Studies with impossible languages falsify LMs as models of human language

Behavioral and Brain Sciences 2026 Open access

Read full text
Authors: Jeffrey S. Bowers, Jeff Mitchell
DOI: 10.1017/s0140525x26104579
Cited by: 0
References: 12

Abstract

, 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.

Fields of Study & Subjects

Language Development and Disorders Neurobiology of Language and Bilingualism Language and cultural evolution

Publication details

Journal / source
Behavioral and Brain Sciences
Publisher
Cambridge University Press (CUP)
Volume
49
Pages
e202–e202
Publication date
1 January 2026
Language
EN