Do LLMs learn a true syntactic universal?

John Hale, Miloš Stanojević


Abstract
Do large multilingual language models learn language universals? We consider a candidate universal much-discussed in the linguistics literature, the Final-over-Final Condition (Sheehan et al., 2017b). This Condition is syntactic in the sense that it can only be stated by reference to abstract sentence properties such as nested phrases and head direction. A study of typologically diverse “mixed head direction” languages confirms that the Condition holds in corpora. But in a targeted syntactic evaluation, Gemini Pro only seems to respect the Condition in German, Russian, Hungarian and Serbian. These relatively high-resource languages contrast with Basque, where Gemini Pro does not seem to have learned the Condition at all. This result suggests that modern language models may need additional sources of bias in order to become truly human-like, within a developmentally-realistic budget of training data.
Anthology ID:
2024.emnlp-main.950
Volume:
Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
Month:
November
Year:
2024
Address:
Miami, Florida, USA
Editors:
Yaser Al-Onaizan, Mohit Bansal, Yun-Nung Chen
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
17106–17119
Language:
URL:
https://aclanthology.org/2024.emnlp-main.950
DOI:
Bibkey:
Cite (ACL):
John Hale and Miloš Stanojević. 2024. Do LLMs learn a true syntactic universal?. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 17106–17119, Miami, Florida, USA. Association for Computational Linguistics.
Cite (Informal):
Do LLMs learn a true syntactic universal? (Hale & Stanojević, EMNLP 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.emnlp-main.950.pdf