Automated feedback for speaking and writing skills: Deep learning in English language assessment
EduLite: Journal of English Education, Literature and Culture · 05.03.2026 · 11 (1) · 67
DOI: 10.30659/e.11.1.67-85 ↗Kurz gefasst
Fachartikel in der Zeitschrift EduLite: Journal of English Education, Literature and Culture, online erschienen am 05.03.2026. Verfasst von Andi Asrifan und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
The incorporation of Artificial Intelligence (AI) into language evaluation has revolutionized how learners receive feedback on their speaking and writing abilities. Nevertheless, empirical information about the precision and educational efficacy of AI-generated feedback—especially in advanced language competencies—continues to be scarce. This study seeks to evaluate the efficacy of deep learning–driven automated feedback systems in enhancing English learners' speaking and writing skills. The study utilized a mixed-methods research approach and included 100 undergraduate students participating in an English for Academic Purposes course, focusing on English as a Foreign Language (EFL). Quantitative data were gathered via pre-test and post-test writing and speaking activities evaluated using AI tools (Grammarly, ETS e-rater, and Google Automatic Speech Recognition), whilst qualitative data were derived from surveys and interviews to capture learners' impressions. The findings demonstrate statistically significant enhancements in grammatical accuracy, lexical diversity, coherence, fluency, pronunciation, and intelligibility following exposure to AI-generated feedback. However, inconsistencies were identified between AI and human assessments regarding speech coherence and contextual relevance. The results indicate that AI-generated feedback serves as an excellent additional evaluation instrument, especially for form-focused linguistic elements, however it is constrained in its ability to measure higher-order communication competencies. This study underscores the significance of amalgamating AI-driven feedback with human discernment to establish a more holistic and pedagogically robust language assessment framework.
Zitieren
APA
Asrifan, A., Cardoso, L. M. O. D. B., & Vargheese, K. (2026). Automated feedback for speaking and writing skills: Deep learning in English language assessment. EduLite: Journal of English Education, Literature and Culture, 11(1), 67. https://doi.org/10.30659/e.11.1.67-85
Vancouver
Asrifan A, Cardoso LMODB, Vargheese K. Automated feedback for speaking and writing skills: Deep learning in English language assessment. EduLite: Journal of English Education, Literature and Culture. 2026;11(1):67. doi:10.30659/e.11.1.67-85
BibTeX
@article{Asrifan2026,
author = {Asrifan, Andi and Cardoso, Luís Miguel Oliveira de Barros and Vargheese, K.J.},
title = {Automated feedback for speaking and writing skills: Deep learning in English language assessment},
journal = {EduLite: Journal of English Education, Literature and Culture},
year = {2026},
volume = {11},
number = {1},
pages = {67},
doi = {10.30659/e.11.1.67-85}
}
EUVIMED