Teachers and students alike would benefit from self-access materials to free up lesson time for activities requiring live face-to-face interaction, but such materials must first undergo thorough testing and evaluation of their pedagogical effectiveness. The present study is part of an ongoing project to develop a series of self-paced, interactive online modules on the use of grammatical voice. Previous acceptability judgment data showed these materials to be effective for improving performance on sentences with animate subjects but largely ineffective or even detrimental in cases with inanimate ones. Thus, 10 returning participants were individually recorded re-attempting the same acceptability judgments, this time stimulated by structured interview questions to probe decision-making processes and thereby permit step-by-step analysis of conceptual understanding. The findings included common failure to grasp the role of personification in permitting false agentive subjects in English, conflation of the use of inanimate subjects with passive voice, and unawareness of relationships between grammatical case-marking and thematic relations.[For the complete volume, see ED574893.]