Modeling lingual coarticulation in coronal stops
Master Thesis in Phonetics, Spring 1999
This study presents observations from lateral X-ray recordings of one speaker. The speech material consists of VCV-sequences occurring in meaningful Swedish (C)V1CV2 words where C = dental or retroflex voiced stop. The words were pronounced with the grave accent implying a main stress on the first syllable and an unstressed final syllable.
The articulatory observations were in the form of profile tracings and were quantified using procedures adopted in several previous investigations. In order to get some insight into the articulatory principles underlying coarticulation , an attempt was also made to model the observed data quantitatively. The data were first applied to Öhman's (1967) model. Then they were analyzed in terms of the parameters of the APEX articulatory model (Lindblom & Sundberg, 1971, Stark et al. 1996). Both of these models showed shortcomings, which led to the development of a new framework where the target articulation for the dental and retroflex consonants was taken to be the consonant produced in an «-context. More data are certainly required before any broader conclusions can be drawn from the results. However, good accuracy was obtained with the 3-parameter model. These results can be attributed to the choice of a target articulation for the consonant which assumes a contour with minimal vowel activity. The present findings were compared with those of Houde (1967), McAllister & Engstrand (1992) and Engstrand et al. (1999) and certain parallels were noted, e.g. Houde's 'perturbation movement' and McAllister & Engstrand's 'trough'. Taken together, all of these results encourage us to pursue the present approach further in future analyses of the X-ray database.