Enhancing EFL Speaking Proficiency through TalkPal AI and Automatic Speech Recognition: Meningkatkan Kemampuan Berbicara Bahasa Inggris sebagai Bahasa Asing (EFL) melalui TalkPal AI dan Pengenalan Suara Otomatis

Hasna Nurain Mukhsin (Universitas Muhammadiyah Gorontalo)
Rita Roswita Duyo Duyo (STKIP YPUP Makassar)
Randi Safii (Universitas Muhammadiyah Gorontalo)

ELOQUENCE: Journal of Foreign Language
Vol. 5 No. 2 (2026): AUGUST

Background: The integration of Artificial Intelligence (AI) into language education has created new opportunities to improve English as a Foreign Language (EFL) learners' speaking proficiency through interactive and personalized learning environments. One emerging platform is TalkPal AI, which incorporates Automatic Speech Recognition (ASR) technology to provide immediate feedback on learners' oral performance. Purpose: This study aimed to investigate the effectiveness of TalkPal AI-assisted learning in enhancing EFL university students’ speaking proficiency through Automatic Speech Recognition (ASR)-based learning. Method: A quantitative approach employing a quasi-experimental nonequivalent control group design was used. The participants were 21 undergraduate students enrolled in the English Literature Study Program at Universitas Muhammadiyah Gorontalo, comprising an experimental group (n = 11) and a control group (n = 10). Students' speaking proficiency was assessed through pre-test and post-test speaking tasks using an analytic rubric covering pronunciation, fluency, grammar, vocabulary, and comprehension. The instrument demonstrated excellent inter-rater reliability (ICC = 0.987) and high rating consistency (Cronbach's α = 0.994). Data were analyzed using descriptive statistics, Shapiro–Wilk normality test, Levene's test, paired samples t-test, independent samples t-test, and Cohen's d. Results and Discussion: Both groups showed significant improvement after instruction. However, the experimental group demonstrated a substantially greater improvement (mean difference = 42.09) than the control group (mean difference = 14.25). The independent samples t-test revealed a significant difference in post-test scores between the groups (t = −14.470, p < .001), with a very large effect size (Cohen's d = 2.35). Conclusions and Implications: TalkPal AI-assisted learning supported by ASR technology was associated with significantly higher speaking proficiency than conventional instruction, indicating its potential to enhance technology-enhanced language learning in EFL speaking classrooms.