Abstract The paper compares a morphophonological and a syntactic approach to stress shift and enclisis/proclisis asymmetries. The former analysis envisages the co-occurrence of multiple allomorphs with the same featural content, which are distributed in accordance with phonological rules. The latter analysis, by contrast, is based on the hypothesis that alternations are due to the co-occurrence of items belonging to different functional classes. I argue that the syntactic analyses put forth so far are not superior to previous morphophonological accounts. I acknowledge that the synchronic distribution of allomorphs does not result from productive phonological rules. Nonetheless, this is not per se a valid argument in favor of syntactic, class-based accounts, which must be advanced in compliance with Occam’s razor.