Emotional Effects of Fundamental-Frequency Transformations

Fundamental frequency was derived from the utterances of speakers of American English. Each read a total of 40 simple sentences. The words of these sentences express fear, anger, joy, or “neutrality.” Each sentence was read 4 times by the speakers, who attempted to convey one of these emotions by their manner of speaking, even though the lexical content and the emotion might not correspond. Test stimuli were synthesized by exciting a fixed POVO with several sets of fundamental frequency contours: (1) the contours derived directly from the original signal, and with the mean fundamental frequency raised, normal, or lowered; (2) the original contours with the frequency range compressed, and with the mean fundamental frequency raised, normal, or lowered; (3) the original contours with the frequency range increased, and with the mean fundamental frequency raised, normal, or lowered. These stimuli were first presented to subjects in forced-choice categorization tests in which subjects had to decide whether the POVO's output conveyed fear, anger, joy, or “neutrality.” Listening tests in the free-choice mode were also performed. The effect of these transformations on the emotions conveyed are reported and interpreted in the light of a general theory regarding the function of the prosodic speech features in language.