MultiWord Lexical Units in EUSLEM, a lemmatiser-tagger for Basque

A lemmatiser-tagger must not only lemmatise word-forms consisting of a single lexical element but it must also be able to detect complex units. In this paper we try to delimit linguistically which complex words deserve to be lemmatised as a unit. Then, we propose a formal description for MultiWord Lexical Units (MWLU) in Basque ---resulting of a conscientious analysis of their syntactic and morphological behaviour. Based on that formal description we propose a simple logical formalism to represent those MWLUs so that they can be automatically processed. 1. Introduction A lemmatiser-tagger is a computational tool used for assigning the correct lemma and grammatical category to each token of a corpus. It is a basic device for corpus analysis, automatic indexation, syntactic and semantic analyses etc. For example, the lemmatiser-tagger for Basque (EUSLEM) (Aduriz et al., 96a) is essential for the second phase of the Systematic Compilation of Modern Basque 1 (EEBS) project (Urkia et a..