Successful speech communication requires the extraction of important acoustic cues from irrelevant background noise. In order to better understand this process, this study examined the effects of background noise on mismatch negativity (MMN) latency, amplitude, and spectral power measures as well as behavioral speech intelligibility tasks. Auditory event-related potentials (AERPs) were obtained from 15 normal-hearing participants to determine whether pre-attentive MMN measures recorded in response to a consonant (from /ba/ to /bu/) and vowel change (from /ba/ to /da/) in a double-oddball paradigm can predict sentence-level speech perception. The results showed that background noise increased MMN latencies and decreased MMN amplitudes with a reduction in the theta frequency band power. Differential noise-induced effects were observed for the pre-attentive processing of consonant and vowel changes due to different degrees of signal degradation by noise. Linear mixed-effects models further revealed significant correlations between the MMN measures and speech intelligibility scores across conditions and stimuli. These results confirm the utility of MMN as an objective neural marker for understanding noise-induced variations as well as individual differences in speech perception, which has important implications for potential clinical applications.
Bibliographical noteFunding Information:
This research project was supported in part by the Charles E. Speaks Graduate Fellowship (TKK) , the Bryng Bryngelson Research Fund (TKK and YZ) , the Capita Foundation (YZ) , the Brain Imaging Research Project award (YZ) from the College of Liberal Arts, University of Minnesota . Participant payment was partially supported by NIDCD R01-DC008306 (PBN) . We would like to thank Drs. Edward Carney, Sharon Miller, Yingjiu Nie, and Adam Svec for their assistance. Author contributions: YZ and TKK designed and performed research; TKK, YZ, and BW analyzed data; PBN and HZ provided consulting; TKK and YZ wrote the paper.
- Linear mixed effects model
- Speech-in-noise perception
- Theta band
- Time-frequency analysis