An analysis of language acquisition in English and Dutch focuses on a theory of phrase structure. It is argued that the previously posited phrase structure operations of projection and adjunction can be dispensed with in favor of the single operation of "merge." One version of merge is shown to account for a range of data from child English and child Dutch. It is also proposed that the phrase structure possibilities this version yields also account for previously reported but unsolved syntactic puzzles in child language and for adult uses of lexical items. The primary finding is that merge is, both conceptually and empirically, a more parsimonious operation; a single operation facilitates the mapping of phrase markers to children's earliest utterances. Contains 14 references. (Author/MSE)