UrBLiMP: A Benchmark for Evaluating the Linguistic Competence of Large Language Models in Urdu

TL;DR

This paper introduces UrBLiMP, a benchmark dataset for evaluating the syntactic capabilities of multilingual large language models in Urdu, revealing performance disparities and highlighting current limitations in low-resource language modeling.

Contribution

The paper presents UrBLiMP, a new dataset of minimal pairs for Urdu, and evaluates twenty multilingual LLMs, providing insights into their syntactic understanding in a low-resource language.

Findings

LLaMA-3-70B achieves 94.73% accuracy on UrBLiMP.

Performance varies significantly across linguistic phenomena.

Current models show limitations in capturing fine-grained Urdu syntax.

Abstract

Multilingual Large Language Models (LLMs) have shown remarkable performance across various languages; however, they often include significantly less data for low-resource languages such as Urdu compared to high-resource languages like English. To assess the linguistic knowledge of LLMs in Urdu, we present the Urdu Benchmark of Linguistic Minimal Pairs (UrBLiMP) i.e. pairs of minimally different sentences that contrast in grammatical acceptability. UrBLiMP comprises 5,696 minimal pairs targeting ten core syntactic phenomena, carefully curated using the Urdu Treebank and diverse Urdu text corpora. A human evaluation of UrBLiMP annotations yielded a 96.10% inter-annotator agreement, confirming the reliability of the dataset. We evaluate twenty multilingual LLMs on UrBLiMP, revealing significant variation in performance across linguistic phenomena. While LLaMA-3-70B achieves the highest average accuracy (94.73%), its performance is statistically comparable to other top models such as Gemma-3-27B-PT. These findings highlight both the potential and the limitations of current multilingual LLMs in capturing fine-grained syntactic knowledge in low-resource languages.