Electronic Dictionaries, Tagsets and Tagging

This monograph examines the state of the art in the area of electronic dictionaries and in particulur their application in natural language processing tasks. The book reviews the (brief) history of electronic dictionaries, the approaches and underlying assumptions of existing attempts at exhaustive lexical coverage for languages like English, German, French and others. The author also seriously challenges the adequacy of current practices in the area of tagging; it is shown that not only is the information in tagsets far from sufficient, the fact that tagging methods are typically restricted to single word forms and to syntactic categories (instead of to complex lexical units and their semantic categories) diminishes their usefulness tremendously when it comes to parsing sentences. Given the kind of lexical units used in tagging, it comes as no surprise that very few parsing results can really be considered to be correct. The book also contains interesting new statistics about the distribution of lexical units in large corpora and the problem of how to deal with ambiguity in texts.