Figure 2.

Average percent correct categorization for familiar training, novel generalization, and novel isolated target word stimuli in Experiment 1 (fluent Korean speech) and Experiments 2 and 3 (spectrally rotated English speech) (^* indicates that p ≤ .001). The line at 25% indicates chance-level performance. The learning exhibited across all experiments and all test stimuli was significantly above chance (p ≤ .001). Error bars indicate standard errors of the mean.