Speech-Induced Suppression for Delayed Auditory Feedback in Adults Who Do and Do Not Stutter

Abstract

Speech-induced suppression is the normal, relative amplitude reduction of the auditory evoked potential for self-, compared to externally-generated, auditory stimulation. It remains controversial as to whether adults who stutter exhibit expected auditory modulation during speech; some studies have reported a significant difference between stuttering and fluent groups in speech-induced suppression during speech movement planning, while others have not. We compared auditory evoked potentials (N1 component) for auditory feedback arising from one's own voice (Speaking condition) with passive listening to a recording of one's own voice (Listening condition) in 24 normally-fluent speakers and 16 adults who stutter under various delayed auditory feedback (DAF) time conditions (100 ms, 200 ms, 500 ms, and 1,000 ms). We presented the participant's own voice with a delay, immediately after presenting it without a delay. Our working hypothesis was that the shorter the delay time, the more likely the delayed sound is perceived as self-generated. Therefore, shorter delay time conditions are proposed to result in relatively enhanced suppression of the auditory system. Results showed that in fluent speakers, the shorter the delay time, the more the auditory evoked potential in the Speaking condition tended to be suppressed. In the Listening condition, there was a larger evoked potential with shorter delay times. As a result, speech-induced suppression was only significant at the short delay time conditions of 100 and 200 ms. Adults who stutter did not show the opposing changes in the Speaking and Listening conditions seen in the fluent group. Although the evoked potential in the Listening condition tended to decrease as the delay time increased, that in the Speaking condition did not show a distinct trend, and there was a significant suppression only at 200 ms delay. For the 200 ms delay condition, speakers with more severe stuttering showed significantly greater speech-induced suppression than those with less severe stuttering. This preliminary study suggests our methods for investigating evoked potentials by presenting own voice with a delay may provide a clue as to the nature of auditory modulation in stuttering.

Keywords: EEG; auditory evoked potentials; delayed auditory feedback; speech-induced suppression; stuttering.

Figure 1

Speaking condition in the electroencephalography (EEG) experiment. Participants were instructed to vocalize /a/ very lightly soon after that a gray circle drawn on a black background changed to a gray square. Three midline electrodes Fz, Cz, and Pz were used for calculating auditory evoked potential in response to auditory feedback of speech. The speech signal was sent into an auxiliary EEG channel for offline-extraction of the onsets of individual speech.

Figure 2

Auditory evoked potentials in fluent speakers. (A) Averaged auditory evoked potentials under each delay time condition. The blue line represents the Listening task and the red line represents the Speaking task. (B) N1 amplitude extracted from each participant’s auditory evoked potentials. Graphs represent mean ± SEM. There were significant differences between Listening and Speaking conditions for the N1 amplitude under the 100 ms and under 200 ms (p < 0.05) delay conditions with Tukey’s HSD test, but no significant differences under the 500 ms or 1,000 ms delay conditions.

Figure 3

Auditory evoked potentials in the stuttering group. (A) Averaged auditory evoked potential under each delayed auditory feedback (DAF) time condition. The blue line represents the Listening task and the red line represents the Speaking task. (B) N1 amplitude extracted from each participant’s auditory evoked potential. Graphs represent mean ± SEM. Only the 200 ms delay condition showed a significant difference in comparison with Tukey’s HSD test between the N1 amplitude for Listening and Speaking conditions (p < 0.01).

Figure 4

Magnitude of the speech-induced suppression under the 200 ms condition of the stuttering group. The stuttering group (n = 12) was divided into two subgroups (n = 6 vs. 6) by the median value of %SS. Graphs represent mean ± SEM. There was a significant difference between these subgroups (t₍₁₀₎ = 2.702, p < 0.05), suggesting that more severely stuttered speakers suppressed the perception of their auditory feedback voice more when vocalizing /a/ under the 200 ms DAF condition. Star (*) indicates p < 0.05.

Figure 5

Delay time dependence of auditory evoked potentials in the fluent group (A) and stuttering group (B). The dotted line represents the transition of the average of the evoked potential under each delay time condition and the solid line represents the average of the regression lines estimated from each participant. Note that because the Y-axis is inverted, the beta value is opposite in sign to the slope of the regression line. Star (*) indicates p < 0.05.