Table 1.

Categorization training routine

Step	Trained tokens		Minimum accuracy (%)
	Category A	Category B
I	1	7	90 × 1
II	1, 2	6, 7	90 × 1
III	1, 2, 3	5, 6, 7	90 × 3
IV	3	5	90 × 3

Note: In each of the 4 sessions, participants were trained to categorize the phonemic and then the nonphonemic sounds in 4 steps, starting with the end tokens and gradually introducing token pairs closer to the category boundary. In the final step, the boundary tokens were further reinforced. Ninety percent accuracy on a block of 24 trials was required once for steps I and II and three consecutive times for steps III and IV before advancing to the next step. The duration of a typical training session was approximately 1 h.