ABSTRACT

Presupposing some survival value in being able to anticipate the appearance of particular kinds of objects, imagine that the creatures have a lexicon which enables them to name each of the objects on the conveyer belt. In natural language, the word symbols of speech still appear one after the other in a linear temporal sequence, but the relationships among concepts and things communicated by natural-language speakers are much more varied and complicated. Skinner suggested that the language user can learn to discriminate the grammatical categories of individual words. Linguists and some psychologists have readily adopted underlying structures as formal description of language data while disregarding the question of how such structures, generating algorithms, might be acquired by the principles of learning. The set of grammatically correct sentences of a natural language is somewhat analogous to set A in the lever sequence experiment. Words and connection codes are entered, reinforced, and extinguished during the language-acquisition process.