A computational model of discourse production
pp. 125-136
in: Robin N. Campbell, Philip T. Smith (eds), Recent advances in the psychology of language, Berlin, Springer, 1978Abstract
A central problem in the psychology of language is to explain as precisely as possible how the speaker of a natural language expresses himself in words. The problem is really twofold. In the first place, how does the speaker decide what to say? Secondly, how does he find the words in which to say it? Existing grammatical theories (Chomsky 1965; Lakoff 1971) address the latter problem by positing a semantic representation for each sentence, and attempting to prescribe, for a given language, a set of rules for mapping such semantic representations on to syntactic structures, or vice versa. This approach, however, does not cast any light upon the first problem, namely how the semantic representations are created in the first place; the main difficulty here is in specifying the message which is to be conveyed independently of the linguistic structures which must be created in order to express it.