On the From of Default Rules in Phonology

The theory of Lexical Phonology proposed by Kiparsky (1982; 1985; 1993; 1995)has toally abandoned the distinction between types of phonological rules and those of redundancy ru1es. By the theory of Underspecification, the feature specifications are underspecified at the level of underlying representation if they are predictable by rules, including phonological and redundancy rules. The Trisyllabic Shortening Rule(TSS), which has been regarded as a kind of phonological rule, fills in [-long] in non-derived phonological environments, and it may change the feature specification [+long] into [-long]in derived phonological environments: in the former case of rule application TSS functions as a redundancy rule, while in the latter case it behaves as a phonological rule. Thus TSS performs the double duty of (i) feature-filling and (ii)feature-changing functions. Within the framework of Lexical Phonology, the modes of rule application are governed by the interaction of principles of Underspecification and Structure Preservation: TSS applies lexically to fill in [-long] in such words as Italy([ølong]təli])and to change [+long] into [-long]in such words as sanity([sæniti]).