PERCEPTUAL CATEGORIZATION OF SYNTHESIZED ENGLISH VOWELS FROM BIRTH TO ADULTHOOD

Lucie Ménard1, Barbara L. Davis2 & Louis-Jean Boë3
1Laboratoire de phonétique, Université du Québec à Montréal; 2Department of Communication Sciences and Disorders, The University of Texas at Austin; 3Institut de la Communication Parlée

ID 1290
[full paper]

The experiment reported in this paper is aimed at determining the influence of non uniform vocal tract growth on the ability to reach acoustic-perceptual targets in English. An articulatory-to-acoustic model integrating non uniform vocal tract growth has been exploited to synthesize 342 5-formant vowels, covering maximal vowel spaces produced by speakers at 5 growth stages: newborn, 4 years old, 10 years old, 16 years old, and 21 years old (adult stage). 37 American English speakers served as subjects in a perceptual categorization experiment. Results show that the three cardinal vowels /i u a/ can be produced by speakers with a newborn-like vocal tract. It is suggested that articulatory-to-acoustic relationships for a given vowel may differ across growth stages.