Translating Similar Languages: Role of Mutual Intelligibility in Multilingual Transformers
This work addresses translation challenges for similar language pairs in low-resource settings, but it is incremental as it applies existing methods to a specific task.
The paper tackled low-resource translation between similar languages, finding that back-translation improved performance by over 3 BLEU points and that mutual intelligibility positively correlated with model performance, with Spanish-Catalan achieving the best results.
We investigate different approaches to translate between similar languages under low resource conditions, as part of our contribution to the WMT 2020 Similar Languages Translation Shared Task. We submitted Transformer-based bilingual and multilingual systems for all language pairs, in the two directions. We also leverage back-translation for one of the language pairs, acquiring an improvement of more than 3 BLEU points. We interpret our results in light of the degree of mutual intelligibility (based on Jaccard similarity) between each pair, finding a positive correlation between mutual intelligibility and model performance. Our Spanish-Catalan model has the best performance of all the five language pairs. Except for the case of Hindi-Marathi, our bilingual models achieve better performance than the multilingual models on all pairs.