Identifying collocations using cross-lingual association measures

Lis Pereira, Elga Strafella, Kevin Duh, Yuji Matsumoto

Abstract: We introduce a simple and effective cross-lingual approach to identifying collocations. This approach is based on the observation that true collocations, which cannot be translated word for word, will exhibit very different association scores before and after literal translation. Our experiments in Japanese demonstrate that our cross-lingual association measure can successfully exploit the combination of bilingual dictionary and large monolingual corpora, outperforming monolingual association measures.