Listening to noise-masked speech results in interrupted speech “glimpses” requiring auditory and cognitive factors for speech recognition. Older adults with normal (ONH) or impaired (OHI) hearing completed measures of vocabulary knowledge, working memory (WM), pure-tone detection, and fluctuating masker benefit (FMB). They also completed various conditions of degraded sentences in speech-modulated noise and an interrupted speech task. An ideal binary mask identified glimpsed and non-glimpsed keywords based on proportions of spectrotemporal speech fragments above −6 dB signal-to-noise ratio. Measures of auditory and cognitive function were entered into a dominance analysis separately for ONH and OHI, which determined the relative importance of each predictor in the presence of all other predictors. Auditory and cognitive measures accounted for 51%–69% (ONH) and 91% (OHI) of the variance in keyword recognition, with most of the variance explained by pure-tone thresholds. Greater contributions were observed from vocabulary knowledge for ONH for non-glimpsed keywords and, for OHI, from FMB and WM for glimpsed and non-glimpsed keywords. Thus, ONH listeners appeared to capitalize on linguistic cues, whereas OHI listeners were limited by their ability to glimpse and process degraded speech in WM. Preliminary results indicate that an interrupted-speech task may provide a simplified assessment of keyword glimpsing. Work supported by NIH/NIDCD.
Fogerty et al. (Wed,) studied this question.