A Comparative Study on Textual Saliency of Styles from Eye Tracking, Annotations, and Language Models

TL;DR

This paper introduces eyeStyliency, a new eye-tracking dataset for stylistic text processing, compares it with human annotations and model importance scores, and explores its potential to evaluate NLP models' cognitive plausibility.

Contribution

The paper presents a novel eye-tracking dataset for stylistic language processing and analyzes its relationship with human annotations and model interpretability metrics.

Findings

Eye-tracking data provides unique insights into stylistic text processing.

There is an intersection between eye-tracking data, human annotations, and model importance scores.

Eye-tracking data can be used to evaluate the cognitive plausibility of NLP models.

Abstract

There is growing interest in incorporating eye-tracking data and other implicit measures of human language processing into natural language processing (NLP) pipelines. The data from human language processing contain unique insight into human linguistic understanding that could be exploited by language models. However, many unanswered questions remain about the nature of this data and how it can best be utilized in downstream NLP tasks. In this paper, we present eyeStyliency, an eye-tracking dataset for human processing of stylistic text (e.g., politeness). We develop a variety of methods to derive style saliency scores over text using the collected eye dataset. We further investigate how this saliency data compares to both human annotation methods and model-based interpretability metrics. We find that while eye-tracking data is unique, it also intersects with both human annotations and model-based importance scores, providing a possible bridge between human- and machine-based perspectives. We propose utilizing this type of data to evaluate the cognitive plausibility of models that interpret style. Our eye-tracking data and processing code are publicly available.