Investigations in Optimality Theory
Typology, Learning, and Modeling
The goal of this research is to improve understanding of phonology, the study of the sound systems of human languages. What do native speakers of a language know about its sound system? How is that knowledge acquired? Optimality Theory is an influential approach to studying phonology and other aspects of language. This project explores the results of changing two of Optimality Theory's basic assumptions. One is the assumption that the output of a grammar is reached in a single step from the input. The alternative, called Harmonic Serialism, derives the output of the grammar by a succession of steps, each of which improves wellformedness. The other is the assumption that constraints on well-formedness are ranked in a strict hierarchy. The alternative, called Harmonic Grammar, assigns numerical weights to constraints, allowing for additive effects.
The principal goal of the project is to determine whether these alternatives, individually or together, offer a better approach to phonology than standard Optimality Theory does. Preliminary research has produced some evidence in support of the alternatives as well as some challenges to them. The work of the project involves studying the evidence, the challenges, and the theories themselves, with the aim of settling this issue and thereby improving understanding of phonology. The methods employed in this project include typological and computational research on phonological phenomena, examination of longitudinal data from child phonology, and computer modeling.