The role of prosody and voice quality in indirect storytelling speech: Annotation methodology and expressive categories

Storytelling speech entails a particular speaking style that contains subtle expressive nuances represented by specific acoustic cues. In the literature, different pieces of research have addressed its analysis following diverse approaches, some of them studying these subtleties by attending only to prosodic variations. In this work, we propose an analysis methodology to annotate this particular speaking style at the sentence level based on storytelling discourse modes (narrative, descriptive and dialogue), besides introducing narrative sub-modes denoted as expressive categories. Following this annotation methodology, we analyse the role of both prosody and voice quality in indirect storytelling speech through statistical and discriminant analyses. The results obtained on the oral interpretation in Spanish of a tale show the viability of the annotation methodology, since most of the sentence-level utterances are successfully classified after several inter-annotator agreement verifications. Moreover, the outcome of the statistical and discriminant analyses prove that voice quality is as important as prosody to characterize storytelling expressive categories. Finally, we also show evidences that such categories can be conveyed with subtle expressive variability.