PROSODY PROCESSING IN SPEECH PRODUCTION: PRE-EVALUATION OF A fMRI STUDY

Due to noise emission of the MR scanner during functional imaging it is impossible to evaluate subjects' performance on speech production tasks online. Therefore we choose a pre-evaluation paradigm to extract the relevant phonetic parameters that characterize subjects' utterances. All experimental tasks were first recorded in an anechoic chamber before subjects passed on to the fMRI session. All subjects were recorded under two conditions: (1) silence and (2) with fMRI noise presented over headphones. We applied this procedure to answer two questions: (1) Does the scanner noise influence speech production, particularly the prosodic features of the speech signal? (2) Is the subjects' performance on experimental tasks as intended? Our results indicate that MR scanner noise does not significantly interact with prosody generation and that our experimental paradigms yield the intended results.