Immediate and Delayed Effects of ChatGPT-Enhanced Vocabulary Instruction on Saudi EFL Learners

Research on AI-mediated vocabulary learning has expanded rapidly, yet the delayed effects of GPT-4-based ChatGPT model via a web interface on lexical retention remain underexplored, particularly in English as a Foreign Language (EFL) contexts such as Saudi Arabia. This study investigates whether ChatGPT is associated with differences in sustained vocabulary learning compared with traditional instruction, and the short- and delayed effects of ChatGPT on vocabulary learning and retention among Saudi EFL learners using an explanatory sequential mixed-methods design. Forty undergraduate learners were assigned to experimental and control groups and studied the target vocabulary sets during a four-week intervention, followed by a four-week retention interval. Vocabulary development was measured at three intervals with a Vocabulary Knowledge Scale, which was administered at the pre-test, post-test and delayed post-test stages. Results showed higher VKS scores at the immediate and delayed post-tests in the ChatGPT group compared with traditional instruction. Qualitative findings indicated that learners perceived ChatGPT as supporting motivation, confidence, and personalised engagement, while also identifying challenges related to interpreting complex AI feedback. These findings highlight the pedagogical promise of integrating AI tools into vocabulary instruction while also indicating the need for guided teacher mediation. Implications for EFL pedagogy and recommendations for future research on AI-supported language learning, such as comprehending the role of AI in developing language learning by investigating other language skills, engaging wider samples, and applying longer interventions and retention periods are also discussed.