Briefly Noted

In Les linguistiques de corpus, newcomers as well as experts in the field will find a very interesting, well-written, and concise survey of corpus linguistics. The book gives a good overview of the multiple methods used and the problems involved in the creation and the annotation of a corpus. It also presents proposed applications for these corpora. The authors do not relate their own work, but present to the reader in an organized way (by topic) the work of many researchers. The book is written in French, but most of the works presented are based on corpora of English texts. Among other things, the authors discuss what makes a "good corpus," how representative and reliable a corpus can be, and what corpora annotated at the lexical, syntactic, or semantic level presently exist. They analyze the interesting problem of the dependency of a corpus annotation on the foreseen applications for the corpus. The authors present applications such as word sense disambiguation and lexical acquisition from text, but they also present different applications that are of more concern to the literary community, such as discovering changes in the usage of a word between periods or between authors. In the authors' view (which I certainly share and consider very positive), corpus linguistics leaves linguists (and language specialists) and computer scientists no choice but to finally come together and use each other 's strengths (deep understanding of language and deep understanding of computer technology); they must share their expertise to push the field forward.--Caroline Barri~re, School of Information Technology and Engineering, University of Ottawa, Canada Text Retrieval and Filtering: Analytic Models of Performance