Tulling, M. 1 , Pablos Robles, L. 2 & Gryllia, S. 2
1 New York University, USA
2 Leiden University, the Netherlands
Successful communication is achieved when the listener understands the intention behind the proposition uttered. However, important cues about intention often appear sentence finally, namely in the form of discourse particles or boundary tones. This is true for Dutch where final discourse particles hè and hoor mark the proposition with contradictory intentions: either as an agreement-seeking or corrective utterance. This is puzzling for two reasons. First, given that intention precedes other production steps, as most models of speech production assume, it is less optimal that the speaker delays the intentional cue until the end of the sentence. Second, it seems less productive for the listener to process intentional information only after the entire sentence has been heard. In this study, we test the hypothesis that speakers already incorporate intentional cues by means of prosody before the end of the sentence and that listeners use this prosodic information to anticipate intentional cues prior to hearing the final particles. To test whether speakers cue the listeners by using prosodic information before the final particle, we recorded 10 Dutch speakers producing sentences with intentional particles (hè 'agreement-seeking' and hoor 'corrective') and sentences ending in non-intentional elements (man 'addressee marker' and zo 'deictic element'). To determine whether listeners use prosodic cues to anticipate intentional elements, we tested 24 listeners on an audio gating sentence completion task. Production results revealed a significant difference in pitch and duration preceding the particles but listeners were not able to use these prosodic cues to anticipate the different intentional particles in the perception task. These results are compatible with an account of sentence processing in which intention is only incorporated after the propositional content has been processed.