Worth a thousand words: absolute and relative decodability of nonlinguistic affect vocalizations

The authors compared the accuracy of emotion decoding for nonlinguistic affect vocalizations, speech-embedded vocal prosody, and facial cues representing 9 different emotions. Participants (N = 121) decoded 80 stimuli from 1 of the 3 channels. Accuracy scores for nonlinguistic affect vocalizations and facial expressions were generally equivalent, and both were higher than scores for speech-embedded prosody. In particular, affect vocalizations showed superior decoding over the speech stimuli for anger, contempt, disgust, fear, joy, and sadness. Further, specific emotions that were decoded relatively poorly through speech-embedded prosody were more accurately identified through affect vocalizations, suggesting that emotions that are difficult to communicate in running speech can still be expressed vocally through other means. Affect vocalizations also showed superior decoding over faces for anger, contempt, disgust, fear, sadness, and surprise. Facial expressions showed superior decoding scores over both types of vocal stimuli for joy, pride, embarrassment, and "neutral" portrayals. Results are discussed in terms of the social functions served by various forms of nonverbal emotion cues and the communicative advantages of expressing emotions through particular channels.