Many language learners have difficulty practicing listening skills using authentic materials, and thus use captions to map text with speech, and they benefit from reading along while listening to comprehend content. However, many learners over-rely on reading the text and many have difficulty in dividing their attention to the multimodal input. We have proposed a captioning tool, Partial and Synchronized Captions (PSC), which detects the useful words to be shown in the caption for addressing learners' listening difficulties. To handle individual learner demands, PSC should adapt its word selection criteria. This study proposes an Adaptive PSC (APSC), which improves its word selection and retrains itself on-the-fly by applying learner feedback on the generated caption to provide individualized and effective assistance that satisfies the learners' requirements. Preliminary results revealed that the system was relatively successful to adapt itself to the demand of the L2 learner, which raised learner satisfaction on the resultant captions. [For the complete proceedings, see ED600837.]