The American English Flapping Rule and the Effect of Stress on Stop Consonant Durations

The experiment described in this paper concerns the American English Flapping Rule, whereby non-ward-initial iintervocalic  /t/ and /d/  are flapped when preceding an unstressed vowel, as in the words 'metal' and 'pyramidal'; in contrast the /t/ preceding a stressed vowel in 'metallic' is not flapped. Acoustically, a major difference between flapped and non-flapped alveolars is durational: flaps are considerably shorter than their non-flapped counterparts. In this study we measured the durations of intervocalic stops in both the non-flapping and flapping environments as well as the durations of vowels that preceded the alveolars in order to answer the following questions:

1. Does a general timing mechanism underly the rule? It may be the case that the extreme shortening found in /t/ and /d/ is an instantiation of a more general timing principle. We discuss possible mechanisms far the length differences between poststress and prestress stops as well as implications for the formulation of the American English flapping rule.

2. Are vowels preceding flaps that were underlying /d/'s longer than vowels preceding flaps that were underlying /t/'s?  It is well-known that vowels preceding voiced stops tend to be longer than vowels preceding voiceless stops (e.g. Chen, 1970). Longer vowels before flapped /d/'s than before flapped /t/'s would suggest that lengthening before voiced segments is due to the underlying phonological distinction ([+voice] vs. [-voice]) rather than to the phonetic realization of the consonant (which, in the case of most flaps, is as a voiced segment). In fact, one study (Fox and Terbeek, 1977) found that vowels preceding flapped /d/'s were longer than vowels preceeding flapped /t/'s.

3. Does a longer phonological phrase affect segment durations?