Alexander Clark
The Handbook of Contemporary Semantic Theory
Shalom Lappin (editor)
School of Oriental and African Studies, University of London
Oxford, UK: Blackwell, 1996,xv + 670pp.
Hardbound, ISBN 0-631-18752-9, $83.95
Reviewed by
Alexander Clark
University of Sussex
This book is a collection of articles by semanticists at the forefront of the field: together they provide an excellent overview of the cutting edge of semantic research. This `Handbook' (at nearly 700 pages and several pounds in weight, it is hardly very handy) is a fascinating snapshot of where the action is in contemporary semantics. Where syntactic formalisms are used, they are generic unification grammars or HPSG.
The book has 22 chapters which are grouped into 11 sections. The book begins with a historical survey of formal semantics as used in linguistics by Barbara H. Partee. This is largely concerned with the legacy of Montague grammar in formal semantics, and provides a very useful intellectual context for the rest of the volume. The next section is entitled `Generalized Quantifier Theory', and consists of two articles. The first, by Edward L. Keenan is an analysis of various formal properties of natural language quantifiers, and the second, by Robin Cooper, is a discussion of the role of situations in generalised quantifier theory. The third section is on the syntax/semantics interface, which starts with Pauline Jacobson on Categorial Grammar, and then has Robert Fiengo and Robert May's paper on coreferencing and coindexing in the use of pronominals. The sixth chapter is by Shalom Lappin :`The Interpretation of Ellipsis'. This is a careful treatment of several different types of ellipsis, including Antecedent Contained Ellipsis, and ellipsis in parasitic gapping constructions. The seventh chapter is by Jeroen Groenendijk, Martin Stokhof and Frank Veltman on Dynamic Semantics. It provides definitions and derivations of the base concepts and then proceeds to deal with a set of notoriously recalcitrant problems of modality and pronouns. This is an admirably concrete introduction to an extremely promising advance in semantics: the idea that `meaning' should be characterised not truth-conditionally but in terms of what is called `information change potential'.
There are then a group of more narrowly focussed papers on key semantic areas such as tense, plurality, focus, presupposition and negation. Craige Roberts' on `Anaphora in Intensional Contexts' is a rehashing of those old favourites, the `donkey' sentences. Jean Mark Gawron's `Quantificational Domains' is a dynamic semantics treatment of quantification, and the subsequent paper by Mats Rooth on `Focus' is a semantical analysis of intonational focus in English. Laurence R. Horn wrote chapter 11 - Presupposition and Implicature, which is a reworking of the Russellian analyses of the presuppositions of noun phrases. The next 5 chapters - `Negation and Polarity Items' by William A. Ladusaw, `Tense and Modality' by Mürvet Enç, `The Semantics of Questions' by James Higginbotham, `Interrogatives: Questions, Facts and Dialogue' by Jonathan Ginzburg, and `Plurality' by Fred Landman are all, as their titles suggest, analyses of particular semantic contructions. I would pick out Higginbotham's and Ginzburg's as being of particular interest to computational linguists -- they are both exhaustive explorations of the possibilities of their respective subjects.
The next paper by John Nerbonne on Computational Semantics, is a careful examination of how theoretical constraints on semantic theories, such as compositionality, need not make any empirical claims about the processing of semantics. That is to say, that having a compositional semantics does not commit one to computing the semantics in a naive, bottom-up way. Next up is a paper by Beth Levin, and Malka Rappaport Hovav on Lexical Semantics. The last section has papers by Ray Jackendoff and Jerrold Katz. Unfortunately they both seem more interested in refighting old debates from the Seventies than contributing to, or even reading, new research. The other papers in this section are by Gila Sher on Logic, and Ruth Kempson on a formal treatment of pragmatics.
The most striking thing about this collection is the uniformly high quality of the contributions. All of the contributors are leaders in the field. All of the papers are not merely standard over-specific conference papers but combine a general overview of the area with a set of specific examples. This combined with the rather encyclopaedic sweep of the book, makes it an enormously useful overview of the whole field. It is particularly interesting to see the growing influence of Amsterdam-style dynamic semantics. The only serious gap in the scope of this collection is that there is no mention or discussion of any of the ideas coming out of connectionist or PDP approaches. Philosophically there is a great deal of interest in this field, and its complete omission is striking. There is, moreover, no work on vagueness or metaphor. There seems to be no recognition of the strongly held belief by many philosophers that these `traditional' symbolic approaches will never succeed. Most of the approaches taken in this book are far from being usable in real systems: researchers looking for semantic formalisms to use in broad coverage systems will be disappointed.
The division of the volume into sections is rather ad hoc : it highlights rather than conceals the rather incoherent nature of the volume. This incoherence is not really a fault - contemporary semanticists are not part of a coherent large-scale research program. For all the cutting edge methods that are applied, it is not clear how this patchwork of specific, admittedly impressive, techniques will fit together into a coherent whole. Nonetheless it will be a useful addition to the library of any semantically inclined computational linguist: it provides an excellent snapshot of current semantical research. The bibliography is excellent, and will prove a valuable research tool for anybody whose work intersects with this field. It does assume a familiarity with the basic ideas of model-theoretic semantics, but a student who has worked through a standard work on Montague Grammar, such as Dowty, Wall and Peters (1981), and has encountered the standard linguistic formalisms would also find the work an extremely useful introduction to the field as a whole and the current state of the art.
Alexander Clark is an M.Sc. student at the School of Cognitive and Computing Sciences, University of Sussex. Clark's address is: School of Cognitive and Computing Sciences, University of Sussex at Brighton, Falmer, Brighton, BN1 9QH, UK;e-mail:alexc@cogs.susx.ac.uk.
