Consequential Parameters Research Articles

L'encodage labial des voyelles du Français

The examination of the speech signal for correlates to the phonological system led us to elaborate two constraint levels between the correlate and the feature, where the property and the cue, units for the decoding process, may be tested (Abry & Boë, 1979 and 1981b). The most discriminating correlate valid for all contexts is generally given priority due to the system's distinctive function, all other things being equal. In contrast to this search for context free invariants, we have shown the usefulness of taking into account the structure of cue distribution in the treatment of voicing ( ibid). We now propose to challenge, beyond the criterion of invariance, the criterion of discrimination, which in effect gives preference to the decoding process, regardless of encoding needs. Our data concern rounding in French vowels (Abry et al., 1980a). From the start, this data showed lip width (distance between corners) as best single discriminator, and thus as the primary feature [round] physically defined as such by Ladefoged in 1975. However, the introduction of articulatory constraints shows this discriminator is merely the consequence of protrusion-retraction and opening-closing maneuvers by the lips. Moreover, an examination in this frame of mind of the French phonological system shows that protrusion can be predicted from closure, notwithstanding important effects of co-articulation (e.g.: [ staggered: i w ] following [∝, 3]). It is therefore possible to propose a descriptive framework for the generation of lip slit shapes in French, beginning with no more than classifying feature [round] and specified, from the phonological level on, in closure maneuvers, then in positions, and finally in terminal shapes. These shapes include geometric characteristics required to preserve acoustic and visual information, both of which, as correlates, may be validated as properties and cues. The fact that, at decoding, a consequential parameter (such as lip width) has a chance of keeping its discriminatory power is thus confusing only to those for whom decoding is the exact reverse of encoding.