Linguistic exploitation of syntactic databases: The use of the Nijmegen Linguistic Database program By Hans van Halteren and Theo van den HeuVEL (review)

BOOK NOTICES 869 also an appendix containing a glossary, references , and index. The book is poorly written and edited and somewhat difficult to understand. Examples are frequently unglossed, which may make the data difficult to understand unless one already knows the language. There are also minor problems with the content. For instance, the author asserts that 'there is ample reason to believe' that language parsing operates from left to right, 'considering that we write in that direction' (157 and also n. 53), without mentioning the existence ofwriting systems such as those of Arabic and Hebrew. There is also a statement that English has no Inversion rule of the type responsible for sentences like German Im Garten liegt eine Puppe 'In the garden lies a doll', overlooking examples such as Joseph E. Emonds' At none of the beaches are the lifeguards alert (A transformational approach to English syntax . New York: Academic Press, 1976, p. 28). 'Eurogrammar' is an odd term for Germanic and Romance, since it implies that all European languages belong to these two groups. As for the overall purpose, the author claims that translators, for example, will benefit from access to this type of work, and cites studies to support this claim. However, I can't imagine the wider audience profiting from the work, apart from the data. I expect it to be much more useful to theoretical syntacticians, including comparativists and typologists. The real strength of the book lies in the large quantity of data it presents, and it is valuable for this reason alone. [Carolyn Harford, Michigan State University.] Linguistic exploitation of syntactic databases : The use of the Nijmegen Linguistic Database program. By Hans van Halteren and Theo van den HeuVEL. (Language and computers : Studies in practical linguistics , 5.) Amsterdam & Atlanta, GA: Rodopi, 1990. Pp. 207. Cloth $33.50. The Nijmegen Corpus is a body ofrecent British English prose approximately 140,000 words long. It is distinguished from other English corpora such as Brown and Lancaster-Oslo/Bergen by the fact that it has been given a full syntactic analysis in tree form. The incorporation ofthis corpus with detailed operational software constitutes the Nijmegen Linguistic Database (abbreviated LDB). This book is a detailed manual of instructions in the use of LDB for various kinds oflinguistic analysis made possible by the ability to specify all kinds of grammatical constituents for comparison, frequency counts, and other statistical operations. The book is divided into two sections, a 'Tutorial Part' of four chapters and a 'Reference Part' of two, plus a brief bibliography and full index. As the subtitles indicate, the Tutorial Part is a 'how-to' section, taking the reader step by step through various processes of using the LDB, and the Reference Part is a description of the powers of the software—what it can do when properly used. There is inevitably considerable overlap between the two parts, since various aspects of the programs are treated in each, though from different points of view. Presumably the authors' intent was for readers to work their way through the Tutorial Part, preferably with the LDB on a minicomputer, and to use the Reference Part to check on various functions and refresh their memories on how to use them. A demonstration diskette, furnished with the book, gives the user the option of watching an automatic demonstration or taking part in an interactive one. The syntactic analysis is in effect an immediate constituent structure, with some provision for discontinuous constituents but otherwise displaying surface structure. Every sentence (or 'utterance') is in the form of a tree, with the root node to the left and numbered branches indicating constituents ending in 'leaves', the ultimate words. In the case of long sentences, the display may be scrolled vertically and horizontally . Provision is made to focus on any node, which is highlighted on the screen and described in terms of its components. Two modes of viewing the structure are provided, the 'tree view' and the 'pattern view'. The latter gives the major components by function and category. Thus the pattern CAT = 'NP' FUN PREM' FUN = 'POM' describes a noun phrase containing a premodi- 870 LANGUAGE, VOLUME 68, NUMBER 4 (1992) fier and a postmodifier, as in...