Enhancing Vocabulary and Spelling Accuracy through AI-Generated Stories by Integrating Dictation Activities for A2-Level EFL Learners

Abstract

This study evaluated the pedagogical and affective impact of incorporating artificial intelligence (AI)-generated contextualized stories into dictation activities for secondary English as a Foreign Language (EFL) learners. Traditional dictation exercises often use static, decontextualized texts that lead to learner monotony, persistent orthographic difficulties, and minimal engagement. To address this challenge, an 8-week mixed-methods intervention was conducted with 54 CEFR A2-level students at a bilingual institution in Ecuador. Participants completed weekly dictations using level-adapted narratives generated Gemini with specific syntactic constraints and embedded target vocabulary. Quantitative pre- and post-tests evaluated orthographic accuracy and target word mastery, while qualitative focus groups and a post-intervention survey explored student motivation and engagement. Quantitative results demonstrated substantial linguistic gains: orthographic accuracy increased by an absolute mean of +17.7%, and target word composite mastery scores rose from 10.80 to 16.75 out of 20 points (+5.95 gain) across receptive and productive dimensions. Survey findings revealed high student engagement across behavioral (M=4.35), affective (M=4.28), and cognitive (M=4.12) domains. Thematic analysis identified narrative curiosity and immediate self-correction feedback as key driving factors for motivation, while acoustic delivery speed and cognitive writing fatigue emerged as primary hindering factors. The findings demonstrate that AI-generated story dictations effectively improve traditional dictation by enhancing orthographic precision and target vocabulary retention while maintaining high learner engagement when supported by appropriate instructional scaffolding.