Release from informational masking by time reversal of native and non-native interfering speech

In many studies, the influence of intelligibility of the interfering speech is avoided by reversing it in time. Usually, intelligibility with time-reversed interfering speech indeed is higher compared to that with normal interfering speech. However, due to the nature of speech, reversed speech also gives rise to increased forward masking. The latter will result in a decrease in intelligibility. Thus, differences in intelligibility as a consequence of reversing speech in time are due to two opposite effects. This paper describes a speech reception threshold (SRT) test with intelligible and unintelligible interfering speech played normally and time-reversed. With Dutch listeners, Swedish reversed interfering speech gave a rise in SRT of 2.3 dB compared with the Swedish interfering speech (played normally). The difference can be attributed to differences in forward masking. Dutch time-reversed interfering speech gave a decrease in SRT of 4.3 dB compared to (intelligible) Dutch interfering speech. The latter is the result of both a release from informational masking and an increase in forward masking. Therefore, the amount of informational masking is larger than 4.3 dB and, if one assumes similar differences in forward masking for Dutch and Swedish speech, may amount to 6.6 dB.