In this study, a speaker-adaptive articulatory model was constructed by fitting point-wise articulatory positions measured by Electromagnetic Articulography (EMA) to the framework of Childerfs vocal tract model (2000) to customize the standard vocal tract model with speaker-dependent articulatory features. With the speaker-adaptive articulatory model, the area functions of oral and nasal vowel pairs (/a/, /i/, /u/) were simulated. The differences of area functions between oral and nasal vowels were decomposed into orthogonal modes (Story et al., 1998) to account for the primary articulatory changes related to nasalization. The relationship between the principal articulatory modes and the lowest two formant frequencies (F1, F2) was examined to indicate the effect of oropharyngeal articulatory alterations on the acoustics of nasal vowels.

Index Terms: articulatory modeling, nasalization, electromagnetic articulography