ABSTRACT
This chapter outlines a theory of what is learned in acquiring word classes. It is not a model of how learning takes place. Although the theory has a natural fit with my sieve-memory acquisition model, it would no doubt be quite compatible with other models of the acquisition process and is therefore open for borrowing. The theory is a special case of what Pinker has called a "bootstrapping" theory. It shows that if we take the distinction between a predicate and its arguments for granted as a primitive notion that is within the child's cognitive competence at the outset of language acquisition, then it is possible to build a theory of the acquisition of word classes that is based primarily on the child's sensitivity to the distribution of elements in argument phrases and predicate phrases.
