14th International Congress of Phonetic Sciences (ICPhS-14)
San Francisco, CA, USA
For American English /r/, a large number of articulatory configurations produce essentially equivalent acoustic profiles of F1, F2 and F3. In previous work, recently acquired physiological data from Magnetic Resonance Imaging studies were used to develop a simple tube model for /r/ that gives reasonable estimates for the first four formants. All the /r/ configurations showed bunching of the tongue dorsum. Using articulatory data from several sources, we show in this paper that the simple tube model also gives reasonable estimates of F4 and F5 which can vary substantially depending upon the tongue shape. Our acoustic data show that the traditional retroflex /r/ (with the tongue tip raised and the tongue dorsum lowered) has an acoustic profile where F3 and F5 track each other. In this case, there is also some lowering in F4. However, when the tongue dorsum is raised (regardless of whether the tongue tip is raised or lowered), there is no such lowering of F4 or F5. With reasonable estimates of vocal tract dimensions, we show that the model can explain both types of formant patterns.
Bibliographic reference. Espy-Wilson, Carol / Boyce, Suzanne (1999): "A simple tube model for American English /r/", In ICPhS-14, 2137-2140.