LLMs do not exhibit the neglect-zero effect in structural priming tasks unlike humans.
Priming reveals similarities and differences between three purported cases of implicature: Some, number and free choice disjunctions
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Analysis of the Neglect-Zero Effect in Large Language Models
LLMs do not exhibit the neglect-zero effect in structural priming tasks unlike humans.