Influence of Semantic Context on Speech-in-Noise Performance: Evaluating the Quick Speech-in-Noise Test.

Abstract

Purpose: Sentences are encoded with sematic context, which facilitates audiologic ability to navigate background noise, or speech-in-noise (SIN), conditions. To examine how semantic context contributes to performance on one commonly used SIN test, the Quick Speech-in-Noise Test (QuickSIN) by Etymotic Research, Inc. (henceforth "QuickSIN"), we use a novel experimental paradigm that isolates semantic information.

Method: Ten college-aged monolingual participants with typical hearing listened to 72 sentences delivered in 0, 5, 10, 15, 20, or 25 dB SNR followed by a choice between two visual words. One word was related to the overall sentence meaning, and the other word was unrelated. The reaction time (RT) to correctly select related targets was measured to index usage of semantic information.

Results: Participant's RTs to select a correct response were compared across different signal-to-noise ratios (SNRs). We found that less favorable noise conditions (0, +5 dB SNR) elicited a greater usage of semantic information than more favorable noise conditions (20, 25 dB SNR). Transformed RT data were analyzed with nonparametric tests that assessed the homogeneity of variance within responses to each SNR condition. Results indicated that participants' RTs were consistently varied within each SNR condition, except to sentences in +20 dB SNR, indicating an imbalance in the degree of semantic context used in the sentences in that SNR level.

Conclusions: Respondents to the QuickSIN use semantic context to facilitate processing especially at less favorable SNR levels, which is consistent with research supporting a greater role of semantic information during suboptimal listening conditions. Differences in context use across noise conditions means test performance also reflects language processing and should be considered for updated tests of speech-in-noise performance. Critically, responses to sentences at the +20 dB SNR used in the QuickSIN are so inconsistently varied in their degree of semantic usage as to prohibit a clinical interpretation alongside the other conditions. These findings warrant the development of a quick-to-administer SIN test with stimuli that are balanced for semantic expectancy to avoid language effects.