| ROA: | 193 |
|---|---|
| Title: | A case for Enriched Inputs |
| Authors: | Ronald Sprouse |
| Comment: | |
| Length: | 6 |
| Abstract: | In this paper I anlayze complex alternations which require an intermediate representation between UR and output and therefore appear to pose a serious problem for OT and other two-level theories of phonology. I argue for the need for an intermediate representation in OT and propose the Enriched Input model, which provides this representation without creating a true three-level theory. An intermediate representation is motivated by cases of compensatory lengthening triggered by affixation as well as an opaque relationship between epenthesis and deletion in Turkish, both of which require a link between non-surfacing segments and syllable structure that is not present in UR. The Enriched Input model provides a set of candidate inputs of varying degrees of syllabification, all of which are required to unify with the underlying representation. Candidate input-output mappings are evaluated on the well-formedness of the input and output and Faithfulness of the input-output mapping. The evaluation of candidate input-output mappings is accomplished through use of a tableau des tableaux of the type already proposed for Lexicon Optimization. The paper concludes by showing how the Enriched Output model can be applied to other types of opacity, such as the relationship between vowel harmony, lowering, and epenthesis in Yowlumne (Yawelmani). This document is a handout prepared for TREND VII, held May 3, 1997 at UCSC. |
| Type: | Paper/tech report |
| Area/Keywords: | |
| Article: | Version 1 |