Whistling in alveolar fricatives: Evidence from two speech modes

The tongue constriction observed in alveolar fricatives, forming an unstable jet and boundary layer at the teeth, can be accompanied by a front cavity feedback mechanism that reinforces this instability, generating a narrow-bandwidth peak as observed in whistling. This study presents new evidence of a generalised use of this mechanism in speech, based on acoustic recordings of eight female speakers from the same dialectal region in Portugal. Sustained fricatives /s, z/, six disyllabic words, six sentences and a phonetically balanced text, were produced in voiced and whispered speech modes. Variation of the multitaper Power Spectral Density of not whistled/ whistled fricative pairs was assessed using Functional Principal Component Analysis. The first of two principal components explained 93.3% of the /s/ variance and 90.4% of the /z/ variance. Results also revealed the following regarding the first shape descriptor used to model the curves: /s/ – sd = 969.55, R2 = 0.170, p = 0.000, /z/ – sd = 931.41, R2 = 0.105, p = 0.000; and the second shape descriptor: /s/ – sd = 257.32, R2 = 0.009, p = 0.248, /z/ – sd = 299.57, R2 = 0.051, p = 0.007. Whistled fricatives were observed both in voiced and whispered speech, so this does not seem to be a mechanism used to compensate for a weaker source strength as typically observed in whispered speech. Nevertheless, reconstructed not whistled/whistled curve pairs were significantly different.