Comparing Reading Behavior across Reader Expertise and Text Complexity: Insights from the French Eye-Tracking Corpus (FETA)

Abstract

This study examines how readers process general and medical texts with varying levels of complexity and how text simplification affects reading behavior. Using eye-tracking data, we compared two participant groups – common population and speech therapy students – as they read French medical, clinical, and general-domain texts in both original and simplified versions. We applied unsupervised clustering to identify patterns in reading behavior and investigate whether these patterns differ across participant groups, text types and complexity. The analysis identified between two and four clusters per group and condition, revealing distinct reading strategies ranging from effortful re-reading behavior to fluent, streamlined processing. The results reveal that medical and clinical texts elicit longer fixations and more regressions, indicating greater processing effort, while simplification produces shorter fixations and more fluid reading. Speech therapy students generally exhibit more efficient and stable gaze patterns, reflecting greater metalinguistic awareness and familiarity with the field. The dataset is a novel resource for modelling cognitive aspects of text complexity in French.

Resources

Details