‘‘Lexicon’’
Which types of form–meaning pairings were distinguished by animals was to a large measure contingent on what they had been taught. The artificial languages used to train bottle-nosed dolphins, for example, included objects, actions, and properties, so these were the units that the animals comprehended and reacted to (Herman et al. 1984; Herman 1987, 1989). But different researchers did not always arrive at the same conclusions on lexical classification. Analyzing videotaped human–chimpanzee inter action, where the animals had been taught ASL, Rivas (2005) found only the following classes of signs: objects, actions, request markers (GIMME, HURRY), the deictic sign THAT/THERE/YOU, and the chimpanzee’s own name sign, although the relevant animals had been taught a larger range of unit types by the Gardners, their original caretakers.
One domain where explicit training of animals yielded particularly notable results is that of the ‘‘lexicon’’, or at least of what corresponds to lexical categories in human languages. The orangutan Chantek had acquired 26 signs at age 2;2, he could use 56 signs when he was roughly three and a half years old and eventually acquired 150 signs (Miles 1983: 49–50; 1990; Miles and Harper 1994), and the gorilla Koko is said to have acquired a ‘‘vocabulary’’ of 100 signs in American Sign Language (ASL) within 30 months, and 243 signs in 50 months of training (Patterson 1978a: 72, 1978b: 173). The chimpanzee Nim had acquired 40 signs at age 2;2, learned to use a vocabulary of 125 signs at age 4, and eventually knew 140 signs (Terrace 1983: 23). The grey parrot Alex acquired more than 90 form–meaning pairings (or vocalizations) with explicit training, including labels for foods and locations (Pepperberg 1994). Teaching the chimpanzee Washoe started when she was about 11 months old, and 51 months later she had a vocabulary of 132 signs of ASL, using space, eye gaze, facial expression, and repetition (Gardner and Gardner 1978). As we noted in “Concepts”, these authors claim that Washoe used her signs not for specific objects or events but rather for classes of referents; for example, the sign for ‘dog’ was used to refer not only to live dogs and pictures of dogs of many breeds, sizes, and colors, but also for the sound of barking made by an unseen dog.
Overall, trained apes have been found to acquire between 120 and 200 different form–meaning pairings, with the gorilla Koko topping the list, having a reported inventory of 243 learned signs. These units include most of all concrete objects and physical actions, while there is a conspicuous lack of items for non-tangible objects and abstract concepts.
Most of these form–meaning pairings were acquired via transmission from human caretaker to animal. However, some of the animals showed the ability to create signs on their own; thus, the chimpanzee Washoe and the orangutan Chantek created five, and the gorilla Koko four new signs (Miles and Harper 1994). Koko was taught a sign TICKLE by Patterson (1978a: 84), but she instead created a more iconic gesture, done by drawing her index finger across her underarm. Furthermore, after three months of training Koko appears to have created a sign to express polar (sentence) questions, by using gestural intonation. That new creations are likely to be iconically motivated is also suggested by the sign for ‘stethoscope’ that Koko created by placing an index finger to each ear (Patterson 1978a: 79, 1978b: 193).
While apes cannot vocalize speech, Savage-Rumbaugh and Lewin (1994: 150) argue that they do understand it. These authors performed three testing sessions with the bonobo Kanzi in which their requests alternated between spoken English and lexigrams. The tests included thirty-five different items, used in 180 trials in English and 180 with lexigrams. Kanzi scored 95 percent correct on the lexigram trials and 93 percent on the English trials; he understood 150 spoken form–meaning pairings at the end of the seventeen-month period.
There is, however, a caveat to some of these findings. We noted above that the chimpanzee Washoe had acquired a vocabulary of 132 signs of ASL. But roughly two decades later, Rivas (2005) found on the basis of 612 videotaped utterances that Washoe distinguished no more than 43 different signs, and it remains largely unclear how this difference is to be accounted for.
In spite of such observations, we assume that most of the findings summarized above are empirically sound. That Washoe and the other chimpanzees were really capable of comprehending lexical distinctions can be demonstrated perhaps more convincingly by the errors they made than by their correct responses. Most of the semantic errors made by the chimpanzees studied by the Gardners (1969, 1978) involved cases where a sign was used incorrectly for a similar meaning, for example COMB for BRUSH, while the form errors concerned mostly pairs of contrasting meanings but signs that were similar in their physical form, like MEAT and OIL. This behavior would seem to suggest that these animals had some understanding of the nature of the sign language they were taught (Fouts 1987: 64).
Numbers A much discussed question in the cognitive sciences is whether animals can represent numerosity: (a) Can they identify a property of the stimulus that is defined by the number of discriminable elements it contains, (b) can they count, and (c) can they use numerical representations recursively?
There are findings that indicate that question (a) can be answered in the affirmative. Many animal taxa can discriminate stimuli differing in number, such as pigeons, parrots, rats, dolphins, monkeys, and chimpanzees (see Brannon and Terrace 1998). Ravens (Corvus corax) and jackdaws (Corvus monedula) succeeded on numerical match-to-sample tests on quantities up to 8, and tested on sets up to 4, the chimpanzee Sheba demonstrated ordinality and labeled, with a card depicting an Arabic numeral, the sum of two arrays separated in time and space (Koehler 1943, 1950; Pepperberg 1987a; Boysen and Berntson 1989). The trained grey parrot Alex replied correctly to the question ‘How many?’ (Pepperberg 1999b: 131) and learned to produce vocal numerical labels for sets of two to six objects and showed remarkable abilities in handling numerical quantities. For example, when presented with two pieces of cork or five pieces of wood, the responses of Alex would be two cork and five wood (Pepperberg 1987a, 1987c: 42, 1994).
But it seems that questions (b) and (c) have to be answered in the negative: Neither do non-human animals show a concept of counting, nor do they appear to have the capacity to create open-ended generative systems—that is, numbers are not acquired by animals in paradigms involving a successor function (Hauser, Chomsky, and Fitch 2002: 1577). The studies available overwhelmingly suggest that non-human animals do not have a natural ability to discriminate numerosity, attending to it only as a ‘‘last resort’’, if other bases for discrimination, such as shape, color, size, frequency, or duration of a stimulus, are eliminated (Brannon and Terrace 1998: 746). Rhesus monkeys (Macaca mulatta) can spontaneously represent the numbers of novel visual stimuli and they can extrapolate an ordinal rule to novel numerosities, but it seems that they do not do so using a counting algorithm (Brannon and Terrace 1998: 748). And while the parrot Alex showed remarkable skills in handling numbers, Pepperberg (1999b: 110) concludes that her data did not indicate if Alex could count; there is no evidence that he is capable of a counting process or of a recursive understanding of numbers.
