Dynamic visuospatial support (DVS) is an effective tool for helping people learn linguistic tones (Hannah et al., 2017, Morett et al., 2022, Zhen et al., 2019), and learn words with these tonal categories (Baills et al., 2019, Morett et al., 2015). Progress in the field of embodied cognition and learning has led to recent emphasis on the pitch gesture (PG), a hand gesture representing the pitch movement of a tone spatially. However, some non-embodied DVS, like a red dot tracing the same pitch movement, has shown comparable tone learning benefits to PGs (Morett et al., 2022), leaving the role of gesture embodiment unclear. Moreover, how much of a human agent is visible in DVS varies among recent studies (head+hand, torso+hands, head+torso+hands)--through a learning and media theory lens, this agent embodiment could have a significant learning impact. In this dissertation, I explore the role of embodiment further through a 2x2, pretest-posttest learning intervention study. During 90--120 minutes of typing-based gameplay, 68 adult naive learners of Mandarin Chinese viewed a total of 60 DVS videos, which conveyed the pronunciation of 15 2-syllable vocabulary items, and represented linguistic tone in 1 of 4 ways: (1) a human agent with a hand gesturing, (2) invisible agent with a visible hand gesturing, (3) a still human agent with a red dot gesturing, and (4) no agent with a red dot gesturing. Assessments measured pre- and posttest tone recognition for 1- and 2-syllable nonce words, and posttest word recall of the vocabulary items.First, an interaction effect of gesture embodiment and agent embodiment for word recall revealed that learners benefited significantly more when DVS was either fully embodied or fully non-embodied, than when partially embodied. Findings suggest that the word learning benefit of high embodiment in a DVS agent outweighs other cognitive hindrance introduced by the agent, in line with principles for pedagogical agents from media theory (Mayer, 2017). Conversely, findings reinforce the notion that the visuospatial metaphor is a powerful word learning scaffold, comparably effective to embodied DVS, and more effective than DVS with compromised embodiment. Second, non-significant differences in tone recognition learning gains between conditions suggest that the scaffolding of tone category learning is not sensitive to embodiment effects in the same way as word learning. Post-hoc tests showed significant positive correlation between gameplay progress and word recall, suggesting that using the vocabulary items for problem solving means was particularly helpful for participants. In light of the findings, I suggest educators and designers use highly embodied or fully non-embodied DVS when supporting tonal word learning, especially in hands-on L2 problem solving activities, like digital games. [The dissertation citations contained here are published with the permission of ProQuest LLC. Further reproduction is prohibited without permission. Copies of dissertations may be obtained by Telephone (800) 1-800-521-0600. Web page: http://bibliotheek.ehb.be:2222/en-US/products/dissertations/individuals.shtml.]