Dental flaps, vowel duration and rule ordering in American English

Abstract: This paper presents the results of an experiment which investigated dental flaps and their effect upon preceding vowel duration in American English. Twenty speakers read a randomized list of 40 disyllabic words, 20 with medial /t/ and 20 with medial /d/. Most tokens of /t/ and /d/ were pronounced as flaps, as expected. The data, collected from spectrograms, included 1st and 2nd vowel duration, presence or absence of flap (as opposed to stop [t] or [d]), duration of consonant, and extent of voicing during the flap. Our results provide evidence bearing on two phonological questions: (1) Does the rule which changes /t/ and/d/ to flaps (F) precede or follow the rule which lengthens vowels before voiced consonants (VL)? If VL precedes F, the vowels preceding flaps with underlying /d/ will have greater duration than vowels preceding flaps with underlying /t/. If F precedes VL, there will be no such difference. Our results indicate a small, but significant durational difference. (2) Do underlying /t/ and /d/, in fact, give rise to the same voiced flap consonant, as is usually assumed? The data suggest that not all /t/-flaps are identical, nor are they necessarily voiced.