Catherine L. Rogers
Robert F. Port
Dept. of Linguistics, Indiana Univ., Bloomington, IN 47405
This study sought to explore previous claims made about the interaction of several linguistic timing factors, and subsequently, to develop rules for the duration of vowels in several metrical environments. Two-word phrases that resembled personal names, like Tuck Tuckhouse, Tackery leTack and Ticky Tick, were constructed. The vowels [i, (small capital eye), (ae ligature), (inverted vee)] were used as the stressed vowel in the word types Tuck, leTuck, deleTuck, Tucky, Tuckery, and Tuckhouse. For each vowel, the six word types were fully crossed, creating a list of 36 phrases. Three speakers produced three repetitions each. Measurements were made of the major segments in the stressed syllable of each word (t-voice-onset time, the vowel, and k closure), as well as of the word and phrase as a whole. The incompressibility property was observed for the interaction of word identity by position-in-phrase [cf. D. H. Klatt, J. Acoust. Soc. Am. 59, 1208--1221 (1976)], while the constant ratio hypothesis was found for the vowel by position-in-phrase interaction [partly replicating R. F. Port, J. Acoust. Soc. Am. 69, 262--274 (1981)]. A formal timing model, suitable for use in speech-synthesis-by-rule, is proposed to model these results.