TY  - RPRT
TI  - Why are language models less surprised than humans? Testing the Parse Multiplicity Mismatch Hypothesis
AU  - William Timkey
AU  - Brian Dillon
AU  - Tal Linzen
PY  - 2026
UR  - https://arxiv.org/abs/2605.15440
ID  - 2605.15440
ER  -