ChakmaNMT: Machine Translation for a Low-Resource and Endangered Language via Transliteration

We present the first systematic study of machine translation for Chakma, an endangered and extremely low-resource Indo-Aryan language, with the goal of supporting language access and preservation. We introduce a new Chakma-Bangla parallel and monolingual dataset, along with a trilingual Chakma-Bangla-English benchmark for evaluation. To address script mismatch and data scarcity, we propose a character-level transliteration framework that exploits the close orthographic and phonological relationship between Chakma and Bangla, preserving semantic content while enabling effective transfer from Bangla and multilingual pretrained models. We benchmark from-scratch MT, fine-tuned pretrained models, and large language models via in-context learning. Results show that transliteration is essential and that fine-tuning and in-context learning substantially outperform from-scratch baselines, with strong asymmetry across translation directions.

• Publication year

  2024

• Venue

  Unknown venue

• Publication date

  2024-10-14

• Fields of study

  Linguistics, Computer Science

• Identifiers
  arXiv 2410.10219
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Self-ChakmaNet: A deep learning framework for indigenous language learning using handwritten characters

  2023cited by this paper
• Scaling Speech Technology to 1, 000+ Languages

  2023cited by this paper
• BanglaNLG and BanglaT5: Benchmarks and Resources for Evaluating Low-Resource Natural Language Generation in Bangla

  2022influential reference
• Language Models are Few-Shot Learners

  2020cited by this paper
• Not Low-Resource Anymore: Aligner Ensembling, Batch Filtering, and New Datasets for Bengali-English Machine Translation

  2020influential reference
• Multilingual Denoising Pre-training for Neural Machine Translation

  2020cited by this paper
• mT5: A Massively Multilingual Pre-trained Text-to-Text Transformer

  2020cited by this paper
• The FLORES Evaluation Datasets for Low-Resource Machine Translation: Nepali–English and Sinhala–English

  2019cited by this paper
• SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural Text Processing

  2018cited by this paper
• A Call for Clarity in Reporting BLEU Scores

  2018cited by this paper
• Trivial Transfer Learning for Low-Resource Neural Machine Translation

  2018cited by this paper
• Iterative Back-Translation for Neural Machine Translation

  2018influential reference
• Attention is All you Need

  2017influential reference
• Six Challenges for Neural Machine Translation

  2017cited by this paper
• Introduction of the Asian Language Treebank

  2016cited by this paper
• Google’s Multilingual Neural Machine Translation System: Enabling Zero-Shot Translation

  2016cited by this paper
• Improving Neural Machine Translation Models with Monolingual Data

  2015cited by this paper
• On Using Monolingual Corpora in Neural Machine Translation

  2015cited by this paper
• chrF: character n-gram F-score for automatic MT evaluation

  2015cited by this paper
• Effective Approaches to Attention-based Neural Machine Translation

  2015cited by this paper
• Neural Machine Translation by Jointly Learning to Align and Translate

  2014cited by this paper
• Hindi-to-Urdu Machine Translation through Transliteration

  2010cited by this paper
• Understanding the difficulty of training deep feedforward neural networks

  2010cited by this paper
• Parallel corpora for medium density languages

  2007cited by this paper
• Statistical Phrase-Based Translation

  2003influential reference

• No citing papers are available for this paper.