ChakmaNMT: Machine Translation for a Low-Resource and Endangered Language via Transliteration
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866918276874895360 |
|---|---|
| author | Chakma, Aunabil Chakma, Aditya Hasan, Masum Khisa, Soham Tripura, Chumui Shahriyar, Rifat |
| author_facet | Chakma, Aunabil Chakma, Aditya Hasan, Masum Khisa, Soham Tripura, Chumui Shahriyar, Rifat |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_10219 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | ChakmaNMT: Machine Translation for a Low-Resource and Endangered Language via Transliteration Chakma, Aunabil Chakma, Aditya Hasan, Masum Khisa, Soham Tripura, Chumui Shahriyar, Rifat Computation and Language 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. |
| title | ChakmaNMT: Machine Translation for a Low-Resource and Endangered Language via Transliteration |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2410.10219 |