Two ways into the hall of mirrors: Language exposure and lossy memory drive cross-linguistic grammaticality illusions in language models

Kate McCurdy, Katharina Christian, Amelie Seyfried, Mikhail Sonkin


Abstract
Readers of English — but not Dutch or German — consistently show a grammaticality illusion: they find ungrammatical double-center-embedded sentences easier to process than corresponding grammatical sentences. If pre-trained language model (LM) surprisal mimics these cross-linguistic patterns, this implies that language statistics explain the effect; if, however, the illusion requires memory constraints such as lossy context surprisal (LCS), this suggests a critical role for memory. We evaluate LMs in Dutch, German, and English. We find that both factors influence LMs’ susceptibility to grammaticality illusions, and neither fully account for human-like processing patterns.
Anthology ID:
2025.babylm-main.17
Volume:
Proceedings of the First BabyLM Workshop
Month:
November
Year:
2025
Address:
Suzhou, China
Editors:
Lucas Charpentier, Leshem Choshen, Ryan Cotterell, Mustafa Omer Gul, Michael Y. Hu, Jing Liu, Jaap Jumelet, Tal Linzen, Aaron Mueller, Candace Ross, Raj Sanjay Shah, Alex Warstadt, Ethan Gotlieb Wilcox, Adina Williams
Venue:
BabyLM
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
226–236
Language:
URL:
https://aclanthology.org/2025.babylm-main.17/
DOI:
Bibkey:
Cite (ACL):
Kate McCurdy, Katharina Christian, Amelie Seyfried, and Mikhail Sonkin. 2025. Two ways into the hall of mirrors: Language exposure and lossy memory drive cross-linguistic grammaticality illusions in language models. In Proceedings of the First BabyLM Workshop, pages 226–236, Suzhou, China. Association for Computational Linguistics.
Cite (Informal):
Two ways into the hall of mirrors: Language exposure and lossy memory drive cross-linguistic grammaticality illusions in language models (McCurdy et al., BabyLM 2025)
Copy Citation:
PDF:
https://aclanthology.org/2025.babylm-main.17.pdf