Book Reviews: Linguistic Databases

Linguistic Databases is an edited collection of papers on the use of databases in linguistics. It comprises a selection of 12 contributions to the conference with the same title, which was held at the University of Groningen on 23-24 March 1995. The need for data management tools in linguistics is evident. Although collections of linguistic data grew rapidly in the past, the development of suitable database structures and management systems is still in an early stage. The articles presented in the book introduce a variety of approaches to several kinds of applications in different fields of linguistics. They are generally based on existing encoding schemes and data management standards. The fields of study considered here include data-organization approaches to syntactic corpora and phonetic data, the management of theoretical linguistic data, such as syllable structures and nominal argument structures, applications to linguistic problems such as the simplification of texts, and the extension of existing systems and the interaction between them. Two articles consider the management of "test suites" of linguistic phenomena. In the first article, by Stephan Oepen et al., linguistic examples of language-specific phenomena, such as complementation and agreement, were organized in a relational database structure that can be linked to a syntactic parser and grammar. The project emphasizes the development of a consistent annotation scheme and introduces two implementations based on standard tools and a public-domain C library, respectively, based on the commercial DBMS FoxPro (Microsoft). In the second article, the application of SGML for the annotation of linguistic test suites is discussed. The author, Martin Volk, shows the applicability of SGML for this task but also points out problems with redundancy and efficiency when using an SGML annotation scheme. A third article on syntactic data management introduces an approach that combines standard database query systems with SGML-encoded texts. A newly defined query language, SgmlQL, is described, which represents an extension of the standard query language for relational databases SQL. In this language, database queries can be formulated and applied to process hierarchic SGML structures (tree manipulation), for instance, in order to extract information. Furthermore, a freely available prototype was developed that implements a subset of this query language. Four articles deal with phonetic data. Werner Deutsch et al. introduce an implementation of a database management system (S-Tools) for acoustic data. This system includes graphical editors and specialized tools for classifying and handling phonetic data. The implementation is based on existing DBMSs (askSam and MS Access) and is applied to a database of Austrian German and a database of child speech in four