: In this paper, we propose an improved context-aware system for supporting to learning Japanese mimicry and onomatopoeia (MIO) using sensor data. In our two previous studies, we proposed context-aware language learning assistant systems named JAMIOLAS (JApanese MImicry and Onomatopoeia Learning Assistant System). We have used wearable sensors and sensor network respectively to support learning Japanese MIO. On the disadvantage of previous systems, we propose a new learning model that can support learning MIO with the sensor network to carry out context-aware learning mainly in ways of creating context initiatively and detecting context automatically.