Abstract :
[en] The acoustic properties of nasal vowels differ strongly from those of their oral counterpart.
First, nasal coupling induces a general damping of energy for nasalized vowels. In particular, energy is lost around F1 (bandwidth widening) and around F3 (up to cancelling). Second, many additional articulations are involved in the phonetic realization of French nasal vowels. Differing across vowels, these articulations include lip rounding, tongue lowering, tongue retraction, etc. But they share a common acoustic consequence : the lowering of F2 (Delvaux et al. 2002, among others).
This study concerns two perceptual experiments that have been run (1) to test the relevance of these acoustic cues for the detection of the [nasal] contrast, and (2) to address the issue of the perceptual integration (vs independence) of the acoustic dimensions.
Both experiments use 5-by-5 arrays of synthetic stimuli. The stimuli are C1VC2 items, where C1 [m,t] and C2 [t] are natural sounds, and V are synthetic [?, ? ,?]. The spectral properties of the vowel are varied along two dimensions per array, from an oral to a nasal(ized) endpoint. Two acoustic dimensions have been selected for manipulation : (1) 'compactness', which involves F1 broadening, F2 sharpening and F3 cancelling, and (2) 'gravity', which consists in the lowering of F2. Sixteen Belgian French listeners took part in the experiments, divided in three groups, one per vowel.
First, a trading-relations experiment has been carried out, in order to test the hypothesis that both compactness and gravity are taken into account by French listeners in deciding whether a vowel is oral or nasal. Huge boundary shifts from one identification function to another show that the differences in the oral articulations are as much a part of the contrast as the soft palate lowering. The lowering of F2 is even more necessary for NVC items, supposedly to enhance the contrast in a nasal context.
Second, the integration of dimensions has been measured using a fixed classification paradigm on a subset of the stimuli (MacMillan et al.1999). Results show that the covariation of nasal coupling and oral articulations largely increases the perceptual distance between vowels for French listeners.
The perception of the same stimuli by American English listeners has also been tested in each of these tasks to determine whether these processes are language-specific or not. If the perceptual integration is also attested for those listeners, French speakers can be considered as taking advantage of a universal perceptual property in order to maximize the [nasal] contrast within their crowded vocalic system.
References
MacMillan N.A., Kingston J., Thorburn R., Dickey A.W., Bartels C. (1999). Integrality of nasalization and F1. II Basic sensivity and phonetic labeling measure distinct sensory and decision-rule interactions, JASA 106 (5), 2913-2932.
Delvaux V., Metens T., Soquet A. (2002). French nasal vowels: articulatory and acoustic properties. Proceedings of the 7th International Conference on Spoken Language Processing, Denver, 1, 53-56.
