# On Romanization for Model Transfer Between Scripts in Neural Machine Translation

arXiv: https://arxiv.org/abs/2009.14824 · HTML: https://mlsanity.com/paper/2009.14824/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 30 Sep 2020.

## Built on by (most cited first, 9 of 9)

- [Language Modelling with Pixels](https://mlsanity.com/paper/2207.06991/) (2022)
- [A Large-Scale Study of Machine Translation in the Turkic Languages](https://mlsanity.com/paper/2109.04593/) (2021)
- [Romanization-based Large-scale Adaptation of Multilingual Language Models](https://mlsanity.com/paper/2304.08865/) (2023)
- [Does Transliteration Help Multilingual Language Modeling?](https://mlsanity.com/paper/2201.12501/) (2022)
- [Subword Mapping and Anchoring across Languages](https://mlsanity.com/paper/2109.04556/) (2021)
- [Towards a Better Understanding of Variations in Zero-Shot Neural Machine Translation Performance](https://mlsanity.com/paper/2310.10385/) (2023)
- [Beyond Shared Vocabulary: Increasing Representational Word Similarities across Languages for Multilingual Machine Translation](https://mlsanity.com/paper/2305.14189/) (2023)
- [Assessing the Importance of Frequency versus Compositionality for Subword-based Tokenization in NMT](https://mlsanity.com/paper/2306.01393/) (2023)
- [Enhancing Cross-Lingual Transfer through Reversible Transliteration: A Huffman-Based Approach for Low-Resource Languages](https://mlsanity.com/paper/2509.17493/) (2025)

## Abstract (arXiv)

Transfer learning is a popular strategy to improve the quality of low-resource machine translation. For an optimal transfer of the embedding layer, the child and parent model should share a substantial part of the vocabulary. This is not the case when transferring to languages with a different script. We explore the benefit of romanization in this scenario. Our results show that romanization entails information loss and is thus not always superior to simpler vocabulary transfer methods, but can improve the transfer between related languages with different scripts. We compare two romanization tools and find that they exhibit different degrees of information loss, which affects translation quality. Finally, we extend romanization to the target side, showing that this can be a successful strategy when coupled with a simple deromanization model.