I have often met linguists and computational linguists willing to read a short, simple, and well-documented textbook about the use of Prolog for natural-language understanding. I feel that this well-written book mainly devoted to Prolog and logic grammars to a large extent meets the challenge. I will here summarize and com­ment on each chapter, and conclude by pointing out aspects I would have liked to see included in this introductory book.

Besides the presentation of concepts and examples, each chapter contains several exercises left to the reader and exhaustive bibliographic notes. I only regret that no answers to the exercises are provided as is usually the case for textbooks. These exercises are often nontrivial for the beginner and commented an­swers would have made them much more attractive and useful. Another interesting point is the internal struc­ture of each chapter: two themes (Prolog and logic grammars) are tackled in a parallel and progressive way.

Chapter 1 is a very short introduction to logic pro­gramming, Prolog, and natural language. It also includes a list of prerequisites (which does not constitute a real constraint for a novice in the area) and some general historical material about the field and related fields.

Chapter 2 begins with an introduction to Prolog, viewed as a means to encode knowledge in a declarative way. The presentation is simple and classical. Section 2.3, about the logic of Prolog, is probably the best presentation of the topic I have ever seen, apart from the excellent book The Art of Prolog (Sterling and Shapiro 1986). To conclude this brief introduction, a problem section deals with the representation of seman­tic networks. This chapter ends by the development of an implementation method for context-free grammars in Prolog. A trace of the parsing of a short sentence helps the user to understand how the proof procedure of Prolog is used as a parsing mechanism.

Chapter 3, entitled "Pure Prolog: Theory and appli­cation," observes the same structure as the previous chapter. It first introduces more complex concepts, techniques, and data-structures of Prolog, illustrated by typical short programs, most of them belonging to the classics of the field. Then, more material about formal aspects of Prolog (e.g., substitutions and unification, resolution and the semantics of Prolog) is given. Finally, definite-clause grammars (DCGs) are introduced. They constitute the basic form of logic grammars from which several formalisms have been developed, motivated either by linguistic considerations or by the need of providing grammar writers with more transparent for­mulations of grammars. DCGs have themselves partly emerged from metamorphosis grammars (Colmerauer 1978), which is a general framework to deal with type-0 grammars. Besides a basic implementation of DCGs, the authors introduce fundamental techniques to build parsing trees, to deal with PP-attachment and to express different kinds of agreements.

Chapter 4 is divided into two parts: some very basic principles for semantic interpretation are given, and then the syntactic coverage of the grammar given in the previous chapter is extended. The first section of this chapter shows how the basic principles of computa­tional semantics are dealt with in Prolog by means of logical variables. The authors then focus on quantified noun phrases and quantifier scoping. Although the programming techniques given in this latter section are of much interest, the idea of generating all possible quantifier scopings without any further refinements seems to me to be a little too superficial, given the actual state of the art. Next, the syntactic coverage of the previous grammar is extended. Of particular interest is the way filler-gap dependencies are expressed in an efficient way for simple linguistic constructions. The technique of difference lists is then introduced to permit an efficient treatment of those dependencies. The im­plementation of simple island constraints is also given to emphasize the relevance of the difference-list tech­nique. Finally, the problem section addresses grammar extensions such as noun complements and modifiers. This chapter is the central part of the book and provides a good basis for the reader who is willing to write his own grammar. However, I feel that the semantic-interpretation section should have been a little more developed, since 1. the main goal of language process­ing is to understand a sentence or a text via its semantic representation, and 2. the semantic interpretation of other constructions like modifiers requires improve­ments of the basic construction method presented here and would be worth presenting at this level! The prob­able reason for this lack of development is that rela­tively few works have been undertaken in the area of semantic representation and interpretation using Pro­log.

Chapter 5 introduces metalogical facilities offered by standard Prolog (e.g., built-in predicates like call, cut, set of) and then proposes a simple dialogue program. Chapter 6 is entirely devoted to the construction of interpreters for DCGs. Interpreters mainly permit one to circumvent computational problems like left-recur­sion in grammar rules, implementing parsing strategies

different from the strategy imposed by Prolog, and partly automating the writing of rules. For example, arguments having a regular and predictable structure (parse tree building, logical form construction, etc.) can be automatically included in grammar rules. Such inter­preters are written in Prolog and they transform a Prolog-like program into one that is directly executable. Fundamental techniques are proposed and illustrated for DCGs, such as automated construction of proof trees, interpretation of filler-gap dependencies, partial execution, and tabular parsing (i.e., chart parsing). The book ends with annotated sample programs that inte­grate most of the material presented in the preceding chapters. The bibliography is relevant and comprehen­sive, given the aims of the book.

To conclude, I feel that this book includes most of the material one would wish to see in an introductory book. By a small number of linguistic concepts and constructions, the authors present the main techniques a grammar writer can use in a large number of situa­tions. There are, however, some points I would have liked to see included in this book. Very few parsers directly use the linguistic framework proposed here but rather are inspired by GPSGs, LFGs, HPSGs, UGs, etc. Thus it would have been of particular interest to have guidelines on Prolog techniques useful to such computational-linguistics frameworks. Some aspects, like feature percolation and control in GPSGs, are probably not trivial to deal with.

In spite of these nonessential restrictions, I feel that this book should be read by everyone in the field having a minimum interest in Prolog. This very convincing book also suggests the reading of more complex publi­cations in the field like Dahl and St.-Dizier (1985, 1988) and JLP (1986).