A discriminative syntactic model for source permutation via tree transduction

A major challenge in statistical machine translation is mitigating the word order differences between source and target strings. While reordering and lexical translation choices are often conducted in tandem, source string permutation prior to translation is attractive for studying reordering using hierarchical and syntactic structure. This work contributes an approach for learning source string permutation via transfer of the source syntax tree. We present a novel discriminative, probabilistic tree transduction model, and contribute a set of empirical upperbounds on translation performance for English-to-Dutch source string permutation under sequence and parse tree constraints. Finally, the translation performance of our learning model is shown to outperform the state-of-the-art phrase-based system significantly.