Percepção de correlatos prosódicos da distinção dado-novo em português brasileiro

Resumo

The experiment described here is based on previous descriptive work on prosodic patterns in coreference chains which showed that new referents tend to be more prominent than given ones. The hypothesis being tested in the present experiment states that listeners use prosodic prominence as cue to referents' status. Sentences taken from the production experiment corpus were acoustically manipulated to create three artificial prominence distributions other than the natural New-Given: Given-Given, Given-New amd New-New. In the experiment listeners had to compare a natural sentence with one of the manipulated versions. Results show that listeners prefer the natural sentences to the manipulated ones. The most rejected manipulation is the one that reverses the prominence levels of new and given referents. Manipulations in which both referents have the same prominence level tend not to be rejected.