This chapter introduces the central premises of Optimality Theory. The chapter begins (§1.1) with the overall structure of OT, as proposed by Prince and Smolensky (1993). It continues with some general remarks about the nature of constraints (§1.2) and their modes of interaction through ranking (§1.3). These threads are joined to some practical suggestions for doing OT in §1.4. Readers encountering OT for the first time are advised not to read this chapter straight through; see “How to Use This Book” for a better plan of attack.

Basic Architecture

Candidate Comparison

Many theories of language can best be described as operational, rule based, or transformational: they take an input and apply some procedure that changes it into an output. But the primary action in OT is comparative: the actual output is the optimal member of a set of candidate output forms. Interesting analytic and theoretical results in OT come from understanding the details of how candidates are compared.

Candidates are compared by applying a hierarchy of violable constraints. The constraints assess the form of a candidate and its relationship to the input. Candidates inevitably differ in performance on various constraints. Of two candidates, the more harmonic is the one that performs better on the highest-ranking constraint that distinguishes between them. The actual output – the most harmonic or optimal candidate – is the one that is more harmonic in all its pairwise competitions with other candidates.

Because constraints are violable, the output typically disobeys at least some of the lower-ranking constraints.