Nonhuman Primates, vol. 4, A. M. Schrier and F. Stollnitz, eds. New York: Academic Press, chap. 4.

Riesen, A. H., and Kinder, E. F., 1952. Postural Development in Infant Chimpanzees. New Haven: Yale University Press.

Rumbaugh, D.; Gill, T. V.; and Glasersfeld, E. von.; 1973. Reading and sentence completion by a chimpanzee (Pan). Science, 182:731-33.

Schlesinger, H. S., and Meadows, K. P., 1972. Sound and Sign, Childhood Deafness and Mental Health. Berkeley: University of California Press.

Stokoe, W. C.; Casterline, D. C.; and Croneberg, C. G.; 1965. A Dictionary of American Sign Language on Linguistic Principles. Washington, D.C.: Gallaudet College Press.

van Hooff, J. A. R. A. M., 1971. Aspects of the Social Behavior and Communication in Human and Higher NonHuman Primates. Rotterdam: Bronder-Offset H.V. van Lawick-Goodall, J., 1968. The behaviour of freeliving chimpanzees in the Gombe Stream Reserve. Animal Behavior Monographs, 1:163-311.

Witmer, L., 1909. A monkey with a mind. In: The Psychological Clinic, vol. 3., no. 7. pp. 189-205.

Yerkes, R. M., 1925. Chimpanzee Intelligence and Its Vocal Expression. Baltimore: Williams and Wilkins.

Yerkes, R. M., 1927. Almost Human. New York: Century.

Yerkes, R. M., and Yerkes, A. W., 1929. The Great Apes. New Haven: Yale University Press.