Prosodic Cues to Emotion: Perceptual and Acoustic Analyses

The present study examined the perceptual categorization of spoken sentences designed to convey five different emotions via semantic and prosodic cues. As well, the acoustic cues associated with the sentences that varied in emotional prosody were measured. Based on previous findings listeners were predicted to be able to recognize all emotions with a better than chance accuracy, and second, that the accurate perception of emotions would imply that there are emotion-specific patterns of voice cues (Johnstone & Scherer, 2000). 45 English listeners were presented the sentences under two conditions: a) English sentences that varied in semantic affect but which had neutral prosodic affect (semantic condition), and b) French sentences that varied in prosodic affect but which had no semantic information for English listeners unfamiliar with French (prosodic condition). The listeners were asked to categorize the sentences they heard according to five categories: happy, sad, anger, fear, and neutral. Listeners were able to identify all affect categories at a level above chance for both conditions, exception for `fear' and `neutral' sentences. Seven acoustic cues (e.g., mean F0) were measured in the prosodic stimuli and were compared to the categorization data in order to determine if specific cues were linked to individual emotions. The results partially replicated findings in previous studies linking acoustic cues to emotions. Overall, the results extend knowledge of which acoustic cues are important for discriminating emotions and provide a baseline for emotion categorization performance that can be compared with populations that show deficits in perceiving affect in spoken language.