Automatic Evaluation and Analysis of Idioms in Neural Machine Translation
This work addresses a major open problem in neural machine translation for improving translation quality of idiomatic expressions, though it is incremental as it builds on existing methods with new evaluation techniques.
The authors tackled the problem of literal translation errors for idiomatic expressions in neural machine translation by proposing a novel automatic metric to quantify these errors, and found that monolingual pretraining yields substantial targeted improvements, such as reducing literal translation errors by up to 40% in certain conditions.
A major open problem in neural machine translation (NMT) is the translation of idiomatic expressions, such as "under the weather". The meaning of these expressions is not composed by the meaning of their constituent words, and NMT models tend to translate them literally (i.e., word-by-word), which leads to confusing and nonsensical translations. Research on idioms in NMT is limited and obstructed by the absence of automatic methods for quantifying these errors. In this work, first, we propose a novel metric for automatically measuring the frequency of literal translation errors without human involvement. Equipped with this metric, we present controlled translation experiments with models trained in different conditions (with/without the test-set idioms) and across a wide range of (global and targeted) metrics and test sets. We explore the role of monolingual pretraining and find that it yields substantial targeted improvements, even without observing any translation examples of the test-set idioms. In our analysis, we probe the role of idiom context. We find that the randomly initialized models are more local or "myopic" as they are relatively unaffected by variations of the idiom context, unlike the pretrained ones.