תקציר:

We examined the impact of variability in speech stimuli on improvement of general performance and on accessibility to low-level information as a function of practice. Listeners had to discriminate between two similar words in noise in two configurations that differed only in their low-level binaural information, which was either null or maximal. The difference in performance quantifies the use of binaural low-level information. These configurations were presented in three training protocols: in separate blocks; in a consistently interleaved manner; and in a randomly mixed manner. The first protocol enabled optimal use of the low-level binaural cues already at the first training session. The second, consistently interleaved protocol required more than one training session to reach the same performance. The final, mixed protocol did not enable optimal use of the low-level cues even after multi-session training. Interestingly, training with the first two protocols transferred to the mixe