Demonstrates that iSTART, a system which attempts to automate Self-explanation strategies training, performs about as well as the training delivered by humans in McNamara, D. (2004). SERT: Self-explanation reading training. Discourse Processes, 38(1), 1–30. In this experiment, there’s moderate improvement, but only for text-based questions.
Students are given a knowledge test, as in (McNamara, 2004), but no extended analysis here about high-vs-low-knowledge students, other than a note that high-knowledge students performed better.