New CALQUE benchmark tests how multilingual LLMs handle constructional borrowing across languages
A new arXiv paper introduces CALQUE, a benchmark evaluating how well multilingual large language models apply morphosyntactic strategies from one language when expressing meanings in another, a phenomenon the authors call constructional bleeding.
Read the full story at arXiv cs.CL ↗
Timeline · 1 report
- 2026-11-10 05:00 · arXiv cs.CL
I play of the cello: Evaluating Constructional Bleeding in Multilingual LLMs with CALQUE