Bilingual lexicon extraction from comparable corpora for closely related languages