Abstract

This thesis investigates speech production and self-monitoring by using a newly constructed Real-time Speech Exchange system. This system makes it possible to covertly manipulate the auditory feedback that participants receive of their own voice. First a word or utterance made by the participant is recorded, and later in the test this recording is inserted into their auditory feedback with precise timing, while the feedback of what they are actually saying is simultaneously blocked. This way, a situation is created where participants say one thing but hear themselves, in their own voice, saying something else. Two main studies were performed. In the first study, participants performed a Stroop test while single color-word utterances were covertly exchanged. When the timing was good, a majority of the voice exchanges were undetected, and in 85% of non-detected trials participants experienced the inserted word as self-produced. This demonstrates how there is a strong inferential component to the sense of agency for speech. Importantly, it shows how auditory feedback is actively used to specify the meaning of what the speaker is saying. In the second study, an argument for the so-called internal monitoring channel was investigated. It has been assumed that no speech errors that were interrupted quicker than 350 ms could have been detected with the auditory monitoring channel. Instead, it has been proposed that a prearticulatory internal channel is used. In this study, RSE was used to simulate speech errors. Participants had been told that their auditory feedback would sometimes be manipulated, and that they were supposed to interrupt their speech as quickly as possible when this happened. Results showed that more than 50% of the interruptions were made within 350 ms, and some were even made before 100 ms. This demonstrates that swift interruptions can be made based on auditory feedback alone. Consequently, it might not be necessary to postulate an internal monitoring channel to explain brief interruption times.