Speech skull: From the bony structure of the head to soft tissues of
the vocal tract

Jean-Louis Heim*, Louis-Jean Boë**, Denis Autesserre**

*Musée de l'Homme, CNRS, Paris, France
** ICP Université Stendhal, INPG, CNRS, Grenoble, France

Email : heim@mnhn.fr

Within the frame of language origin debate*, this project aims at
defining possible speech prerequisites in the geometry, the musculature,
and the control of the vocal tract.  In order to answer these questions
one should be able (i) to describe the structural relationships between
landmarks of the skull and the jaw (the supporting structures of vocal
tract); (ii) to reconstruct the geometrical limits of the vocal tract
from these bony structures; (iii) to adapt an anthropomorphic model of
the vocal tract to a given skull; (iv) to generate the corresponding
maximal vowel space.  This work is the first step of a larger project
aimed at reconstructing the vocal tract from a cranium, modern or
fossilized, in particular Neandertal.  We present results derived from
xeroradiographic data of 45 subjects (males and females) uttering /i a/
vowels or in a resting position.  We used 19 landmarks corresponding to
skull, jaw, C2 to C5, hyoid bone and extremities of the vocal tract
(lips, pharyngeal wall and glottis).  The data were acquired with the
help of Craniomat, a software specially developed in collaboration with
radiologists, anthropologists and speech scientists.  To predict soft
tissues landmarks and avoid a simple fitting of the data observations,
we use the jackknife test.  We take away one skull from the learning set
for the remaining skulls, we learn a linear relation to predict the soft
tissue landmarks from bony landmarks, we calculate the error of
prediction on this skull, then we repeat learning and prediction for all
the skulls and evaluate the mean error.  We present the best predictors
and error estimation for training and prediction.  For lip protrusion,
hyoid height, and glottis height, errors of prediction are assessed in
relation with the range of observed data.  This first step permitted us
to show that it is possible to predict the extremities of the vocal
tract (lips, pharyngeal wall, glottis) given the bony structure.  In
order to corroborate and extend our method we present first results
concerning the predictions to other skulls going back in time: Egyptian
mummies (2,000 years B.P.), CroMagnon (20,000 years B.P.) and Neandertal
(45,000 years B.P.). 

----------------------------------------------------------------------------------

* This research is a part of Orofacial control in communication in human
and non human primates, EUROCORES program, being part of The Origin of
Man, Language and Languages (OMLL) project funded by European Community.