Studies with impossible languages falsify LMs as models of human language
Computation and Language
2025-11-17 v1
Abstract
According to Futrell and Mahowald [arXiv:2501.17047], both infants and language models (LMs) find attested languages easier to learn than impossible languages that have unnatural structures. We review the literature and show that LMs often learn attested and many impossible languages equally well. Difficult to learn impossible languages are simply more complex (or random). LMs are missing human inductive biases that support language acquisition.
Cite
@article{arxiv.2511.11389,
title = {Studies with impossible languages falsify LMs as models of human language},
author = {Jeffrey S. Bowers and Jeff Mitchell},
journal= {arXiv preprint arXiv:2511.11389},
year = {2025}
}
Comments
Commentary on Futrell, R., & Mahowald, K. arXiv:2501.17047 (in press). How linguistics learned to stop worrying and love the language models. Behavioural and Brain Sciences