Morphological constraints for phrase pivot statistical machine translation

Morphological Constraints for Phrase Pivot Statistical Machine Translation Ahmed El Kholy ame2127@columbia.edu Center for Computational Learning Systems, Columbia University Nizar Habash nizar.habash@nyu.edu Computer Science, New York University Abu Dhabi Abstract The lack of parallel data for many language pairs is an important challenge to statistical ma- chine translation (SMT). One common solution is to pivot through a third language for which there exist parallel corpora with the source and target languages. Although pivoting is a ro- bust technique, it introduces some low quality translations especially when a poor morphology language is used as the pivot between rich morphology languages. In this paper, we examine the use of synchronous morphology constraint features to improve the quality of phrase pivot SMT. We compare hand-crafted constraints to those learned from limited parallel data between source and target languages. The learned morphology constraints are based on projected align- ments between the source and target phrases in the pivot phrase table. We show positive results on Hebrew-Arabic SMT (pivoting on English). We get 1.5 BLEU points over a phrase pivot baseline and 0.8 BLEU points over a system combination baseline with a direct model built from parallel data. 1 Introduction One of the main challenges in statistical machine translation (SMT) is the scarcity of parallel data for many language pairs especially when the source and target languages are morphologi- cally rich. A common SMT solution to the lack of parallel data is to pivot the translation through a third language (called pivot or bridge language) for which there exist abundant parallel cor- pora with the source and target languages. The literature covers many pivoting techniques. One of the best performing techniques, phrase pivoting (Utiyama and Isahara, 2007), builds an in- duced new phrase table between the source and target. One of the main issues of this technique is that the size of the newly created pivot phrase table is very large. Moreover, many of the produced phrase pairs are of low quality which affects the translation choices during decoding and the overall translation quality. In this paper, we focus on improving phrase pivoting. We introduce morphology constraint scores which are added to the log linear space of features in order to determine the quality of the pivot phrase pairs. We compare two methods of generating the morphology constraints. One method is based on hand-crafted rules relying on the authors knowledge of the source and target languages; while in the other method, the morphology constraints are induced from Proceedings of MT Summit XV, vol.1: MT Researchers' Track Miami, Oct 30 - Nov 3, 2015 | p. 104 available parallel data between the source and target languages which we also use to build a direct translation model. We then combine both the pivot and direct models to achieve better coverage and overall translation quality. We show positive results on Hebrew-Arabic SMT. We get 1.5 BLEU points over a phrase-pivot baseline and 0.8 BLEU points over a system combination baseline with a direct model built from given parallel …

Content truncated to 500 words