Emergence of articulatory-acoustic systems from deictic interaction games in a "vocalize to localize" framework - CNRS - Centre national de la recherche scientifique Accéder directement au contenu
Chapitre D'ouvrage Année : 2011

Emergence of articulatory-acoustic systems from deictic interaction games in a "vocalize to localize" framework

Résumé

Since the 70's and Lindblom's proposal to "derive language from non-language", phoneticians have developed a number of "substance-based" theories. The starting point is Lindblom's Dispersion Theory and Stevens's Quantal Theory, which open the way to a rich tradition of works attempting to determine and possibly model how phonological systems could be shaped by the perceptuo-motor substance of speech communication. These works search to derive the shapes of human languages from constraints arising from perceptual (auditory and perhaps visual) and motor (articulatory and cognitive) properties of the speech communication system: we call them "Morphogenesis Theories". More recently, a number of proposals were introduced in order to connect pre-linguistic primate abilities (such as vocalization, gestures, mastication or deixis) to human language. For instance, in the "Vocalize-to-Localize" framework that we adopt in the present work (Abry & al., 2004), human language is supposed to derive from a precursor deictic function, considering that language could have provided at the beginning an evolutionary development of the ability to "show with the voice". We call this type of theories "Origins Theories". We propose that the principles of Morphogenesis Theories (such as dispersion principles or the quantal nature of speech) can be incorporated and to a certain extent derived from Origins Theories. While Morphogenesis Theories raise questions such as "why are vowel systems shaped the way they are?" and answer that it is to increase auditory dispersion in order to prevent confusion between them, we ask questions such as "why do humans attempt to prevent confusion between percepts?" and answer that it could be to "show with the voice", that is, to improve the pre-linguistic deictic function. In this paper, we present a computational Bayesian model incorporating the Dispersion and Quantal Theories of speech sounds inside the Vocalize-to-Localize framework, and show how realistic simulations of vowel systems can emerge from this model.

Domaines

Informatique
Fichier principal
Vignette du fichier
moulin-frier11.pdf (505.07 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-00961125 , version 1 (19-03-2014)

Identifiants

  • HAL Id : hal-00961125 , version 1

Citer

Clément Moulin-Frier, Jean-Luc Schwartz, Julien Diard, Pierre Bessière. Emergence of articulatory-acoustic systems from deictic interaction games in a "vocalize to localize" framework. Anne Vilain and Jean-Luc Schwartz and Christian Abry and Jacques Vauclair. Primate communication and human language: Vocalisations, gestures, imitation and deixis in humans and non-humans, John Benjamins Pub. Co., pp.193-220, 2011, Advances in Interaction Studies. ⟨hal-00961125⟩
473 Consultations
204 Téléchargements

Partager

Gmail Facebook X LinkedIn More