It is proposed that, in contrast to Chomsky's argument, it is possible to arrive at an empirically grounded definition of innate linguistic competence that guides the child in the construction of grammar, particularly when this process is viewed as developmental. This approach treats language acquisition as a process of change. It is suggested that it is possible, from existing data on child language, to arrive at general operating principles for language acquisition, or strategies for language construction, along with a characterization of the child's natural and growing organization of semantic and phonological space and basic syntactic principles. Recent research in this area has accounted for the development of morphological paradigms, canonical sentence forms, placement of operators, and various patterns of over- and underextension of meaning. Research on grammatical morphemes is providing evidence that linguistic forms and constructions have a long developmental history closely tied to discourse functions. Attention to the mutual relationship between form and function is seen as essential to understanding both child language and historical language change. (MSE)