Despite considerable work concerning relationships between cognition and language in the second year of life, these relationships remain obscure. Clarification depends on resolution in at least three major areas. First, studies should be expanded beyond the scope of cognitive prerequisite models that have been difficult to conceptualize and that may prove too narrow to guide research. Second, in order to test models that go beyond correlation and evaluate structural correspondences across domains, researchers must be able to specify the sequence of development in both language and nonlanguage abilities and propose theoretically sound correspondences. Third, a realistic model for language acquisition must take account of other noncognitive, biological, and social variables influencing language, thus replacing simple cause-effect predictions with a more multidimensional model. The multidimensional model of language acquisition includes biological, experiential, and cognitive variables and their interactions. Recent longitudinal and cross-sectional studies offer evidence of the usefulness of the model. These studies, bearing on the development of combinational abilities in play and language, suggest ways the model might be employed. (RH)