Among the increasing number of books devoted to Prolog and logic programming, there are scarcely any that do not offer at least a few exercises to illustrate the possibility of analyzing some fragments of natural language (NL). On the other hand, many meetings--and their proceedings--have included specific discussions of the use of logical methods in computational linguistics (for example, see [1]).

This book, by Fernando C.N. Pereira and Stuart M. Shieber, is, to my knowledge, the first monograph in which the reader can find a systematical and virtually complete introduction to the logical processing of NL--with the distinguished exception of [2], which had a more theoretical purpose about the properties of logical grammars.

Pereira and Shieber provide a two-part introduction: first, to Prolog and logic programming for NL, extended in some examples to the writing of interpreters for logical grammars; and second, to unification-based grammars, specifically to definite-clause grammars (DCGs), which are presented as a linguistic theory providing a unitary formal framework for the description of the syntactic behavior of sentence constituents, the pertinent features of the lexicon, and for the expression of a compositional semantics. Solutions are given for some problems about the scope of quantifiers and reference attribution for descriptions. The relation between syntax and semantics follows from Richard Montague’s ideas (cf. [3]).

In fact, the computational and linguistic components are interwoven thoughout the book. After an intuitive exposition of the bases for unification and resolution, the authors show how these procedures can be used to build parsers for restricted fragments of English. The account of Prolog and its programming techniques gains ground with the complexity of the grammars necessary to describe more extended fragments of NL. In this way, the close interdependence between linguistic targets and occupational investment is clearly defined. Unfortunately, the crucial need for vast amounts of basic linguistic knowledge, as, for instance, the description of specific categories of verbs or adjectives [4] is not made tangible to the reader.

As was pointed out before, the book starts with elementary notions in logic, logic programming, and linguistics and winds up with quite sophisticated concepts (e.g. beta-reduction and gapping grammars) and programs. The whole development is well justified by intuition and experimental considerations. In most cases, formalization is circumscribed to grammars and programs.

This choice is pertinent for the didactical purpose of the book. It is accessible, but the reader should have at least an elementary knowledge of first order calculus and the formal syntax and semantics of NL. In that case, the reader will have direct access to the practice of programming in Edinburgh Prolog.

All the chapters have a problem section. Some of these provide useful complements to the exposition. Unfortunately, the authors give no solutions even for the most difficult exercises. This is the main limitation to the didactical excellence of the book and quite probably it will be amended in future editions.

The reader can find in the bibliography most of the English references for the active research field of logic computational linguistics. This coverage is not as good for work from Japan and Continental Europe, especially Germany [e.g., 5] and France [e.g., 6].

In summary, with this book, Pereira and Shieber will certainly dramatically increase the ranks of logic enthusiasts.