(F08)
Attina, Virginie
Cathiard, Marie-Agnès
Beautemps, Denis
(Institut de la Communication Parlée, UMR CNRS 5009, Grenoble)
French Cued Speech production : giving a hand to speech
Human speech is multimodal by nature : it is now generally accepted
that the perception of the acoustic speech signal is largely enhanced
by visible speech face and lips gestures. In addition, non-verbal spontaneous
gestures are an integral part of language (McNeill, 1992).
The coproduction of sound, face and manual gestures allows optimal
communication and understanding. The mechanisms underlying the
relationships between gestures and speech are at the center of many
studies. Concerning timing, the gestures onset is found to precede
speech emission.
This study aims to extend the “sound-gestures” timing relationship
retrieved for spontaneous speech as detailed previously to additional
hand gestures necessary to improve oral speech reception for deaf
people. This work focuses on French Cued Speech production. Cued
Speech (CS ; Cornett, 1967) is an effective system that uses hand
cues placed near the face of the speaker to disambiguate lip shapes.
While uttering, the speaker codes each consonant-vowel (CV) syllable
with a manual cue : the shape of the hand distinguishes among consonants
and its position around the face is devoted to vowel disambiguation.
In order to enlarge the understanding of CS effectiveness
(demonstrated for speech perception and for language acquisition),
it is important to know how CS is produced.
Four trained French CS speakers have been video recorded while
uttering and coding CV syllabic sequences. Data are obtained by
video tracking of hand and oral gestures in order to extract signal
features related to lip contours and hand movements. Lip area, x and
y hand position and acoustic signal are manually labeled at onsets
and offsets times. Results reveal a similar temporal pattern for the 4
subjects : the hand gesture onset precedes the syllable onset and the
hand reaches its cue position during the first part of the consonant,
so well before the vowel labial target.
