Anchoring consonantal perturbation of fundamental frequency

The larynx is raised during voiceless obstruents, elevating fundamental frequency (F0) on the following vowel perceptibly above its Svalue after voiced obstruents or nasals. Though vertical larynx movement is not closely synchronized with vowel onset [C. J. Riordan, Report of the Phonology Laboratory, UC, Berkeley 5, 114–126 (1980)], it does not follow a uniform contour independent of consonant type either, but is loosely anchored to oral release. I measured F0 of vowels following glottal stop, voiced and voiceless (oral) stops, ejectives, and nasals produced by a Tigrinya speaker. Glottal stop, voiceless stops, and ejectives all elevated F0, but voiced stops and nasals did not. F0 fell 6–7 Hz during the first 50 ms after all elevating consonants and rose 2–3 Hz after nonelevators. Taking nasal values as a baseline, F0 at vowel onset was 16 Hz higher after glottal stop, 13 Hz higher after voiceless stops, but only 8 Hz higher after ejectives. F0 should be higher after ejectives than voiceless stops because the larynx is actively raised during ejectives to compress air in the oral cavity. However, voice onset time (VOT) for ejectives is 1.5 times that of oral stops, so the larnyx may have fallen back before voicing begins. The F0 contour after an ejective fits that following a voiceless stop if it is offset by the difference in their VOTs.