Book Reviews Local Constraints vs. Economy David E. Johnson and Shalom Lappin (IBM Thomas J. Watson Research Center and King's College, University of London) Stanford, CA: CSLI Publications (Stanford monographs in linguistics), 1999, x+150 pp; distributed by Cambridge University Press; hardbound, ISBN 1-57586-183-6, $49.95; paperbound, ISBN 1-57586-182-8, $18.95 Reviewed by Annie Zaenen Xerox Research Centre Europe This monograph belongs to a literary species that flowers every time Chomsky comes up with a new proposal and has as its aim to show that what is proposed is woefully inadequate on empirical and theoretical grounds. This kind of literature has a ped- agogical role to play in theoretical linguistics and card-carrying syntacticians should read the book. However, over the years Chomsky's versions of syntactic theory have become so irrelevant to natural language processing that computational linguists can safely skip this book and the one it is a response to, Chomsky (1995). But, as it is not a long book, some computational linguists might be interested in reading the first chapters as a quick summary of early minimalism and the fourth chapter as an introduction to the HPSGtreatment of wh-constructions. After a brief intro- ductory chapter, the authors summarize the Chomsky 1995 version of global economy in Chapter 2. In Chapter 3, they discuss two versions of local economy, that proposed by Collins (1997) and the one developed by Yang (1997). They contrast the Minimalist Program model with what they call the incremental interface correspondence model, which is exemplified in several recent linguistic frameworks, from Montague grammar to LFG, to GPSG, to TAGS, to cite only a few. In Chapter 4, they compare an account of wh-constructions in Iranian Arabic according to a local version of economy to one based on an HPSG local-constraint approach. The last chapter addresses issues of theory construction in general. The general argument developed in the second chapter is somewhat awkward: the authors want to show that the three main economy principles proposed by Chomsky (Procrastinate, the Smallest Derivation Principle, and "Have an Effect on Output") are unnecessary. They do this mainly by showing that other, more GB-like, mechanisms could account for the same data. Given the limited amount of data discussed (a lot of the discussion centers on the contrast between There seems to be somebody in the room and *There seems somebody to be in the room), it would be rather astonishing if this were not the case. We get here the familiar dialogue des sourds where one party points out descriptive problems and alternatives whereas the other party is arguing about "explanatory" adequacy. The discussion allows the authors to make a couple of points about the computational complexity of the Minimalist Program proposal that would certainly help scare away practically minded computer linguists, but as I said above, I do not think they will be tempted. It makes also the usual points about Chomsky's sloppiness and vagueness. 265