Front Med (Lausanne). 2026 Sep 16;13:1895483. doi: 10.3389/fmed.2026.1895483. eCollection 2026.
ABSTRACT
INTRODUCTION: Evidence-based practice requires students to read, evaluate, and write about research, all of which depend on academic vocabulary. For English as a Foreign Language (EFL) health sciences students studying through English-medium instruction (EMI), this is a persistent challenge that affects their readiness for clinical practice, and AI writing tools have complicated it further since late 2022, as they can produce the academic English that writing courses spend years developing. This study tracks academic vocabulary development across four trimesters, a period coinciding with the phased rollout of ChatGPT in Saudi Arabia.
METHODS: The corpus-based analysis drew on 855,680 running words from the King Saud bin Abdulaziz University for Health Sciences Learner Corpus, produced by 157 female preparatory program students with Arabic as their first language. Academic Word List (AWL) coverage was computed for each trimester as the percentage of running words drawn from the 570 word families of the AWL, and compared against published benchmarks for medical academic texts. Each trimester was mapped to one of four phases of ChatGPT accessibility.
RESULTS: Overall AWL coverage was 7.66%, rising significantly from 3.69% in Trimester 1 to 11.39% in Trimester 4. The steepest gain coincided with the most demanding writing task and with unrestricted ChatGPT access, and Trimester 4 coverage exceeded the 10.07% benchmark for expert medical research articles.
DISCUSSION: Progressive, genre-based English for Academic Purposes (EAP) instruction was associated with substantial vocabulary growth. Because task complexity, cumulative language development, and AI availability all increased over the same period, their effects cannot be disentangled, and the results are interpreted as associations rather than as evidence of an AI effect. The findings carry implications for vocabulary teaching and for the validity of unsupervised writing assessment.
PMID:42819225 | PMC:PMC13623714 | DOI:10.3389/fmed.2026.1895483