Motivated by a consideration of the machine-mediated nature of human interaction in web-based tutoring, we propose the construction of artificial expressions, displays which reflect users' felt bodily experience, to support the development of greater empathy in remote interaction. To demonstrate the concept of artificial expressions we have implemented a system for the real-time visual display of affective signals such as respiration, pulse, and skin-conductivity which, combined with contextual information, may help partners in a learning interaction to estimate one another's level of arousal, stress, or boredom, for example. We have employed this system in a trial learning situation for the remote teaching and learning of Janka, the Chinese characters used in written Japanese. (Contains 6 figures.)