Habitual Speaking Berbasis Deep Learning untuk Meningkatkan Keterampilan Berbicara Bahasa Inggris di SMP Negeri 24 Semarang

Abstract

English speaking skills are essential for developing students' communication competence; however, classroom instruction is often constrained by limited speaking practice, low self-confidence, and learning activities that have not fully encouraged authentic communication. This community service program aimed to optimize English speaking instruction through the implementation of Habitual Speaking based on Deep Learning for teachers and students at SMP Negeri 24 Semarang. The program was conducted on 10–12 June 2026 involving 39 eighth-grade students and three English teachers. Activities included needs identification, preparation of learning materials, classroom implementation through Jumbled Word Game, Tic-Tac-Toe Speaking, and Market Simulation, followed by program evaluation using Likert-scale questionnaires. Evaluation results showed very positive responses from students (mean = 4.58) and teachers (mean = 4.68), indicating that the learning activities were perceived as communicative, contextual, and engaging. The program also provided teachers with practical experience in integrating the principles of Mindful Learning, Meaningful Learning, and Joyful Learning into English speaking instruction. The implementation of Habitual Speaking based on Deep Learning can serve as an alternative approach to support more meaningful and student-centered English speaking instruction in junior high schools.