Full-coverage Identification of English Light Verb Constructions

The identification of light verb constructions (LVC) is an important task for several applications. Previous studies focused on some limited set of light verb constructions. Here, we address the full coverage of LVCs. We investigate the performance of different candidate extraction methods on two English full-coverage LVC annotated corpora, where we found that less severe candidate extraction methods should be applied. Then we follow a machine learning approach that makes use of an extended and rich feature set to select LVCs among extracted candidates.