Optionality and variation in optimality theory: Focus on Korean phonology

Optimality Theory (OT) relies on a strict dominance ranking system among constraints in which parallel evaluation unexceptionally predicts only a single optimal output form mapped onto a single input form. Optimization is thus realized under a one-to-one mapping between input and output. Standard OT, in its essence, has no room for more than one optimal output on the surface. However, this dissertation alternatively proposes a new notion of a ranking-based strategy to handle optional phenomena of phonology while keeping OT's basic spirit. Other previous ranking strategies proposed in the OT literature cannot treat optionality (or free variation) properly. Tied ranking predicts optionality only partially, due to the potential role of low-ranked constraints in the evaluation. Re-ranking (or free ranking) cannot enable optional forms to emerge from a single evaluation process.;Given the input-output correspondence schema, surface multiplicity resulting from optionality or from variation undergoes a different mapping procedure and thus each necessitates a different ranking strategy. Intra-speaker optionality reflects a one-to-many mapping relationship while dialectal (or speech group) variation reflects a one-to-one mapping relationship. For the former, Reversible Ranking is proposed as opposed to multiple constraint hierarchies or co-phonologies while for the latter, re-ranking is invoked. The crucial role of Reversible Ranking in predicting optionality is confirmed by the analysis of Korean vowel phonology. Dialectal variation, on the other hand, is due to different rankings of two constraints. This captures well the variational aspects of Korean Phonology.