Speech preparation in conversational turn-taking

In everyday conversation, speaker changes occur surprisingly quickly, with gaps between turns of around 200-300 ms, compared to reaction times of around 600 ms in laboratory naming studies. To achieve these rapid turn transitions, speakers have to plan the next utterance during the current turn, as soon as they know what they want to say as the next speaker. In this project, we will investigate what triggers the onset of planning and how listeners predict the end of the current speaker’ turn in order to avoid overlaps and excessively long gaps. Until now, the time between the end of the current speaker’s turn and the start of the next speaker’s turn in conversations, called the floor transfer offset (FTO), has been based on measurements of the acoustic signal. The same is true for reaction time (RT, measured from trigger onset to the response onset) in naming experiments. However, speech preparation begins well before the acoustic onset, with an inhalation (at least for longer utterances) and the initiation of the articulatory gestures to produce the initial speech sound(s). Therefore, using the acoustic onset to measure speech onset overestimates planning time in FTO and RT. The aim of this project is to disentangle planning and execution in speech preparation in order to investigate how and when speakers initiate speech. Simultaneous recordings of respiratory activity with Respiratory Inductance Plethysmography (RIP) and of speech movements with Electromagnetic Articulography (EMA) in a sophisticated experimental design will provide precise measures of the timing of speech preparation, allowing us to better understand how turn-taking works. To this end, we will compare respiratory activity and speech organ movements in highly controlled naming experiments and in two dialogue scenarios.