From the earliest days of computer science, researchers have tried to develop computational models of language. They had initial success with formal theories of language, which were useful in producing programming languages. Natural language processing proved to be a less tractable problem, however. Early work in machine translation was considered a failure [1]. Theories for the syntax of natural language did little to address the hard problems of semantics--the underlying meaning of language.

This volume emphasizes the successes: the syntactic side of language. Most of the book is devoted to formal language theory. The author pays particular attention to two-level grammars for English.

At the beginning of chapter 6, on page 299, we read:

In the first five chapters, the focus of discussion was almost completely on the processing of syntactic structure. Yet, if one thinks about language as a means for communication, this emphasis may appear to be somewhat paradoxical. After all, reading sentences in order to verify grammaticality must be at least a means to an end or ends.

The author goes on to discuss the use of symbolic logic to interpret natural language. Finally, chapter 7 reviews some of the related work in question answering, including Cordell Green’s BASEBALL, William Woods’s LUNAR, and Jaime Carbonell’s XCALIBUR. A collection of exercises geared to the preceding seven chapters appears as an appendix.

Throughout the book, the discussion is largely formal and abstract. No computer programs appear. The book does contain a number of algorithms, which are presented formally, not procedurally.

The book is intended as a text for a course in natural language processing for advanced undergraduates or graduate students not only in computer science but in psychology and linguistics.

I readily concede that syntax has played a major role in natural language processing research. Nonetheless, a book titled Computer processing of natural language should spend more time discussing computational techniques for understanding language.

By placing major emphasis on syntactic descriptions of language, this text may cause readers to lose sight of the primary function of language, which is to communicate. Thus, the following judgment (which appears to be a Freudian slip) displays an uncanny insight: “The impact of generative grammar on linguistic research within the past decade or so cannot be underestimated” [2].