Auditory training and subsequent generalization with speech and non-speech stimuli

Abstract

Auditory training studies have varied in degree-of-similarity between the training and testing conditions. Some focused testing on the trained task (e.g., Sweetow and Henderson Sabes, 2007), while others addressed whether the training effects generalize to other tasks. The current study used speech and non-speech stimuli to explore how much training in a specific analytical listening task would generalize to performance improvements for another. Naïve listeners participated. Percent correct was measured as a function of signal-to-noise ratio (SNR) in pre- and post-tests with the “multiple-bursts same” (MBS) paradigm (Kidd et al., 1994) and the speech-target/speech-masker Coordinate Response Measure (CRM) paradigm. There were three training groups in a between-groups design: 1) training in a masked speech identification task (the CRM) with the competitors varying in f0 separation (Darwin et al., 2003); 2) training in the MBS task with the protected-region (frequency region with no maskers) bandwidth around the target varying; and 3) no training. Preliminary data describe individual-subject variability in pre-test performance and indicate performance improvements with auditory training. The degree of generalization will be explored for the trained sub-groups. 1) Int J. Audiol. 46. 374. 2) J. Acoust. Soc. Am. 114, 2913. 3) J. Acoust. Soc. Am. 95, 962.