The use of machine learning in developing learner-adaptive tools for second language acquisition

Abstract

[EN] Advancements in artificial intelligence and machine learning present opportunities to revolutionize language learning tools with learner-adaptive capabilities. These technologies facilitate the creation of trainable systems that can interact with learners, offering personalized learning experiences tailored to individual needs, interests, proficiency levels, backgrounds, and native languages. This study explores the role of machine learning in developing personalized frameworks for second language learning, introducing the Partial and Synchronized Caption (PSC) tool as an example. PSC utilizes automatic speech recognition and natural language processing to identify challenging words for language learners, which are presented in the caption while masking easy words. We used machine learning to personalize the caption for various learners. An experiment involving graduate students learning English as a second language demonstrated the adaptability of PSC's word selection to different learners. While creating entirely personalized captions may be challenging, PSC offers a promising approach to personalized and adaptable listening tools. The data collected from learner interactions also provides valuable insights into individual needs, shaping future language learning tools and pedagogical practices.