The speech laugh spectrum

This paper reports on the acoustic features of the laugh-speech continuum, specifically dealing with formant space, pitch range and voice quality. Results show that laugh vowels are generally reduced and there is a tendency for them to be produced more back and low in the vowel space when compared to speech. The large open quotient and shallow spectral tilt measurements for laughter is consistent with large sub-glottal pressure, more lateral compression of the glottis and greater glottal airflow. High sub-glottal pressure intrinsically raises fundamental frequency and could possibly explain the high F0 values we report for laugh and speech-laughs. Speech-laugh was found to be an intermediate between speech and laugh where some modulation of the opening phase occurs in order to speak, even with high subglottal pressure. 1. General Information Laughter is a special human vocalization chiefly because it induces a positive affective state on listeners. As more research is conducted on laughter more and more interesting facts and phenomena are revealed. Human babies begin to laugh at about 4 months, preceding early forms of babble, which begins at 6months-1 year of age (Sroufe & Waters, 1976). Even deaf and blind babies can laugh in the absence of auditory and visual feedback (Eibl-Eibesfeldt, 1989: referenced in Bachorowski et al., 2004). While laughter is universal there are many forms of human laughter. Bachorowski and colleagues investigating a large database of laughter sounds have differentiated three basic types namely, “song-like”, “snort-like” which contains nasal turbulence, and “grunt-like” which contains laryngeal and oral cavity frication. They also claim that only 30% of all analyzed laughs were voiced. According to Provine (1996) laughter in human communication does not interrupt the “speech stream”, but act somewhat as punctuation marks in written speech, i.e., proceeding or following phrases. However, it is difficult to imagine that laughter does not influence the acoustics of speech even if it does not disrupt the production of speech segments. Nwokah et al., analyzing motherchild interaction find that about 50% of laughs in conversations co-occur with speech, what they refer to as “speech-laughs”, a simultaneous production of laugh and articulation. Further they claim that “neither speech nor laughter is dominant when combined” and the complex vocal production of speech-laughs is a result of maintaining speech like fundamental frequency and laugh like amplitude and rhythm. Ethologists have presented evidence for a direct evolutionary link between human laughter and chimpanzee vocalizations (van Hoof, undated web source, Provine, 1996). Owren & Bachorowski (2003) propose that the acoustics of both animal vocalization and laughter have direct influence on the affective state of the target animal (man in the case of laughter). In fact, Provine (1996) goes to the extent of stating that “the study of laughter provides a novel approach to the mechanisms and evolution of vocal production, perception and social behavior.” If human laughter shows developmental trends from primate vocalization is it possible that speech has developed from laughter? This is a big question? In this report we investigate the acoustic quality of speech and laughter and the intermediate state of speech-laughs as a small step in trying to understand the phonetic difference in speech and laughter.