# Benchmarking Bengali Dialectal Bias: A Multi-Stage Framework Integrating RAG-Based Translation and Human-Augmented RLAIF

arXiv: https://arxiv.org/abs/2603.21359 · HTML: https://mlsanity.com/paper/2603.21359/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Human values and AI feedback alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 20 of its 40 references that are in the map.
- First posted on arXiv 22 Mar 2026.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Human values and AI feedback alignment](https://mlsanity.com/alignment/human-values-and-ai-feedback-alignment/)

## Builds on (most cited first, 20 of 20)

- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment](https://mlsanity.com/paper/2303.16634/) (2023)
- [A Survey on LLM-as-a-Judge](https://mlsanity.com/paper/2411.15594/) (2024)
- [BLEURT: Learning Robust Metrics for Text Generation](https://mlsanity.com/paper/2004.04696/) (2020)
- [COMET: A Neural Framework for MT Evaluation](https://mlsanity.com/paper/2009.09025/) (2020)
- [Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedback](https://mlsanity.com/paper/2305.14975/) (2023)
- [How Contextual are Contextualized Word Representations? Comparing the Geometry of BERT, ELMo, and GPT-2 Embeddings](https://mlsanity.com/paper/1909.00512/) (2019)
- [LLM Evaluators Recognize and Favor Their Own Generations](https://mlsanity.com/paper/2404.13076/) (2024)
- [From Generation to Judgment: Opportunities and Challenges of LLM-as-a-judge](https://mlsanity.com/paper/2411.16594/) (2024)
- [Linguistic Bias in ChatGPT: Language Models Reinforce Dialect Discrimination](https://mlsanity.com/paper/2406.08818/) (2024)
- [Dialect prejudice predicts AI decisions about people's character, employability, and criminality](https://mlsanity.com/paper/2403.00742/) (2024)
- [Judge's Verdict: A Comprehensive Analysis of LLM Judge Capability Through Human Agreement](https://mlsanity.com/paper/2510.09738/) (2025)
- [BenLLMEval: A Comprehensive Evaluation into the Potentials and Pitfalls of Large Language Models on Bengali NLP](https://mlsanity.com/paper/2309.13173/) (2023)
- [An Empirical Study of LLM-as-a-Judge: How Design Choices Impact Evaluation Reliability](https://mlsanity.com/paper/2506.13639/) (2025)
- [L3Cube-IndicSBERT: A simple approach for learning cross-lingual sentence representations using multilingual BERT](https://mlsanity.com/paper/2304.11434/) (2023)
- [Social Bias in Large Language Models For Bangla: An Empirical Study on Gender and Religious Bias](https://mlsanity.com/paper/2407.03536/) (2024)
- [Vashantor: A Large-scale Multilingual Benchmark Dataset for Automated Translation of Bangla Regional Dialects to Bangla Language](https://mlsanity.com/paper/2311.11142/) (2023)
- [A Comparative Analysis of Retrieval-Augmented Generation Techniques for Bengali Standard-to-Dialect Machine Translation Using LLMs](https://mlsanity.com/paper/2512.14179/) (2025)
- [RegSpeech12: A Regional Corpus of Bengali Spontaneous Speech Across Dialects](https://mlsanity.com/paper/2510.24096/) (2025)

## Abstract (arXiv)

Large language models (LLMs) frequently exhibit performance biases against regional dialects of low-resource languages. However, frameworks to quantify these disparities remain scarce. We propose a two-phase framework to evaluate dialectal bias, operationalized as comprehension degradation relative to standard Bengali, in LLM question-answering across nine Bengali dialects. First, we translate and gold-label standard Bengali questions into dialectal variants adopting a retrieval-augmented generation (RAG) pipeline to prepare 4,000 question sets. Since traditional translation quality evaluation metrics fail on unstandardized dialects, we evaluate fidelity using an LLM-as-a-judge, which human correlation confirms outperforms legacy metrics. Second, we benchmark 19 LLMs across these gold-labeled sets, running 68,395 RLAIF evaluations validated through multi-judge agreement and human fallback. Our findings reveal severe performance drops linked to linguistic divergence. For instance, responses to the highly divergent Chittagong dialect score 5.44/10, compared to 7.68/10 for Tangail. Furthermore, increased model scale does not consistently mitigate this bias. We contribute a validated translation quality evaluation method, a rigorous benchmark dataset, and a Critical Bias Sensitivity (CBS) metric for safety-critical applications.