ha-HA-hha? Intensity and voice quality characteristics of laughter

Laughter is one of the most widely-encountered paralinguis-tic phenomena in human interaction and it has been studied from different perspectives throughout the years, including its acoustic-prosodic realization. However, previous studies have mostly focused on fundamental frequency and duration measures, with other prosodic features being less studied. We ex-amine here the acoustic marking of laughter in terms of intensity and voice quality characteristics, by using a corpus of spontaneous conversations. We operationalized the two cues by means of the root-mean-square energy and the cepstral peak prominence, respectively. Examining laughs, speech-laughs and speech instances at two different levels (that of the entire event and at the syllable nucleus level) we observed the least regular phonation for laughs and the most regular one for speech, while intensity was the highest for speech-laughs, followed by laughter and the lowest for speech. Using mixed effect models we determined that all three vocalization classes differ significantly from one another, in terms of both acoustic cues. Moreover, an interesting effect of syllable position was seen for laughter, with phonation becoming more regular for later syllables.