This experimental study examined the design and effectiveness of embodied interactions for learning. The researchers designed a digital learning environment integrating body joint mapping sensors to teach novice learners Chinese characters, and examined whether the embodied interaction would lead to greater knowledge acquisition in language learning compared to the conventional mouse-based interaction. Fifty-three adult learners were randomly assigned to experimental and control groups. The study adopted a pretest, an immediate posttest, and a delayed posttest on knowledge acquisition. Although higher scores were found for the embodied interaction group in both posttests, only the delayed posttest showed a statistically significant group difference. The findings suggested that active embodied actions lead to better knowledge retention compared with the passive visual embodiment. The body-moving process works as an alternative and complementary encoding strategy for character understanding and memorization by associating the semantic meaning of a character with the construction of a body posture.