A Thematic Guide to Optimality Theory (review)

The new Cambridge University Press series, Research Surveys in Linguistics, aims to provide a new tool for linguists: "concise, single-authored, non-textbook reviews of rapidly evolving areas of inquiry." McCarthy's Thematic guide to Optimality Theory fulfills the publisher's goals. The rapidly evolving subject of interest is Optimality Theory, a grammar formalism elaborated by Alan Prince and Paul Smolensky in the early 1990s, which has inspired a massive scholar production, both by its believers and by its detractors. The research survey is single-authored, with McCarthy being one of the leading scholars in OT. McCarthy provides a concise survey of the conceptual foundations of OT, with each of the four chapters ending with detailed and up-to-date recommendations for further readings. Meant to be a useful reference book, A thematic guide to OT displays an extra-fine modular structure, where each section can be consulted independently from the others and yet is bridged to them by a system of cross-references. Both cross-references and bibliographical references are flagged throughout the main text. McCarthy's guide could be easily renamed "The OT Companion": it is the book to consult while thinking in OT terms, while computing an OT analysis, while mastering OT awareness, rather than the book which teaches OT as analytical tool, exercise after exercise. In the guide, for instance, the examples used to illustrate OT concepts are either abstract (e.g., referring to constraints C1 and C2, and outputs [out1] and [out2]), or, when referring to real languages, they are highly simplified (as noted by Odden 2003:167). The guide has been explicitly made a non-textbook by McCarthy and it is a well-designed prototype for the Research Surveys in Linguistics series. However, due to major typos, the errata corrige list, compiled by McCarthy himself and available at people.umass.edu/jjmccart/errata.html, should supplement the book.

The book is organized into four chapters. Chapter 1, "The core of OT", introduces the architecture of the model; Chapter 2, "The context of OT", reconstructs the genesis of the model within the development of phonological theory in particular and linguistic theory in general; Chapter 3, "The results of OT", describes the consequences of doing phonology, morphology, and syntax by means of OT; and Chapter 4, "The connections of OT", discusses the use and usefulness of OT in other areas of linguistic inquiry, such as language acquisition and learnability, and language variation and change.

Chapter 1, "The core of OT", is roughly divided into two major subsections, a theoretical one and a practical one. The theoretical subsection (1.1) outlines the basic architecture of OT by defining the role and function of each of its components. These are GEN (the module generating candidates from a given input), the language-specific constraint hierarchy (a grammar), and EVAL (the module applying the given grammar to the set of candidates generated by GEN in order to find the most harmonic of them, that is, the output). Crucial to this model is the discussion of the theory of constraints (universality of constraints, constraint typology, constraint schemata; 1.2), and of constraints interaction (faithful and unfaithful mappings, blocking effects; 1.3). A practical subsection (1.4), on the other hand, demonstrates "How to do OT"; it is a compendium of rules of thumb. It exemplifies how to rank known constraints, how to select informative candidates, and how to overcome "problems of stinginess" (p. 36), when hierarchies cannot select the right candidate. The discussion of this last point (1.4.4) is extremely welcome, as weak OT analyses usually posit ad hoc constraints to ensure the right output. This praxis undermines the necessary rigorousness of a model of Universal Grammar since constraints must be understood as "claims about UG" (p. 39). McCarthy strongly advises that the introduction of a new constraint should be undertaken only after having carefully considered the interaction of known constraints and/or the modification of known...