Learning bilingual lexicons from comparable English and Spanish corpora

Abstract

Research on extraction of word translation from comparable, nonparallel texts has not been very popular because it produces poor results when compared to those obtained from aligned parallel corpora. Whereas for parallel texts, word translation extraction can reach about 99%, the accuracy for comparable corpora has been around 72 % up to now. The current approach, which relies not on a bilingual dictionary but on the previous extraction of bilingual information from parallel corpora, makes a significant improvement to about 79 % of words translations identified correctly. 1.