Overwriting as optimization

摘要

Recent optimality-theoretic analyses of fixed segmentism reduplication and root-and-pattern morphology invoke a technique of overwriting: Phonotactic constraints prevent the cooccurrence of an affix with the unmodified reduplicant or base word, and the resulting conflict is resolved by replacing part of the lexical material with the affix (Ussishkin 1999; Alderete et al. 1999). Nevins (2005) claims that this type of approach both overgenerates and undergenerates since it predicts unattested overwriting patterns, and cannot account for specific phenomena in Hebrew and Hindi. In this paper, we show that a substantial part of the over-generation argument is empirically flawed, and argue that all remaining problems find a straightforward solution in the independently motivated parametrization of optimality-theoretic constraints and a general metacondition on possible constraint rankings.