Wilks sets out to discuss deep and superficial parsers and to show that “they are not as different as might appear.” For him, deep systems “avoid conventional syntax in general, and aim for semantic representation,” whereas superficial systems “are committed to . . . language as a flat linear phenomenon.” The bulk of the chapter is a survey of Riesbeck’s parser [1], Wilks’ parser (see, e.g., [2]), and Wilensky and Arens’ PHRAN [3]. It points out some theoretical and practical shortcomings of each, for example, the surreptitious and inadequate handling of syntax by conceptual analyzers. Wilks briefly addresses many issues in parsing algorithms and the representation and use of syntactic knowledge, world knowledge, and parsing.

The chapter does usefully compare the systems mentioned, although the failure to contrast them with syntax-based parsers is unfortunate. The distinction between deep and superficial is confusing and useless, and Wilks has “no very clear conclusions to offer,” but many of the observations are thought-provoking. The presentation assumes an advanced reader. Indeed, some comments will be opaque to anyone without detailed knowledge of the systems discussed.