A transparent reanalysis of self-destructive feeding

Opacity can be observed in pairwise rule orderings other than counterfeeding and counterbleeding, one of which being ‘self-destructive feeding’. Like other opacity, self-destructive feeding poses problems to Standard OT. Unlike other rule interactions, all documented self-destructive feeding cases share two characteristics highly dependent on morpheme edges: (1) they all involves instances of non-derived environment blocking; and (2) they all attempt to simplify consonant clusters with a crosslinguistically less common method. In this paper, I reanalyse self-destructive feeding with underspecification and contextual faithfulness constraints, and demonstrate how it can be handled in Standard OT. I argue that the new proposal not only shows that some cases of opacity can be dealt with in Standard OT once the UR is more properly understood, but also offers a new angle to analyse self-destructive feeding: as an epiphenomenon of phonologically-conditioned allomorph optimisation, rather than a separate type of opaque interaction between two rules that generally apply in a language.