Abstract
The standard sets of 8-10 word classes (POS) are defined by a mix of mor- phological, syntactic, and semantic criteria. For some POS the three criteria yield the same result, but POS such as numerals and pronouns end up as heterogeneous classes.
The goal of this contribution is to support the idea of a multidimensional taxonomy of word classes using arguments from the practical domains of corpus and applied linguistics. Annotating corpora with a cross-classifying POS tagset facilitates both cor- pus queries and their use by application tools.
Disparate morphosyntactic annotation of multilingual corpora can be harmonized when the concepts behind language- or theory-specific tagsets are properly located in the 3D space of word classes. Finally, a cross-classification of POS can be applied as a powerful tool for the analysis of texts produced by non-native speakers.