ROG – A Paradigmatic Morphological Generator for Romanian

Abstract

We describe here the process of implementing a morphological generator for Romanian, together with references to the resources and previous work that made this achievement possible. The generator receives a lemma and a morpho-lexical description of the intended word-form and it is supposed to generate that very word-form. It uses a paradigmatic description of Romanian morphology and a large dictionary of lemmas associated with their inflectional paradigms. The work is motivated by an ongoing example-based machine translation system and by its utility in extending the lexical resources created in our group. The generation process distinguishes two major situations, depending on the lemma being recorded or not into the generator's dictionary. In the first case, the generated inflected form is guaranteed to be a real word of Romanian language while in the second case the inflected form might be a potential but not extant word of Romanian.