Free The Articulate Mammal by Jean Aitchison

‘swan’. However, since she was beside a river when she produced this combination, it is possible that she made two separate signs, one referring to the water, the other to the swan.
    In her new home, Washoe was given an infant chimp, Loulis, to adopt, and tried to teach him some signs. On one occasion, Washoe put a chair in front of Loulis, and then demonstrated the CHAIRSIT sign to him five times. And, both through imitating Washoe and other signing chimps, Loulis developed his own repertoire of signs (Fouts et al 1982; Fouts 1983). These days, Washoe, Loulis and two other chimps, Tatu and Dar all live together. They interact with humans and each other by means of signs, though of course also use spontaneous chimp gestures and vocalizations.
    Now the fact that Washoe spontaneously transmitted signs to another chimp is interesting and important, but it does not magically turn these signsinto ‘language’. In brief, we have to conclude that although Washoe’s speech is sometimes creative, and showed semanticity and displacement, it has not been shown to be structure-dependent. We cannot be sure, because Washoe’s ‘speech’ was only ever partly analysed – recording it all was impossible, and any repeated signs were usually ignored by the Gardners.
    But Nim Chimpsky, a male chimpanzee, who was taught a sign system some years later, was attended by a fleet of graduate students who recorded his every sign. He was for several years under the care of Herbert Terrace at Columbia University, New York. Somewhat ironically, Nim’s achievements began to interest psycholinguists mainly after the project ran out of money, and Nim was returned to a chimpanzee colony in Oklahoma. Without Nim around, Terrace found that he had much more time to analyse the material he had collected so far. The data from Project Nim, therefore, have been examined much more carefully than those from any of the other animals. With Nim out of the way, Herbert Terrace was able to sort out and classify the data he had accumulated over the previous 4 years.
    At first sight, Nim’s sign sequences were impressive. Of the 20,000 recorded, approximately half were two-sign combinations, and 1,378 were different. A superficial look at the signs suggested to Terrace that they were structured (Terrace 1979a: 72). For example, of the two-sign utterances which included the word MORE, 78 per cent had MORE at the beginning as in MORE TICKLE, MORE DRINK, and of the two-sign utterances involving a transitive verb (a verb which takes an object), 83 per cent had the verb before the object, as in TICKLE NIM, HUG NIM. But a closer analysis showed that the appearance of structure was an illusion. Nim simply had a statistical preference for putting certain words in certain places, while other words showed no such preference. He preferred to put the word MORE at the beginning of a sequence, the word NIM at the end, and any foods he was requesting at the beginning also. But many other words had a random distribution. Take the word EAT, a high frequency item in his vocabulary. It occurred in the two-, three- and four-sign sequences set out in the tables below.

    It would require a considerable amount of imagination and wishful thinking to detect a coherent structure in such a collection. Looking at the two-sign sequences, we note that EAT NIM, NIM EAT and ME EAT are all very common, making it impossible to claim that there is a firm subject–verb, or verb–subject order. A similar pattern occurs in the three-sign sequences, with EAT ME NIM, NIM ME EAT, ME NIM EAT and EAT NIM ME all occurring a significant number of times. It is particularly noticeable that Nim’s longer utterances were not in any way more interesting and sophisticated than his shorter ones – they were simply more repetitive. Of the thirteen four-sign sequences noted above, ten of them involved repeated items, and five of them were simply a doubling up of two-sign utterances: EAT DRINK EAT DRINK, EAT NIM