ABSTRACT Generative-transformational grammar has often been cited as a possible theoretical model for the psychological processes involved in first language acquisition. Of these processes, the discovery of latent structure is widely considered to be the major one. Such discovery may result primarily from inferential operations, and it is plausible that these operations are influential in second language acquisition as well as in first language acquisition. If the discovery of latent structure through inferencing characterizes second language acquisition, a transformational grammar model may have implications for second language learning and instruction. A language program reflecting this model would emphasize the role of syntax in expressing meaning. Instructional techniques based on sets of transformationally related sentences would facilitate the manipulation of syntax while holding meaning constant. Concentration on basic structural alternatives would involve the learner directly with latent structure and encourage its discovery. It would also allow him to focus on expressing his own meaning by deciding from among the structural alternatives for conveying it. Decisions made at the syntactic level would restrict choices at the morphological and phonological levels. Techniques of language habit theory might be useful in mastering the fixed elements of the closed systems at these levels. This theory would thus be incorporated into the more powerful theory of generative-transformational grammar.