This paper provides some empirical support for the hypothesis that there are two types of language control: compartmentalized control in which performance on discrete point tests or achievement tests is relatively unrelated to performance on communication tests and intergrated control in which the two types of performance are more highly related. Data is presented from three studies: one, a study of individual differences; the other two, controlled experiments in foreign language instruction. The evidence tends, in general, to support the posited distinction, and three factors are suggested as accounting for the two types of competence.Next, the paper considers the hypothesis that integration indicatesacquisitionin Monitor Theory terms, and the data is interpreted in terms of this hypothesis.Finally, a method of teaching for integration is suggested which incorporates the three factors identified in the empirical studies.