Disentangling Linguistic Relatedness from Task Alignment in Cross-Lingual Transfer
Document Type
Conference Proceeding
Role
Author
Published In
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics
Publisher
Association for Computational Linguistics
First Page
683
Last Page
700
Publication Date
7-2026
Abstract
We study cross-lingual transfer by fine-tuning seven large language models (4B–671B parameters) on Arabic and evaluating zero-shot reading comprehension on Semitic languages and non-Semitic controls. Across dense and Mixture-of-Experts architectures, we find no evidence of Semitic-specific transfer: models with weak baselines improve dramatically across all languages, while strong-baseline models show only marginal gains regardless of language family. A chain-of-thought ablation reinforces this finding: the same models that benefit most from fine-tuning benefit equally from inference-time reasoning, suggesting both mechanisms address task-format alignment rather than cross-lingual knowledge transfer.
Suggested Citation
Ahmed, Ahmed H. '26, Zhang, R., & Grissom II, Alvin C. (Computer Science) (2026). Disentangling Linguistic Relatedness from Task Alignment in Cross-Lingual Transfer. In S. T.Y.S.S., J. D. Rodriguez, & O. de Gibert (Eds.), Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (pp. 683–700). Association for Computational Linguistics. https://doi.org/10.18653/v1/2026.acl-srw.62
