Skip to main content
Paper

Investigating Syntactic Biases in Multilingual Transformers with RC Attachment Ambiguities in Italian and English

Authors
  • Michael Kamerath (University of Utah)
  • Aniello De Santo (University of Utah)

Abstract

This paper investigates whether monolingual and multilingual LLMs show human-like preferences when presented with examples of relative clause attachment  ambiguities in Italian and English. We also test whether these preferences can be modulated by lexical factors (the type of verb/noun in the matrix clause) which have been shown to be tied to subtle constraints on syntactic and semantic relations. Our results overall showcase how LLM behavior varies inconsistently across models and languages, and highlight the importance of leveraging subtle syntactic contrasts in exploring these models’ ability to correctly align with human-like preferences

Keywords: Syntactic Ambiguity, LLMs, RC Attachment, English, Italian, Relative Clause

How to Cite:

Kamerath, M. & De Santo, A., (2026) “Investigating Syntactic Biases in Multilingual Transformers with RC Attachment Ambiguities in Italian and English”, Society for Computation in Linguistics 9(1). doi: https://doi.org/10.7275/scil.4041

Downloads:
Download PDF

45 Views

18 Downloads

Published on
2026-06-26

Peer Reviewed