Acoustic cues to perceived prominence levels: Evidence from German spontaneous speech

Abstract

A corpus study on German spontaneous speech proved the robustness of a meaningful perceptual distinction between two levels of prominence, namely fully-fledged versus secondary accents, and its effective use in prosodic annotation. We found that the two levels are characterized by prosodic profiles that mainly differ in gradient phonetic features (F0 range, duration, intensity) and are only to some extent influenced by the choice of phonological accent type. Furthermore, the distinction between the two levels turned out to be largely independent of medial or final accent positions in the phrase.