which the work done by transformational grammars could be done without derivations using relational networks with only two sets of grammatical relations. Postal and Johnson ( 1980), accepting my conclusions that network representations are correct and derivations are unnecessary, have developed a theory of arc-pair grammar that permits sequences of relations.
Again, I think it is no accident that the linguistic results favoring network representations with gestalt properties mesh with the visual results favoring representations with gestalt properties. It suggests the correctness of the experiential paradigm and incorrectness of autonomous paradigm.
Let us now turn to the question of gestalts. One thing that we know about visual representations is that they must have certain gestalt properties. At the very least, the whole must be greater than the sum of the parts and partial matching of visual representations is required. At least one of the AI systems using network representations does incorporate such gestalt properties to a certain extent--namely, Winograd and Bobrow's KRL system. In my paper on "Linguistic Gestalts ( Lakoff, 1977) I gave six arguments to show that syntax must work by partial matching. In addition I gave a variety of examples indicating that the meaning of whole must be greater than the sum of the meanings of the parts for natural language sentences.
I have barely begun to mention, much less elaborate on, the empirical reasons I have for preferring the experiential paradigm. But there is an overriding reason for preferring it. After a generation of research in which it was implicitly assumed that language could be described on its own terms, it has become more interesting to ask how much of the structure of language is determined by the fact that people have bodies with perceptual mechanisms and memory and processing capabilities and limitations, by the fact that people have to try to make sense of the world using limited resources, and by the fact that people live in social groups and have to try to communicate with each other. It seems to me that a great deal of the structure of language is determined by such factors.
Becker J. 1975. In Theoretical issues in natural language processing, ed. R. C. Schank & B. L. Nash-Webber . Arlington, Va.: Tinlap.
Bolinger D. 1976. "Language and memory". Language Sciences.
DuBois J. 1974. "Syntax in mid-sentence". In Berkeley studies in syntax and semantics, vol. 1.
James D. 1972. Some aspects of the syntax and semantics of interjections. Proceedings of the 8th Annual Meeting of the Chicago Linguistic Society.
James D. 1973a. Another look at, say, some grammatical constraints on, oh, interjections and hesitations. Proceedings of the 9th Annual Meeting of the Chicago Linguistic Society.
James D. 1973b. The syntax and semantics of some interjections in English. Ph.D. dissertation, Linguistics Department, University of Michigan.
Kay P., & McDaniel C. 1975. Color categories as fuzzy sets. Unpublished working paper no. 44, Language Behavior Laboratory, University of California, Berkeley.