Effect of case markers during agreement production: A model comparison using Armenian forced choice data

Pranab Bagartti, Samar Husain


Abstract
Agreement attraction errors, where the verb erroneously agrees with a non-subject noun, have been a useful tool for investigating processes that subserve sentence production. Research has shown that case markers play an important role in modulating such errors. These effects have been argued to arise due to an underlying cue-based retrieval system. However, subsequent research in Armenian has challenged this conclusion (Avetisyan et al., 2020), arguing against a cue-based retrieval account. The current paper revisits the Armenian production data through computational modeling. Specifically, we implemented three distinct models and compared their predictions; we compare (a) a cue-based retrieval model, (b) a feature migration model, and (c) a case as markers for agreement prediction model. Our model comparison results show that a case as markers for agreement prediction model followed by an inference component explains the effect of case better than the cue-based retrieval model as well as the feature migration model.
Anthology ID:
2026.scil-main.33
Volume:
Proceedings of the Society for Computation in Linguistics 2026
Month:
July
Year:
2026
Address:
San Diego, CA
Editors:
Rob Voigt, Alex Warstadt, Naomi Feldman, Tal Linzen
Venues:
SCiL | WS
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
353–364
Language:
URL:
https://preview.aclanthology.org/ingest-acl-workshops/2026.scil-main.33/
DOI:
Bibkey:
Cite (ACL):
Pranab Bagartti and Samar Husain. 2026. Effect of case markers during agreement production: A model comparison using Armenian forced choice data. In Proceedings of the Society for Computation in Linguistics 2026, pages 353–364, San Diego, CA. Association for Computational Linguistics.
Cite (Informal):
Effect of case markers during agreement production: A model comparison using Armenian forced choice data (Bagartti & Husain, SCiL 2026)
Copy Citation:
PDF:
https://preview.aclanthology.org/ingest-acl-workshops/2026.scil-main.33.pdf