To Words and Beyond: Probing Large Language Models for Sentence-Level Psycholinguistic Norms of Memorability and Reading Times

TL;DR

This study explores how large language models can estimate sentence-level psycholinguistic norms like memorability and reading times, showing that fine-tuning improves their correlation with human data, but zero-shot performance remains inconsistent.

Contribution

The paper extends LLM-based psycholinguistic norm estimation to sentence-level features, demonstrating the benefits of fine-tuning for better alignment with human judgments.

Findings

Fine-tuning improves LLM estimates of sentence memorability and reading times.

Zero-shot and few-shot performances are inconsistent and require careful application.

Fine-tuned models outperform baseline predictors in predicting human-derived norms.

Abstract

Large Language Models (LLMs) have recently been shown to produce estimates of psycholinguistic norms, such as valence, arousal, or concreteness, for words and multiword expressions, that correlate with human judgments. These estimates are obtained by prompting an LLM, in zero-shot fashion, with a question similar to those used in human studies. Meanwhile, for other norms such as lexical decision time or age of acquisition, LLMs require supervised fine-tuning to obtain results that align with ground-truth values. In this paper, we extend this approach to the previously unstudied features of sentence memorability and reading times, which involve the relationship between multiple words in a sentence-level context. Our results show that via fine-tuning, models can provide estimates that correlate with human-derived norms and exceed the predictive power of interpretable baseline predictors, demonstrating that LLMs contain useful information about sentence-level features. At the same time, our results show very mixed zero-shot and few-shot performance, providing further evidence that care is needed when using LLM-prompting as a proxy for human cognitive measures.