This paper presents a study analyzing the prosody of spontaneous speech exemplified by the OSU Buckeye corpus of American English employing the Fujisaki model and syllable duration analysis. Results show, inter alia, that model parameters are much more strongly influenced by the higher level discourse structure than by their syntactic function. The study shows that modeling strategies can be employed in a similar fashion as for more formal reading style speech.
Bibliographic reference. Mixdorff, Hansjörg (2008): "Quantitative prosodic analysis of spontaneous speech", In INTERSPEECH-2008, 1195.