Morphemes and candidates in Optimality Theory | draft, comments welcome |
This paper is about the nature of morphemes and, more speci cally, the proper relationship between morphemes and candidate sets within the framework of Optimality Theory. I shall argue that the phonological information of morphemes is best encoded in constraints rather than in representations. Optim...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Text |
| Language: | English |
| Published: |
1995
|
| Subjects: | |
| Online Access: | http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.387.2623 http://roa.rutgers.edu/files/44-0195/44-0195-RUSSELL-0-0.PDF |
| Summary: | This paper is about the nature of morphemes and, more speci cally, the proper relationship between morphemes and candidate sets within the framework of Optimality Theory. I shall argue that the phonological information of morphemes is best encoded in constraints rather than in representations. Optimality Theory has rejected the rule-based derivations that characterized generative phonology, yet it has kept another central tenet, the idea of underlying representations (URs), even at the cost of signi cant complications in the overall architecture of the theory. But, given the other machinery available to OT, URs are no longer necessary. Inkelas (1994) observes that \[g]rammar multiplication reduces need for underlying phonological contrasts; taken to the logical extreme, it makes underlying phonological representation unnecessary altogether", though she assumes that the problem must be with the idea of multiple grammars rather than the idea of underlying representations. URs are not necessary to encode the phonological information of morphemes. Nor are they su cient|we still need morpheme-speci c constraints, most obviously constraints using the Align schema of McCarthy and Prince (1993b). But representational coding of morphemes is not the only option that has been explored. There have been three broad approaches within linguistics to the nature of morphemes and to the phonological information associated with morphemes, which we can summarize as: (1) i) morphemes are representations ii) morphemes are rules iii) morphemes are constraints A shorter version of the Nisgha analysis in this paper was presented at the Canadian Linguistics Association in Calgary in June 1994, and appears in the proceedings (Russell 1994). I am grateful to the audience at the CLA for much useful discussion. My understanding of Nisgha grammar was greatly increased by eld |
|---|