Data delimited by the phonology-to-semantics framework of mainstream linguistics are inadequate to account for the subtle influences of first language (L1) upon second language (L2). Unique errors for which there is no L1 correlate are found in samples of written English produced by Japanese and Chinese second language learners. This is due to a variety of language transfer from L1 function to L2 form. In investigating the nature and causes of this variety of transfer, these questions are asked: (1) Does any part of written English that is identifiable as non-native reflect attempts by the learner to match features of English syntax to certain discourse functions? (2) Does this occur especially in cases where such functions are marked differently in the learner's native language than they are in English? (3) Do learner attempts to encode in English the notions topic/comment, theme/rheme, and so on differ from those of a native speaker of English? (4) What syntactic devices do different learners resort to for such purposes? Close study of written discourse will provide answers to these questions and more knowledge about the dynamic processes of language acquisition. (PMJ)