Extending the Movement Expansion Model (MEM) for rounding from French to English Aude Noiray 1 , Lucie Ménard 2 , Marie-Agnès Cathiard 1 , Christian Abry 1 , Jérôme Aubin 2 , Christophe Savariaux 1 1 Institut de la Communication Parlée, Université Stendhal-INPG, UMR 5009 38040 Grenoble cedex 9, France 2 Département de linguistique et de didactique des langues, Université du Québec à Montréal Case postale 8888, succursale Centre-ville, Montréal (Québec) H3C 3P8, Canada Aude.Noiray@icp.inpg.fr, menard.lucie@uqam.ca Abstract. A test of the current labial anticipatory coarticulation models was performed on 8 adults (4 American English and 4 Canadian French). The subjects were audiovisually recorded, uttering [iC u] sequences. In order to describe their anticipatory movement patterns, protrusion as well as constriction temporal functions were obtained simultaneously via two motion captures: Optotrak for flesh-point protrusion, together with the ICP-Lip- Shape-Tracking-System, which can deliver area constriction in addition. A bite-block condition was added in order to control the possible consonantal influence (via the jaw, notably for coronals) on the time course of rounding anticipation. As for French subjects, for whom the Movement Expansion Model (MEM) was first elaborated (and retested recently for children), Canadian French display a lawful MEM anticipatory behaviour. Surprisingly, this is also the case for American English subjects who have been repeatedly reported not to behave in the line of French or Swedish rounding behaviour. We attribute the failure of the preceding American studies to a too small protrusion magnitude (rare in French but evidenced here for 3 of 4 Americans). The techniques they used could not capture the acoustically relevant lip area parameter, which is now clearly controlled, for French and for American rounding, within a MEM anticipatory pattern. n 1. Introduction In the literature on labial anticipatory coarticulation, three models based on protrusion are classically taken as reference to account for the temporal expansion of the labial rounding movement in adults. The Look-ahead Model elaborated by Henke (1966) for English, and tested for French by Benguerel and Cowan (1974), predicts that the rounding gesture would start at the acoustic offset of the unrounded vowel [i], with a movement expansion proportional to the consonantal interval between the two vowels. A diverging interpretation was proposed by Bell-Berti and Harris (1982), who claim that, for a defined speech style, the movement onset is time-locked to the acoustic onset of the rounded vowel whatever the number of intervocalic consonants (Time-locked Model). Finally, according to a Hybrid model (see Perkell and Matthies, 1992), the