Contextualising (Im)plausible Events Triggers Figurative Language

TL;DR

This paper investigates how humans and language models perceive plausibility and literalness in event descriptions, revealing humans' nuanced understanding versus LLMs' shallow contextualization.

Contribution

It systematically compares human and LLM judgments on plausible and implausible events, highlighting differences in contextual understanding and interpretative depth.

Findings

Humans distinguish nuanced (non-)literal and plausibility cues effectively.

LLMs show shallow contextualization with a bias towards non-literal interpretations.

Significant differences exist between human and LLM assessments of event plausibility.

Abstract

This work explores the connection between (non-)literalness and plausibility at the example of subject-verb-object events in English. We design a systematic setup of plausible and implausible event triples in combination with abstract and concrete constituent categories. Our analysis of human and LLM-generated judgments and example contexts reveals substantial differences between assessments of plausibility. While humans excel at nuanced detection and contextualization of (non-)literal vs. implausible events, LLM results reveal only shallow contextualization patterns with a bias to trade implausibility for non-literal, plausible interpretations.