ABSTRACT
In the first few years of life, every child discovers the mappings between sound and meaning that constitute a native language. Theories of language acquisition, however, have typically asked how a learner might begin with strings of words rather than sounds, building up phrase and clause structures from word-strings based on some combination of semantic and distributional cues (e.g., Bates & MacWhinney, 1982; Pinker, 1984). The adequacy of idealizing linguistic input as word-strings rests on the assumption that properties of the speech wave not represented in sequences of words have little to do with the acquisition of grammar.
