Infants’ vowel discrimination in modulated and unmodulated noise.

Abstract

Adults detect and discriminate masked sounds better in temporally modulated maskers than in unmodulated maskers. It is not known whether the same is true of infants. In this study, 7–9-month-olds learned to respond when a repeated vowel changed from /a/ to /i/ or from /i/ to /a/ in speech spectrum noise. Discrimination was assessed using an observer-based procedure. In one condition the noise was unmodulated. In the other, the noise was modulated with the envelope of single talker speech. The level of the noise was 60-dB SPL. For infants in group 1, the level of the vowel in both conditions was set so that d′=1 was achieved in the unmodulated noise; discrimination in modulated noise was the second condition tested. For infants in group 2, the level of the vowel was fixed at the average level used to test the infants in group 1; discrimination in modulated noise was tested first. In both groups, d′ in the modulated noise condition averaged only 0.5. Thus, infants were unable to take advantage of masker modulation to improve speech discrimination. Their speech discrimination was poorer in modulated than in unmodulated noise. [Work supported by NIH R01 DC00396 and P30 DC04661.]