In everyday environments, we often have to attend to one person’s speech (target speech) while ignoring another (competing speech). A competing talker can impair speech processing through both energetic masking (acoustic degradation at the periphery) and informational, cognitively-demanding aspects of the mask. We refer to the latter as informational interference. We hypothesized that informational interference depletes processing resources that could otherwise be allocated to recognizing and understanding target speech. Consequently, informational interference should be more pronounced when the task is more resource-demanding (more or less complex syntax) or when the participants’ own processing demands are elevated (non-native listeners). Finally, modulating the semantic content of the competing talker’s utterances should influence the degree of informational interference.
Using a speeded picture-selection task, we assessed native and non-native listeners’ understanding of spoken sentences varying in syntactic complexity, played with a competing talker or a matched energetic mask, at various signal-to-noise ratios (SNRs). In a follow-up experiment, the semantic content of the competing talker sentences was manipulated to be congruent, incongruent or unrelated to the target sentence. Participants’ performance was measured with accuracy and reaction times from button presses, as well as eye-tracking. Selective attention, short-term and working memory were assessed to determine the contribution of these cognitive factors to informational interference.
Although syntactic complexity affected participants’ performance, the competing talker was not more detrimental than the energetic mask controls, contrary to our hypothesis. This pattern was comparable for native and non-native listeners, and across SNRs. In the follow-up experiment there was no difference between semantically incongruent and neutral competing sentences, but semantically congruent sentences led to faster sentence processing, indicating facilitation or priming. This indicates that the content of the competing talker is not indiscriminately inhibited. Moreover, individual differences in memory and selective attention were not related to differences in the speeded-selection task, regardless of the mask. These results provide little support for the existence of a uniquely informational source of speech masking.
