USING DEEP LEARNING TO IMPROVE SPEAKING SKILLS: A BIBLIOMETRIC REVIEW OF ACADEMIC TRENDS AND RESEARCH DIRECTIONS (2010–2024)

This bibliometric review examines the incorporation of deep learning methodologies in enhancing speaking skills within language instruction. This study evaluates 437 peer-reviewed articles published between 2010 and 2024, utilising data from Scopus and Web of Science. The review used VOSviewer and Bibliometrix (R) to analyse publishing trends, identify leading authors and journals, and delineate significant research clusters and citation effects. The results indicate a notable increase in research activity from 2018 onwards, with substantial contributions from China, the United States, and Europe. Despite the increasing importance of studies utilising convolutional neural networks (CNNs), recurrent neural networks (RNNs), and transformers, deficiencies persist in pedagogical alignment, learner effect, and real-time feedback mechanisms. This review provides a framework for future research at the convergence of AI and language acquisition, emphasising customised and scalable speaking skill development. Keywords: Deep learning, oral proficiency, language acquisition, artificial intelligence, convolutional neural networks, recurrent neural networks, transformer architectures, educational technology, bibliometric analysis