The effect of energetic and informational masking on the time-course of stream segregation: Evidence that streaming depends on vocal fine structure cues

Abstract

To examine the effect of energetic and informational masking on the time-course of stream segregation, we presented listeners with semantically anomalous but syntactically correct target sentences (e.g., “A house should dash to the bowl”) that were masked by a two-talker speech masker or steady-state noise masker. To determine the effect of each masker on the time-course of stream segregation, we measured performance as a function of keyword position (key words in italics). The results from Experiment 1 showed that performance improved as a function of keyword position under speech masking, but was relatively stable across keyword positions under noise masking. The results of subsequent experiments showed that the variation in performance across keywords under speech masking was primarily due to the vocal similarities between the competing talkers, and that interference from the semantic content of the masker played a secondary role in undermining performance. Taken together, these results indicate that stream segregation takes longer to build up when a speech target is masked by other speech in the absence of cues that aid stream segregation (e.g., spatial separation), but that it takes little time to build up when a speech target is masked by a noise or when cues that aid stream segregation are available to listeners.

Keywords:

• Information masking
• Speech masking
• Streaming
• Perceptual segregation

Acknowledgments

This research was supported by the Canadian Institutes of Health Research [MOP-15359] [STP-53875], Natural Sciences and Engineering Research Council of Canada [RGPIN 138472 and RGPIN 9956] and the National Natural Science Foundation of China [30711120563].

Notes

¹Previous studies have indicated that robust effects using these stimuli in this informational masking paradigm can be obtained with as few as 12–16 participants (e.g., Li, Daneman, Qi, & Schneider, 2004, 12 subjects; Freyman, Balakrishnan, & Helfer, 2004, 16 subjects per condition in the priming conditions). Hence, we were confident that 16 participants would be sufficient to explore the effects of word position in these experiments.

²The repetitious nature of the speech masker may have allowed participants to become familiar with this masker over the course of the experiment. Consequently, it is possible that performance might improve over time due to the participants becoming familiar with the masker.

³This spectral difference between the two types of maskers indicates that the amount of energetic masking produced by the two types of maskers may have differed.

⁴ y represents the probability of correctly identifying a keyword, x is the SNR in dB, µ represents the dB SNR level corresponding to 50%-correct performance, and σ determines the slope of the fitted function.

⁵Keyword 1 vs. Keyword 2, mean difference = 0.469 dB, t(30) = 2.713, p>.05; Keyword 1 vs. Keyword 3, mean difference = 0.063 dB, t(30) = 0.365, p>.05; Keyword 2 vs. Keyword 3, mean difference = 0.406 dB, t(30) = 2.350, p>.05.