Modelling [s] to [ ] accommodation in English

When adjacent segments compete for the same articulator a window is provided on how conflicting phonological specifications are reconciled in articulation. One possibility is that the phonological specifications of the two segments simply overlap in time, yielding as output a blend of the two segments. This, broadly, is the prediction of Articulatory Phonology. Another possibility is that one phonological specification is changed in a pre-articulatory cognitive representation so as no longer to conflict. This is assimilation in the sense in which it has traditionally been used in phonology. A previous acoustic study of / s 4 * / sequences by the first two authors suggested that while some outputs are compatible with gestural blending, there are others where the target for the first segment has been changed from [ s] to [ * ]. Doubts have been raised, however, concerning the validity of inferring articulation from acoustic data, given in particular the possibility of ‘quantal ’ ef fects in coronal fricatives. The present paper uses electropalatography (EPG) (a) to check the relation between articulation and acoustics in the [ s – * ] region, and (b) to explore more directly the production of / s 4 * / sequences. The findings of fer support, but not unequivocally, for the claim that a cognitive phonological process of assimilation may sometimes apply