Laughter: Perceptual evaluations of laugh sounds with known acoustic variability

Acoustic analysis of laughter recorded under controlled social contexts revealed significant variability on a number of source- and filter-related acoustic measures. Here, the results of three experiments used to test listeners’ (n=84) evaluations of this acoustic variability are described. The 80 laugh stimuli used in each experiment were selected on the basis of laugher sex, percentage of voicing, F0 contour, and overall laugh duration. Listeners evaluated each stimulus twice using one of three 4-point rating schemes (goodness for inclusion in a laugh track, sincerity, and the extent of positive emotional responding). The results indicate that listeners prefer laughs with voiced, song-like qualities, and suggest that particular patterns of acoustic features in this nonlinguistic human vocal signal are more likely to engender positive emotional responses in listeners. Further analyses will examine the links between listener evaluations, variation in laugh vowel sounds, and acoustic correlates of laugher body size. [Work supported by NSF.]