AIRS 2nd Annual Meeting: 2010

Title: How do singers manage vowels in relation to the laryngeal mechanism? An acoustic and
linguistic inquiry
Authors: Sylvain Lamesch (LAM-IJLRA, Paris), Michele Castellengo (LAM-IJLRA, Paris), Boris
Doval (LAM-IJLRA, Paris), Luiza Maxim (LAM-IJLRA, Paris)

Abstract

Classical singers are trained to use all the vowels regardless of their predominant tessitura and
laryngeal vibratory mechanism. Comparatively, Yodellers generally choose a given set of
vowels depending on the mechanism they use: M1 (chest voice) or M2 (falsetto). In order to
explore the possible relations between the laryngeal source and the resonance cavities, we have
undertaken an acoustical study of Voice Range Profiles (VRP) - using different mechanisms (M1
and M2) and vowels - as well as a linguistic investigation of the singers’ verbal productions. 21
male and female singers produced crescendos and decrescendos on /a/, /i/ and /o/, from C3 to
C5, in mechanisms M1 and M2. We recorded the sound signal in order to compute the vocal
dynamics. VRPs were obtained for each singer and for each vowel, in M1 and M2 separately. At
the end, the subjects filled a questionnaire concerning the vowels they prefer to sing in M1 and
in M2, respectively. The questionnaire prompted individual evaluations and comfort or
technique related comments from singers. The analysis of the questionnaire showed that singers
prefer singing /a/ rather than /i/ in M1, and inversely in M2. The study of the VRPs showed that
the upper limit is 10 dB louder for /i/ than for /a/ in M1, but not in M2. Consequently the vocal
dynamic range is larger for /a/ than for /i/ in M1, and it is larger on /i/ in M2 than in M1, if
compared with the dynamic range for /a/. In M1, producing a given level on /i/ could require a
higher subglottal pressure. The processed acoustical data provided us with elements, which in
turn allowed us to interpret more accurately the singers' verbal descriptions.