Improving English-Bulgarian statistical machine translation by phrasal verb treatment

This work describes an experimental evaluation of the significance of phrasal verb treatment for obtaining better quality statistical machine translation (SMT) results. Phrasal verbs are multiword expressions used frequently in English, independent of the domain and degree of formality of language. They are challenging for natural language processing due to their idiosyncratic semantic and syntactic properties. The meaning of phrasal verbs is often not directly derivable from the semantics of their constituent tokens. In addition, they are hard to identify in text because of their flexible structure and due to ambiguous prepositional phrase attachments. The importance of the detection and special treatment of phrasal verbs is measured in the context of SMT, where the wordfor-word translation of these units often produces incoherent results. Two ways of integrating phrasal verb information in a phrase-based SMT system are presented. Automatic and manual evaluations of the results reveal improvements in the translation quality in both experiments.