Stripping Adjectives: Integration Techniques for Selective Stemming in SMT Systems

In this paper we present an approach to reduce data sparsity problems when translating from morphologically rich languages into less inflected languages by selectively stemming certain word types. We develop and compare three different integration strategies: replacing words with their stemmed form, combined input using alternative lattice paths for the stemmed and surface forms and a novel hidden combination strategy, where we replace the stems in the stemmed phrase table by the observed surface forms in the test data. This allows us to apply advanced models trained on the surface forms of the words. We evaluate our approach by stemming German adjectives in two German→English translation scenarios: a low-resource condition as well as a large-scale state-of-the-art translation system. We are able to improve between 0.2 and 0.4 BLEU points over our baseline and reduce the number of out-of-vocabulary words by up to 16.5%.