Word sense disambiguation is necessary in translation because different word senses often have different translations. Neural machine translation models learn different senses of words as part of an endto-end translation task, and their capability to perform word sense disambiguation has so far not been quantified. We exploit the fact that neural translation models can score arbitrary translations to design a novel cross-lingual word sense disambiguation task that is tailored towards evaluating neural machine translation models. We present a test set of 7,200 lexical ambiguities for German→English, and 6,700 for German→French, and report baseline results. With 70% of lexical ambiguities correctly disambiguated, we find that word sense disambiguation remains a challenging problem for neural machine translation, especially for rare word senses. To improve word sense disambiguation in neural machine translation, we experiment with two methods to integrate sense embeddings. In a first approach we pass sense embeddings as additional input to the neural machine translation system. For the second experiment, we extract lexical chains based on sense embeddings from the document and integrate this information into the NMT model. While a baseline NMT system disambiguates frequent word senses quite reliably, the annotation with both sense labels and lexical chains improves the neural models' performance on rare word senses.

We implement a fully probabilistic model to combine the hypotheses of a Spanish anaphora resolution system with those of a Spanish-English machine translation system. The probabilities over antecedents are converted into probabilities for the features of translated pronouns, and are integrated with phrase-based MT using an additional translation model for pronouns. The system improves the translation of several Spanish personal and possessive pronouns into English, by solving translation divergencies such as ella → she | it or su → his | her | its | their. On a test set with 2,286 pronouns, a baseline system correctly translates 1,055 of them, while ours improves this by 41. Moreover, with oracle antecedents, possessives are translated with an accuracy of 83%.

We built a pipeline to normalize Quechua texts through morphological analysis and disambiguation. Word forms are analyzed by a set of cascaded finite state transducers which split the words and rewrite the morphemes to a normalized form. However, some of these morphemes, or rather morpheme combinations, are ambiguous, which may affect the normalization. For this reason, we disambiguate the morpheme sequences with conditional random fields. Once we know the individual morphemes of a word, we can generate the normalized word form from the disambiguated morphemes. 1

This paper presents a straightforward method to integrate co-reference information into phrase-based machine translation to address the problems of i) elided subjects and ii) morphological underspecification of pronouns when translating from pro-drop languages. We evaluate the method for the language pair SpanishEnglish and find that translation quality improves with the addition of co-reference information.

No abstract