The Dispersion-Focalization Theory (DFT) has allowed us to predict the most frequent n-vowel systems (3 < n < 9) in a 3D acoustico-perceptual space. The present work provides a new step: the basic idea is to integrate an articulatory model in the prediction by the DFT. Results for n = 3 to 5 confirm the efficiency of the approach and consolidate the concept of focalization necessary to predict systems with a prototypical, velo-palatal [u], instead of an atypical velo-pharyngeal one.