Drawing data from a corpus of more that 100,000 spontaneous utterances, Andrew Radford demonstrates that the fundamental characteristic of children's earliest structures is that they are essentially lexical and thematic in nature. They show evidence of the acquisition of lexical but not functional categories, and of thematic but not nonthematic constituents. This hypothesis provides a unified account of a wide range of phenomena in early child English. This detailed study of children's initial grammars suggests a model of acquisition which is essentially maturational. Different modules of the child's grammar come into operation at different stages of development, triggered by relevant aspects of the child's experience. In this, Radford's account sheds significant light on some of the fundamental questions for the theory of language acquisition.