Crossing the Threshold: Idiomatic Machine Translation through Retrieval Augmentation and Loss Weighting
Preprint
Abstract
Idioms are common in everyday language, but often pose a challenge to translators because their meanings do not follow from the meanings of their parts. Despite significant advances, machine translation systems still struggle to translate idiomatic expressions. We provide a simple characterization of idiomatic translation and related issues. This allows us to conduct a synthetic experiment revealing a tipping point at which transformer-based machine translation models correctly default to idiomatic translations. To expand multilingual resources, we compile a dataset of ~4k natural sentences containing idiomatic expressions in French, Finnish, and Japanese. To improve translation of natural idioms, we introduce two straightforward yet effective techniques: the strategic upweighting of training loss on potentially idiomatic sentences, and using retrieval-augmented models. This not only improves the accuracy of a strong pretrained MT model on idiomatic sentences by up to 13% in absolute accuracy, but also holds potential benefits for non-idiomatic sentences.Keywords
All Related Versions
This publication has 15 references indexed in Scilit:
- Automatic Evaluation and Analysis of Idioms in Neural Machine Translation2023
- Understanding Transformer Memorization Recall Through Idioms2023
- Can Transformer be Too Compositional? Analysing Idiom Processing in Neural Machine Translation2022
- The Paradox of the Compositionality of Natural Language: A Neural Machine Translation Case Study2022
- Billion-Scale Similarity Search with GPUsIEEE Transactions on Big Data, 2019
- fairseq: A Fast, Extensible Toolkit for Sequence Modeling2019
- What Level of Quality Can Neural Machine Translation Attain on Literary Text?2018
- Instance Weighting for Neural Machine Translation Domain Adaptation2017
- Idiom Paraphrases: Seventh Heaven vs Cloud Nine2015
- Processing of Idioms by L2 Learners of EnglishTESOL Quarterly, 1999