Browse Books

Reviewer: Fred J. Damerau

Intended for academic linguists who might want to move into industry, this book assumes significant linguistic sophistication, which limits its audience somewhat. Aside from the first introductory chapter and the last summary chapter, each of the chapters has a different author and covers a specific topic. Chapter 2 is a discussion of linguistic domains and their distinguishing features, and also provides advice on how to go about discovering the topics a domain covers. Chapter 3 is an introduction to the Internet and the applications that use it, especially email, chat, and the World Wide Web. For those who use Internet functions like email essentially by rote, reading this chapter should take these topics out of the realm of magic. Chapter 4 is an introduction to the techniques and processes involved in the production of large grammars. The emphasis is on lexical functional grammar (LFG), but the discussion is general enough that previous knowledge of LFG is not really necessary. In spite of a certain looseness of language (“not all taggers are stochastic” (page 153)), this is still a good introduction to read before embarking on a large project for the first time. Chapter 5, “Ontologies,” is a high-level description with some examples. It discusses principles, some example projects, and available tools. This is a good introduction, marred by some distracting typographical errors. As its final summary states, chapter 6, “Text Mining, Corpus Building and Testing,” is largely concerned with contrasting corpus and theoretical linguistics. It provides a brief description of topics such as corpora, corpus annotation, tagging, disambiguation, and partial parsing. Chapter 7, “Statistical Natural Language Processing,” (NLP) is a surface view of how statistical methods are used for some common NLP tasks. There are a number of instances where technical terms are used without definition or reference. For example, “this greediness helps prevent cotraining from falling into a suboptimal local minimum” is probably gibberish to those without computer science and mathematics training. Chapter 8 discusses knowledge representation. In spite of examples, the chapter is quite abstract, and attempts to cover many difficult issues in a rather small space. Its main virtue is its extensive references to primary work. Chapter 9 is a concise description of speech recognition problems with a clear explanation of how existing systems work. There is probably more mathematical content here than the average linguist will be able to follow. Overall, I can recommend this book to its intended audience of theoretically minded linguists. Those with only a limited knowledge of linguistic theory will find some of the discussion opaque, but can still profit from parts of the book. The bibliographies after each chapter are particularly useful. Online Computing Reviews Service