Vowel context, rate, and loudness effects on linguapalatal contact patterns in Hindi retroflex /t/

Abstract

Tongue position for retroflex /t/ has been shown to differ across languages in the contexts of /a/ and certain other vowels. Degree of retroflexion has also been shown to be affected by vowel context [Ladefoged and Bhaskararao, J. Phonet. 11, 291–302 (1983) and references therein]. Since previous studies have not determined whether the place or length of constriction differs as a function of vowel context, this study examined changes in the articulation of Hindi retroflex /t/ as a function of vowel context, and also as a function of speaking rate and loudness. A 96-channel electropalatograph was used to monitor production of retroflex /t/ in the nonsense words /bitíb/, /batáb/, and /butúb/ spoken in carrier phrase by a native speaker of Hindi. The length of constriction (i.e., the number of contacted rows × 2 mm) changed very little (< 1 mm) as a function of the three factors. However, the anterior-posterior location of the /t/ constriction shifted dramatically as a function of vowel context. The center of constriction was 3.8 mm more posterior in /u/ than in /i/, and 5.8 mm more posterior in /a/ than in /i/. The pattern and area of tongue contact observed for retroflex /t/ and dental /t/ were virtually identical in the /i/ but not in the /a/ or /u/ context. A similar vowel context effect on retroflex /t/ was observed in speech produced at a fast rate, and in loud speech. [Work supported by NIH grant.]