Christophe Savariaux
Pascal Perrier
Jean-Pierre Orliaguet
ICP, INPG & Univ. Stendhal, URA CNRS 368, 46 Ave. Felix Viallet, 38031 Grenoble Cedex, France
An experiment is described involving lip perturbation, using a large (20 mm) Plexiglas tube, during the production of the rounded french vowel [u]. Ten naive speakers pronounced isolated repetitions of [u] in three conditions: (1) without perturbation, (2) with the lip-tube after a learning period of 20 repetitions, and (3) with the lip-tube after 20 repetitions with information on the way to deal with the perturbation. X-ray recordings were made at the end of each phase, and formants were extracted from the acoustical record. Compensation abilities were evaluated by considering the first three formants, in reference to the difference limens [Mermelstein, J. Acoust. Soc. Am. 63, 572--580 (1978)]. At the end of the second phase, one speaker was able to compensate for the perturbation. For all other speakers, we observed a strong increase in F2. For the third phase, the speakers were divided into three groups who were given different information in order to help them compensate for the perturbation (first group: articulatory information, second group: acoustic information, third group: both types of information). Speakers in the first group all produced a fairly good [u]. Speakers in the second group showed some enhancement, but the F1--F2 targets were not reached. In the third group, no reliable compensation was observed. Some interpretations are proposed concerning compensation strategies and speech space control.