To What Extent Do Large Language Models Understand Bangla Idioms?

arXiv cs.CL Papers

Summary

This paper introduces a benchmark dataset for Bangla idioms and evaluates recent large language models on idiom-related tasks, revealing substantial variability in performance across models.

arXiv:2609.03410v1 Announce Type: new Abstract: Idiomatic expressions are an integral part of natural language, reflecting cultural nuances and posing unique challenges for computational models, particularly in low-resource languages. In this paper, we present the first large-scale benchmark dataset of Bangla idioms, complemented by a synthetic multiple-choice question (MCQ) dataset for idiom meaning identification. We conduct a comprehensive evaluation of recent large language models (LLMs) across three idiom-related tasks: paraphrasing, idiom span detection, and meaning identification, leveraging zero-shot and few-shot prompting strategies. Our results reveal substantial variability in model performance, with no single LLM consistently outperforming others across all tasks. Notably, Phi-4-mini-instruct excels in paraphrasing, Kimi-K2-32b-instruct in span detection, and Gemini-2.5-flash in meaning identification. We believe that our datasets and analyses will provide valuable resources to guide future research in improving LLM comprehension of idiomatic expressions, particularly in Bangla and other low-resource languages.
Original Article
View Cached Full Text

Cached at: 09/04/26, 05:59 AM

# To What Extent Do Large Language Models Understand Bangla Idioms?
Source: [https://arxiv.org/abs/2609.03410](https://arxiv.org/abs/2609.03410)
[View PDF](https://arxiv.org/pdf/2609.03410)

> Abstract:Idiomatic expressions are an integral part of natural language, reflecting cultural nuances and posing unique challenges for computational models, particularly in low\-resource languages\. In this paper, we present the first large\-scale benchmark dataset of Bangla idioms, complemented by a synthetic multiple\-choice question \(MCQ\) dataset for idiom meaning identification\. We conduct a comprehensive evaluation of recent large language models \(LLMs\) across three idiom\-related tasks: paraphrasing, idiom span detection, and meaning identification, leveraging zero\-shot and few\-shot prompting strategies\. Our results reveal substantial variability in model performance, with no single LLM consistently outperforming others across all tasks\. Notably, Phi\-4\-mini\-instruct excels in paraphrasing, Kimi\-K2\-32b\-instruct in span detection, and Gemini\-2\.5\-flash in meaning identification\. We believe that our datasets and analyses will provide valuable resources to guide future research in improving LLM comprehension of idiomatic expressions, particularly in Bangla and other low\-resource languages\.

## Submission history

From: Mousumi Akter \[[view email](https://arxiv.org/show-email/819e642f/2609.03410)\] **\[v1\]**Thu, 3 Sep 2026 06:15:15 UTC \(552 KB\)

Similar Articles

When English Rewrites Local Knowledge: Global Narrative Dominance in Large Language Models

arXiv cs.CL

This paper introduces CulturalNB, a dataset of Bengali cultural question-answer pairs, and evaluates nine LLMs for cross-lingual cultural bias. Findings show that English prompting increases global narrative substitution and reduces local perspectives, revealing that cultural failures in LLMs are grounding and prioritization issues, not just missing knowledge.

VIVID: A Culturally Grounded Benchmark Exposing the Figurative Language Gap in Vietnamese NLP

arXiv cs.CL

This paper introduces VIVID, the first systematic benchmark for evaluating culturally grounded figurative language understanding in Vietnamese, comprising 1,636 idioms and proverbs. Evaluation of eight state-of-the-art models reveals significant gaps, with Vietnamese-specialized models drastically underperforming multilingual systems and even top models achieving less than 50% correctness on average.