9421404 ABSTRACT Recent developments In linguistic theory have led to a reconsideration of the nature and role of optimality considerations in the overall architecture of grammar. Emerging from this research is the idea that different components of grammar interact to yield the best choice from a set of canidates. This idea departs from more traditional approaches to the output of linguistic levels in generative grammar, by which rules, principles and constraints interact to determine the grammatical status of each linguistic object independent of the status of possible competitors. In the past year or two, interest in the linguistic role of optimality has been sparked by the sharpened notions of "economy" in Chomsky's Minimalist Program and by Prince and Smolensky's Optimality Theory, originally developed for phonology. As work on these notions has explored new analytic possibilities, many questions have arisen. These include new versions of an old debate between constraints on derivations and constraints on representations, and entirely new questions about the nature of the candidate set, in addition to complex and as yet imperfectly formulated questions about learnability and computability. These sorts of questions are coming to occupy a center stage in current debate. Nonetheless, the questions have not been systematically discussed, either in print or in a public workshop. Therefore, a workshop addressing these issues will convene at MIT May 19-21, 1995. The explicit aim of the workshop is to compare approaches and define questions for further research. The discussion will be as focused as possible, in the hope that genuine scientific progress can be made during the three days of the conference.