We introduce SPMT, a new class of statistical Translation Models that use Syntactified target language Phrases. The SPMT models outperform a state of the art phrase-based baseline model by 2.64 Bleu points on the NIST 2003 Chinese-English test corpus and 0.28 points on a humanbased quality metric that ranks translations on a scale from 1 to 5. 1

Large-scale discriminative machine translation promises to further the state-of-the-art, but has failed to deliver convincing gains over current heuristic frequency count systems. We argue that a principle reason for this failure is not dealing with multiple, equivalent translations. We present a translation model which models derivations as a latent variable, in both training and decoding, and is fully discriminative and globally optimised. Results show that accounting for multiple derivations does indeed improve performance. Additionally, we show that regularisation is essential for maximum conditional likelihood models in order to avoid degenerate solutions. 1

We present a hierarchical phrase-based translation model which annotates and generalizes existing phrase translations with syntactic categories derived from parsing the target side of a parallel corpus. We associate target parse trees for each training sentence pair with a search lattice constructed from the existing phrase translations on the corresponding source sentence, and consider techniques to produce a syntactically motivated bilingual synchronous grammar. We describe refinements to a chart based decoder and k-best extraction techniques to effectively parse the resulting grammar, which contains up to 4000 syntax-derivated nonterminals, producing translations that achieve significant improvements over Pharaoh, a stateof-the-art phrase based system, on the Europarl French-to-English task (Koehn and Monz, 2005). 1

While phrase-based statistical machine translation systems currently deliver state-of-theart performance, they remain weak on word order changes. Current phrase reordering models can properly handle swaps between adjacent phrases, but they typically lack the ability to perform the kind of long-distance reorderings possible with syntax-based systems. In this paper, we present a novel hierarchical phrase reordering model aimed at improving non-local reorderings, which seamlessly integrates with a standard phrase-based system with little loss of computational efficiency. We show that this model can successfully handle the key examples often used to motivate syntax-based systems, such as the rotation of a prepositional phrase around a noun phrase. We contrast our model with reordering models commonly used in phrase-based systems, and show that our approach provides statistically significant BLEU point gains for two language pairs: Chinese-English (+0.53 on MT05 and +0.71 on MT08) and Arabic-English (+0.55 on MT05). 1

We investigate some pitfalls regarding the discriminatory power of MT evaluation metrics and the accuracy of statistical significance tests. In a discriminative reranking experiment for phrase-based SMT we show that the NIST metric is more sensitive than BLEU or F-score despite their incorporation of aspects of fluency or meaning adequacy into MT evaluation. In an experimental comparison of two statistical significance tests we show that p-values are estimated more conservatively by approximate randomization than by bootstrap tests, thus increasing the likelihood of type-I error for the latter. We point out a pitfall of randomly assessing significance in multiple pairwise comparisons, and conclude with a recommendation to combine NIST with approximate randomization, at more stringent rejection levels than is currently standard.

We investigate the impact of parse quality on a syntactically-informed statistical machine translation system applied to technical text. We vary parse quality by varying the amount of data used to train the parser. As the amount of data increases, parse quality improves, leading to improvements in machine translation output and results that significantly outperform a state-of-the-art phrasal baseline.

We describe a new approach to SMT adaptation that weights out-of-domain phrase pairs according to their relevance to the target domain, determined by both how similar to it they appear to be, and whether they belong to general language or not. This extends previous work on discriminative weighting by using a finer granularity, focusing on the properties of instances rather than corpus components, and using a simpler training procedure. We incorporate instance weighting into a mixture-model framework, and find that it yields consistent improvements over a wide range of baselines. 1

State-of-the-art statistical machine translation is based on alignments between phrases – sequences of words in the source and target sentences. The learning step in these systems often relies on alignments between words. It is often assumed that the quality of this word alignment is critical for translation. However, recent results suggest that the relationship between alignment quality and translation quality is weaker than previously thought. We investigate this question directly, comparing the impact of highquality alignments with a carefully constructed set of degraded alignments. In order to tease apart various interactions, we report experiments investigating the impact of alignments on different aspects of the system. Our results confirm a weak correlation, but they also illustrate that more data and better feature engineering may be more beneficial than better alignment. 1

We describe a multi-step process for automatically learning reliable sub-sentential syntactic phrases that are translation equivalents of each other and syntactic translation rules between two languages. The input to the process is a corpus of parallel sentences, word-aligned and annotated with phrase-structure parse trees. We first apply a newly developed algorithm for aligning parse-tree nodes between the two parallel trees. Next, we extract all aligned sub-sentential syntactic constituents from the parallel sentences, and create a syntax-based phrase-table. Finally, we treat the node alignments as tree decomposition points and extract from the corpus all possible synchronous parallel tree fragments. These are then converted into synchronous context-free rules. We describe the approach and analyze its application to Chinese-English parallel data. 1

Previous work has shown that Chinese word segmentation is useful for machine translation to English, yet the way different segmentation strategies affect MT is still poorly understood. In this paper, we demonstrate that optimizing segmentation for an existing segmentation standard does not always yield better MT performance. We find that other factors such as segmentation consistency and granularity of Chinese “words ” can be more important for machine translation. Based on these findings, we implement methods inside a conditional random field segmenter that directly optimize segmentation granularity with respect to the MT task, providing an improvement of 0.73 BLEU. We also show that improving segmentation consistency using external lexicon and proper noun features yields a 0.32 BLEU increase. 1