Reversed speech comprehension depends on the auditory efferent system functionality

In the present study we explore the implication of high and low level mechanisms in degraded (time-reversed) speech comprehension in normal hearing subjects. In experiment 1 we compared the loss of intelligibility due to the increasing size of reversion windows in both words and pseudowords. Results showed that words are generally reconstructed better than pseudowords, suggesting the existence of a lexical benefit in degraded speech restoration. Moreover, there was greater variability between individuals when reconstructing pseudowords than words. In experiment 2, we demonstrated that this interindividual variability correlated with the subjects’ medial olivocochlear bundle functionality, as measured by contralateral suppression of otoacoustic emissions (OAEs). Together these experiments highlight the importance of low-level auditory mechanisms in degraded speech restoration. Moreover they put forward the existence of major interindividual variability in the capacity to reconstruct degraded speech, which correlates with the physiological properties of the auditory system (low-level property). In addition, our results also suggest the existence of multiple higher-level strategies that can compensate on-line for the lack of information caused by speech degradation.