Abstract
In this thesis, we propose a method for learning a reordering model for BTG-based statistical machine translation (SMT). The model focuses on linguistic features extracted from bilingual phrases. Our method involves extracting reordering examples as well as features such as part-of-speech and word class from aligned parallel sentences. The features are classified with special considerations of phrase lengths. We then use these features to train the Maximum Entropy (ME) reordering model. With the model, we performed Chinese-to-English translation tasks. Experimental results show that our bilingual linguistic model significantly outperforms the state-of-the-art phrase-based and BTG-based SMT systems, measured with BLEU scores. Our methodology not only reduce the feature size by a large margin, compared to previously proposed lexicalized reordering models, but also improves the translation quality.