This study developed a Kinect-based somatosensory English learning system. The main design concept was to integrate Kinect as an interaction technique with theories of situated learning and the attention, relevance, confidence, and satisfaction (ARCS) model, to design relevant learning activities and materials, thereby enhancing students’ learning outcomes. The proposed system allows for planning and designing learning activities and content according to situated learning components and the ARCS model. The somatosensory interaction system Kinect was used to provide users with a virtual learning environment to achieve actual spatial and physical experiences, assisting learners’ engagement in stories and scenarios as well as enhancing their motivation to learn. English vocabulary related to supermarkets was set as the learning objective and 70 students ranging from third to sixth grade at a learning center in Tainan, Taiwan were selected as participants. During the experiment, participants were divided into two groups: the experimental group, which employed the proposed learning system, and the control group, in which students learned using printed materials coupled with mobile devices. Pre- and posttest scores of the two groups were used to assess learning outcomes and analyze the ARCS model-based questionnaire. The results revealed that the proposed system effectively improved learners’ motivation to learn and learning outcomes.