Comparative articulatory modelling of the tongue in speech and feeding
Résumé
Two of the major functions of the human vocal tract are speaking and feeding. Ontogenetically and phylogenetically feeding tasks precede speaking tasks and it has been hypothesised that speaking movements constitute a subset of feeding movements. From ElectroMagnetic Articulography data we have extracted two estimates of the degrees of freedom of the tongue, one from a speech task and one from a feeding task. These have been used to build two corresponding articulatory models. For both tasks, about 95% of the tongue variance can be explained by 4 parameters. Contrary to the accepted behaviour for speech, for feeding tasks, the tip, mid and back regions of the tongue appear to be controlled independently. The reconstruction errors of speech and feeding articulations by the two models show slightly better efficiency for the feeding model, suggesting that the speech movements might realistically be considered a subset of the feeding movements.