Improving Neural Machine Translation of Indigenous Languages with Multilingual Transfer Learning

Abstract

Machine translation (MT) involving Indigenous languages, including those possibly endangered, is challenging due to lack of sufficient parallel data. We describe an approach exploiting bilingual and multilingual pretrained MT models in a transfer learning setting to translate from Spanish to ten South American Indigenous languages. Our models set new SOTA on five out of the ten language pairs we consider, even doubling performance on one of these five pairs. Unlike previous SOTA that perform data augmentation to enlarge the train sets, we retain the low-resource setting to test the effectiveness of our models under such a constraint. In spite of the rarity of linguistic information available about the Indigenous languages, we offer a number of quantitative and qualitative analyses (e.g., as to morphology, tokenization, and orthography) to contextualize our results.

Temporal Domain Adaptation for Historical Irish

“…Ma et al (2020) achieve a better performance on the low-resource Tibetan language by training cross-lingual Chinese-Tibetan embeddings. Generally, transfer learning is a popular approach in neural machine translation when it comes to the lack of data, as described in Zoph et al (2016); Nguyen and Chiang (2017); Kocmi and Bojar (2018); Maimaiti et al (2019); Chen and Abdul-Mageed (2022). However, the cross-lingual transfer aimed at overcoming data scarcity is not limited to related languages (Adams et al, 2017;Agić et al, 2016).…”