1,729 vs. 1729: The Effect of Scripts and Formats on LLM Numeracy

Researchers at the Association for Computational Linguistics found that large language models' numerical accuracy drops significantly when numbers are presented in underrepresented scripts or formats, even when the underlying math is identical. The study, presented at ACL 2026, shows that targeted prompting strategies like few-shot prompting and explicit numeral mapping can narrow this performance gap, highlighting an overlooked challenge in multilingual numerical reasoning.

1,729 vs. 1729: The Effect of Scripts and Formats on LLM Numeracy https://aclanthology.org/2026.findings-acl.518.pdf Varshini Reddy /people/varshini-reddy/unverified/ , Craig W Schmidt /people/craig-w-schmidt/ , Seth Ebner /people/seth-ebner/ , Adam Wiemerslage /people/adam-wiemerslage/unverified/ , Yuval Pinter /people/yuval-pinter/ , Chris Tanner /people/chris-tanner/ Abstract Large language models LLMs have achieved impressive proficiency in basic arithmetic, rivaling human-level performance on standard numerical tasks. However, little attention has been given to how these models perform when numerical expressions deviate from the prevailing conventions present in their training corpora. In this work, we investigate numerical reasoning across a wide range of numeral scripts and formats. We show that LLM accuracy drops substantially when numerical inputs are rendered in underrepresented scripts or formats, despite the underlying mathematical reasoning being identical. We further demonstrate that targeted prompting strategies, such as few-shot prompting and explicit numeral mapping, can greatly narrow this gap. Our findings highlight an overlooked challenge in multilingual numerical reasoning and provide actionable insights for working with LLMs to reliably interpret, manipulate, and generate numbers across diverse numeral scripts and formatting styles.- Anthology ID: - 2026.findings-acl.518 - Volume: Findings of the Association for Computational Linguistics: ACL 2026 /volumes/2026.findings-acl/ - Month: - July - Year: - 2026 - Address: - San Diego, California, United States - Editors: Maria Liakata /people/maria-liakata/ , Viviane P. Moreira /people/viviane-p-moreira/unverified/ , Jiajun Zhang /people/jiajun-zhang/unverified/ , David Jurgens /people/david-jurgens/ - Venue: Findings /venues/findings/ - SIG: - Publisher: - Association for Computational Linguistics - Note: - Pages: - 10679–10696 - Language: - URL: https://aclanthology.org/2026.findings-acl.518/ https://aclanthology.org/2026.findings-acl.518/ - DOI: - Cite ACL : - Varshini Reddy, Craig W Schmidt, Seth Ebner, Adam Wiemerslage, Yuval Pinter, and Chris Tanner. 2026. 1,729 vs. 1729: The Effect of Scripts and Formats on LLM Numeracy https://aclanthology.org/2026.findings-acl.518/ . In Findings of the Association for Computational Linguistics: ACL 2026 , pages 10679–10696, San Diego, California, United States. Association for Computational Linguistics. - Cite Informal : 1,729 vs. 1729: The Effect of Scripts and Formats on LLM Numeracy https://aclanthology.org/2026.findings-acl.518/ Reddy et al., Findings 2026 - PDF: https://aclanthology.org/2026.findings-acl.518.pdf https://aclanthology.org/2026.findings-acl.518.pdf