# Diacritic Restoration for Low-Resource Indigenous Languages: Case Study with Bribri and Cook Islands M=aori

arXiv: https://arxiv.org/abs/2512.19630 · HTML: https://mlsanity.com/paper/2512.19630/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 10 of its 63 references that are in the map.
- First posted on arXiv 22 Dec 2025.

## Builds on (most cited first, 10 of 10)

- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural Text Processing](https://mlsanity.com/paper/1808.06226/) (2018)
- [The State and Fate of Linguistic Diversity and Inclusion in the NLP World](https://mlsanity.com/paper/2004.09095/) (2020)
- [ByT5: Towards a token-free future with pre-trained byte-to-byte models](https://mlsanity.com/paper/2105.13626/) (2021)
- [MLS: A Large-Scale Multilingual Dataset for Speech Research](https://mlsanity.com/paper/2012.03411/) (2020)
- [AISHELL-3: A Multi-speaker Mandarin TTS Corpus and the Baselines](https://mlsanity.com/paper/2010.11567/) (2020)
- [How can NLP Help Revitalize Endangered Languages? A Case Study and Roadmap for the Cherokee Language](https://mlsanity.com/paper/2204.11909/) (2022)
- [A Multitask Learning Approach for Diacritic Restoration](https://mlsanity.com/paper/2006.04016/) (2020)
- [Evaluating Pixel Language Models on Non-Standardized Languages](https://mlsanity.com/paper/2412.09084/) (2024)
- [Don't Touch My Diacritics](https://mlsanity.com/paper/2410.24140/) (2024)

## Abstract (arXiv)

We present experiments on diacritic restoration, a form of text normalization essential for natural language processing (NLP) tasks. Our study focuses on two extremely under-resourced languages: Bribri, a Chibchan language spoken in Costa Rica, and Cook Islands M=aori, a Polynesian language spoken in the Cook Islands. Specifically, this paper: (i) compares algorithms for diacritics restoration in under-resourced languages, including tonal diacritics, (ii) examines the amount of data required to achieve target performance levels, (iii) contrasts results across varying resource conditions, and (iv) explores the related task of diacritic correction. We find that fine-tuned, character-level LLMs perform best, likely due to their ability to decompose complex characters into their UTF-8 byte representations. In contrast, massively multilingual models perform less effectively given our data constraints. Across all models, reliable performance begins to emerge with data budgets of around 10,000 words. Zero-shot approaches perform poorly in all cases. This study responds both to requests from the language communities and to broader NLP research questions concerning model performance and generalization in under-resourced contexts.