How the corpus-based Basque Verb Index lexicon was built

This article describes the method used to build the Basque Verb Index (BVI), a corpus-based lexicon. The BVI is the result of semiautomatic annotation of the EPEC corpus with verb predicate information, following the PropBank-VerbNet model. The method presented is the product of a deep study of the syntactic–semantic behaviour of verbs in EPEC-RolSem (the EPEC corpus tagged with verb predicate information). During the process of annotating EPEC-RolSem, we have identified and stored in the BVI lexicon the different role-patterns associated with all verbs appearing in the corpus. In addition, each entry in the BVI is linked to the corresponding verb entry in well-known resources such as PropBank, VerbNet, WordNet and FrameNet. We have also implemented a tool called e-ROLda to facilitate the process of looking up verb patterns in the BVI and examples in EPEC-RolSem as a basis for future studies.