부산대학교 도서관

Previous reconstructions of fossil human vocal tract (VT) anatomy have primarily been based on single skeletal indicators such as basicranial flexion of hyoid morphology. These studies have produced conflicting results. A few reconstructions have used combinations of indicators to predict VT morphology, but these have employed relatively subjective methods. In an attempt to better understand fossil human VT anatomy, we developed a new predictive technique that uses relationships between VT landmarks and associated skeletal landmarks of living humans as the basis for reconstruction. We believe that the use of these landmarks is more likely to produce reliable results than any single indicator, and that our method of analyzing relationships between skeletal and soft-tissue anatomy is less subjective than previous techniques. Using the software package C2000cepha v.2.1.B, we collected two large sets of landmark data from a sample of human clinical CT scans provided by the Clinique Pasteur in Toulouse, France. The first consisted of skeletal landmarks located on the basicranium, vertebral column, dentition, mandible, nasopharynx, nasal cavity, and nasal aperture. The second consisted of soft-tissue landmarks located along the VT. Through the combined use of generalized procrustes analysis, principal component analysis, and multiple regression, we derived multiple formulae that allow us to predict the position of soft-tissue VT landmarks based on associated skeletal landmarks. Preliminary work indicates that this technique facilitates a working 3-D approximation of the supralaryngeal VT from skeletal landmarks, and that it holds promise for the reconstruction of VT soft-tissue anatomy in Neandertal and other fossil specimens.