Search papers, labs, and topics across Lattice.
This paper introduces a Poly-Dialectal Neural Machine Translation System designed to address the challenges posed by the diverse regional dialects of Bangla, which significantly differ from Standard Colloquial Bangla. By compiling the largest multi-dialect parallel corpus for Bangla, consisting of over 51,000 sentence pairs, the authors fine-tune the BanglaT5 model using Weight-Decomposed Low-Rank Adaptation (DoRA) to achieve state-of-the-art translation performance. The results indicate that the system not only surpasses existing models like NLLB-200 and mBART-50 but also facilitates effective cross-dialectal transfer, promoting digital inclusion for marginalized communities.
Achieving a BLEU score of 29.26, this system outperforms leading models while translating across 12 distinct Bangla dialects without relying on a standard pivot.
Regional dialectal variation poses a fundamental challenge to natural language processing (NLP) in Bangla, where over 240 million speakers communicate across diverse regional variants that diverge significantly from Standard Colloquial Bangla (SCB) in phonology, morphology, and lexicon. Contemporary neural machine trans- lation (NMT) architectures and large language models (LLMs) predominantly as- sume a homogeneous language distribution, resulting in severe performance degra- dation when translating low-resource regional dialects. In this work, we present a unified Poly-Dialectal Neural Machine Translation System capable of multi-directional translation across 12 Bangla regional dialects without routing through an inter- mediary standard pivot. We compile the largest multi-dialect parallel corpus for Bangla to date, comprising 51,531 non-null parallel sentence pairs across 12 di- alects, incorporating 2,500 expert-verified, bidirectional parallel sentence pairs for five previously unaddressed dialects. Evaluating sequence-to-sequence architec- tures under Weight-Decomposed Low-Rank Adaptation (DoRA), our fine-tuned BanglaT5 model achieves state-of-the-art translation performance (29.26 BLEU, 57.26 chrF++), outperforming NLLB-200 (615M) and mBART-50 (611M) while preserving morphological coherence. Furthermore, we conduct a systematic cross- dialectal transfer analysis and dataset scaling study, establishing empirical thresh- olds for low-resource dialect adaptation. Finally, we deploy the optimized INT8- quantized model as an open-access web application to promote digital inclusion for marginalized dialect communities. The complete dataset is publicly available at Mendeley Data (https://data.mendeley.com/datasets/v9cf66fk2t/2).
