Abstract
The increasing availability of magnetic resonance imaging (MRI) as a research, and even clinical, tool in speech production makes possible a wide range of quantitative methods in vocal tract measurement. In these initial stages of application, it is essential that the limits of the method be identified. The present investigation was designed to apply the techniques of digital image analysis and volumetric measurement to MRls obtained for the vocal tract during production of continuant speech sounds, and to apply these measures to a well-established and thoroughly tested model of acoustic transmission (Stevens & House, 1955). The results demonstrated that, although there were several sources of relatively large error and measurement bias, the vocal tract volumes obtained from MRIs were significantly predictive of vocal tract resonance frequencies. These results are discussed with respect to limits and potential for future application of MRI to speech production research.

This publication has 9 references indexed in Scilit: