The aha effect is a memory advantage for initially ambiguous stimuli (e.g., "The notes were sour because the seam split.") that are subsequently resolved ("Bagpipe") over stimuli that are comprehensible from the outset (see P M. Auble, J. J. Franks, & S. A. Soraci, 1979; T. W. Wills, S. A. Soraci, R. A. Chechile, & H. A. Taylor, 2000). The authors examined the influence of learning context on the aha effect by manipulating whether generative tasks were completed in a group setting or alone. In Experiment 1, participants in a group context either discussed difficult aha sentences, or overheard the discussion. In Experiments 2 and 3, lone participants were exposed to the sentences, and either spoke their thoughts aloud or thought silently. Participants in a group context did, as predicted, exhibit the aha global deletion of quotation marks effect. Moreover, in all experiments an aha effect was found for the active, speaking condition only. The authors discuss implications for generative learning and collaborative discourse processes.