Argument structure
Are animals able to form sentences that can be said to be homologs or analogs of what one finds in human language; in particular, do they have the ability to acquire an argument structure? We will say that an argument structure is built around a predication where there is a verb (or predicate) having at least one (nominal) argument.
Kanzi developed two kinds of productive combining: The first concerneda lexigram to specify an action (e.g. tickle), subsequently a pointing gesture was used to specify the agent. Savage-Rumbaugh and Lewin (1994: 161) found that Kanzi’s ‘‘rule’’ of action first, using lexigrams, and agent second, using gestures, represented the opposite order of the spoken English that was used around him all the time, and they argue that this fact provides strong evidence for creative productivity in this animal. One might argue, however, that this deviance from the model that Kanzi was exposed to could be due to the different modalities that were recruited by the animal. Nevertheless, this ‘‘rule’’ appears to have been productive: Once Kanzi had learned the action–object ‘‘rule’’, he appears to have extended it to new situations. Such a situation obtained, for example, when Kanzi asked someone to play tickle-ball, meaning to tickle him by rubbing the ball all over his body. In this new combination, he followed exactly the ordering he had demonstrated earlier (Savage-Rumbaugh and Lewin 1994: 159).
A second kind of combining arrangement that Kanzi acquired was putting together two action units, such as tickle-bite, which he himself had created—caretakers almost never used such combinations (Savage Rumbaugh and Lewin 1994: 161). Kanzi produced almost exclusively only two elements in a proposition; in the rendering of the authors concerned (Greenfield and Savage-Rumbaugh 1990; Kako 1999: 9), he comprehended ‘‘both the words and their relations to one another’’, as well as the influence of the action (‘‘verb’’) meaning on the argument (‘‘thematic role’’)—in other words, he appeared to have the ability to understand the nature of an elementary one-argument proposition.
An understanding of distinctions in participant roles has also been observed in other Pan species. The chimpanzee Dar was observed to vary the place of articulation of his signs to include agents, locations, and instruments (Rimpau, Gardner, and Gardner 1989; Casey and Kluender 1998). And the chimpanzee Sarah is said to have learned to distinguish between case functions (‘‘subject’’ vs. ‘‘object’’): Premack (1976: 322) says that Sarah comprehended ‘name of’ when it was used as part of an accusative phrase although, in all of her previous experience, it was confined to the nominative phrase. She was equally successful in transferring the quantifiers from the nominative phrase in which they were taught her to the accusative phrase in which they were later presented. For example, she had been trained to produce sentences of the form ‘Some cracker is round,’ ‘All cracker is PL square,’ etc., and later performed correctly when instructed ‘Sarah take some cracker,’ etc. Thus, although both ‘name of’ and the quantifiers were learned originally as parts of ‘‘nominative phrases’’, she comprehended them when she later experienced them as part of ‘‘accusative phrases’’. The problem with some animals is that the possible link between syntactic position and thematic role is not always clear: Are the animals able to predict some semantic properties of arguments from syntactic information?
Conversely, it has also been claimed that chimpanzees are not able to acquire argument structure. Rivas (2005) concludes that the five chimpanzees studied by him, who had been taught ASL, lacked a semantic or syntactic structure in combining signs.
Combining patterns suggestive of argument structure is not restricted to primate species. The grey parrot Alex learned to use the structures [wanna go + location unit] and [want/wanna + object unit] at least with a range of form–meaning categories (Pepperberg 1987c). While he distinguished volitional propositions (want X, wanna go X) and commands (e.g. go X, you tickle me), it is not entirely clear whether these come close to reflecting an argument structure involving stable arguments in combination with predicates.
Bottle-nosed dolphins have been taught to understand propositional structures whose constituents were object terms (O), nonrelational action terms (A), relational action terms (R), and the optional modifier (M) terms LEFT and RIGHT, and Herman (1989: 24, 36) proposes a set of five rules that the dolphin Ake learned, for example when being directed to take Object 2 (¼ O2) to Object 1 (¼O1). Her man concludes that Ake had an understanding of argument structure because when, for example responding to LEFT OVER, with the object term O missing, she jumped over the ball to her left, and when given the sign FETCH alone with no object sign preceding it, she cleared the tank of objects, often bringing back two or three at once to her trainer.
