Prosodic domains and ambisyllabicity in optimality theory

Selkirk 1984, 1986, Nespor & Vogel 1986 and others independently argue that PrWd structure is built from morphological structure. In Optimality Theory, Generalized Alignment (McCarthy & Prince 1993b) has been successful in encoding the close relationship between prosodic structure and morphological structure. However, surface syllabification renders morphological boundaries opaque in a compound /CVC-VC/, in which each Root of the compound is identified as a separate PrWd in a language such as Korean (Kang 1992). In such case, owing to the requirement of ONSET, the final consonant of the first Root of the compound must be syllabified as an onset, leading to PrWd-Root misalignment. However, we show that unique-onset syllabification is empirically not tenable in Korean; a variety of phonological phenomena suggest that this consonant must be syllabified as a coda, seemingly requiring an abstract syllabification and thus posing a challenge to Optimality Theory, in which abstract syllabification is impossible. This dissertation proposes that a Root juncture is in fact non-crisply aligned with a PrWd juncture. We further argue that the PrWd-final consonant which is followed by a vowel across a PrWd juncture is realized as ambisyllabic. This proposal is strongly supported by several Korean phonological phenomena and English flapping. We demonstrate in Korean that /n/-insertion is compelled to avoid an ambisyllabic consonant before a high front vocoid. We also show in Korean that overapplication of Coda Neutralization and underapplication of primary palatalization of a PrWd-final consonant before a high front vocoid across a PrWd juncture are due to the ambisyllabicity of the PrWd-final consonant. Additionally, we analyze primary palatalization, secondary palatalization and Umlaut in Korean (Iverson 1993, Kiparsky 1993 for Korean and Hume 1990). We will demonstrate that Umlaut is blocked across a secondarily palatalized coronal consonant before a high front vocoid. We propose that Umlaut and secondary palatalization are a single phonological phenomenon and secondary palatalization blocking of Umlaut results from a conspiracy to force the V-place of a high front vocoid to spread only once