Lexical Threshholds in Medical English II: AI-Assisted Text Simplification and Its Reletionship on Vocabulary and Reading Comprehension

Abstract

This study explores how AI-assisted text simplification may influence the relationship between medical vocabulary knowledge and reading comprehension in English for Medical Purposes (EMP). Building on a previous phase, the study adopts a two-group design within a single cohort of 84 second-year medical students, comparing performance on original and simplified versions of the same biochemistry texts. The simplified texts were generated using ChatGPT. Participants completed a 45-item test consisting of 25 vocabulary recognition items and 20 reading comprehension questions. The analysis focuses on overall performance, the relationship between lexical knowledge and comprehension, and the extent to which learners appear to transfer vocabulary knowledge to reading tasks.The results suggest generally high levels of performance in the AI-assisted condition, with mean scores of 91% for vocabulary and 84% for comprehension. A positive relationship was observed between vocabulary knowledge and comprehension (r = .74). When compared with the original-text condition (r = .56), this pattern may indicate a somewhat stronger alignment between lexical knowledge and comprehension under simplified conditions. Transfer-efficiency results (M = 0.92) suggest that students were generally able to apply their lexical knowledge in reading, although with notable individual variation. At the same time, differences in performance between texts remained, suggesting that text-related factors may continue to play a role even under simplified conditions.

Ähnliche Arbeiten