We present Minimum Bayes-Risk (MBR) decoding for statistical machine translation. This statistical approach aims to minimize expected loss of translation errors under loss functions that measure translation performance. We describe a hierarchy of loss functions that incorporate different levels of linguistic information from word strings, word-to-word alignments from an MT system, and syntactic structure from parse-trees of source and target language sentences. We report the performance of the MBR decoders on a Chinese-to-English translation task. Our results show that MBR decoding can be used to tune statistical MT performance for specific loss functions. 1

Translation systems are complex, and most metrics do little to pinpoint causes of error or isolate system differences. We use a simple technique to discover induction errors, which occur when good translations are absent from model search spaces. Our results show that a common pruning heuristic drastically increases induction error, and also strongly suggest that the search spaces of phrase-based and hierarchical phrase-based models are highly overlapping despite the well known structural differences. 1

Hypergraphs are used in several syntaxinspired methods of machine translation to compactly encode exponentially many translation hypotheses. The hypotheses closest to given reference translations therefore cannot be found via brute force, particularly for popular measures of closeness such as BLEU. We develop a dynamic program for extracting the so called oracle-best hypothesis from a hypergraph by viewing it as the problem of finding the most likely hypothesis under an n-gram language model trained from only the reference translations. We further identify and remove massive redundancies in the dynamic program state due to the sparsity of n-grams present in the reference translations, resulting in a very efficient program. We present runtime statistics for this program, and demonstrate successful application of the hypotheses thus found as the targets for discriminative training of translation system components. 1

This paper proposes cross-lingual language modeling for transcribing source resourcepoor languages and translating them into target resource-rich languages if necessary. Our focus is to improve the speech recognition performance of low-resource languages by leveraging the language model statistics from resource-rich languages. The most challenging work of cross-lingual language modeling is to solve the syntactic discrepancies between the source and target languages. We therefore propose syntactic reordering for cross-lingual language modeling, and present a first result that compares inversion transduction grammar (ITG) reordering constraints to IBM and local constraints in an integrated speech transcription and translation system. Evaluations on resource-poor Cantonese speech transcription and Cantonese to resource-rich Mandarin translation tasks show that our proposed approach improves the system performance significantly, up to 3.4 % relative WER reduction in Cantonese transcription and 13.3 % relative bilingual evaluation understudy (BLEU) score improvement in Mandarin transcription compared with the system without reordering. 1