L'interférence et la Théorie Phonologique

The purpose of this paper is to argue that the facts of interphonology support a division of processes relating sound alternations into two distinct types: those that cannot cause negative transfer and those that can and often do. This distinction has been captured in two ways in recent phonological theory: (i) by proposing an autonomous morphophonology {cf. Hooper 1976) and (ii) by positing levels and strata (cf. Mohanan 1982 and Kiparsky 1982). Both internal and external evidence, such as the phonological behaviour of nonsense words {cf. Gussmann 1980) and the necessity of doing at least some inflection in the lexicon (cf. Lieber 1981), argues against the former model. The latter model succeeds in avoiding autonomous morphophonology, but only at the cost of postulating levels and strata, constructs for which interpho-nology offers no support. The facts of negative transfer or interference, it seems to us, are best accounted for by a theory that accounts for non-global "morphophonology" directly in the morphological component of the grammar (cf. Ford and Singh 1984 and Singh à paraître).