Phonetics and phonology of checked phonation, syllables, and tones
The term “checked” has been used to describe constituents whose voicing ends abruptly, often with glottalization. However, there is still no phonological definition of “checked.” This dissertation aims to define what “checked” means phonologically, and to describe the phonetic nature of checked constituents in production and perception.
The first part of the dissertation focuses on what it means to be “checked” in two language families where the term is frequently invoked: Zapotec and Chinese. Surveying the checked vowels in 20 Zapotec languages, I argue that, in Zapotec, “checked” is a phonation type, represented as a late-phased [+constricted glottis] feature on vowels. Checked phonation notably contrasts with rearticulated phonation with earlier glottalization. Moreover, checked phonation in most Zapotec languages contrasts with unchecked phonations independently of tone.
Surveying the checked syllables in 95 Chinese languages, I argue that, in Chinese, “checked” is a descriptor of the phonotactic constraint between stop-closed syllables and tones. Checked syllables are those closed by stop codas and bearing different tones (i.e., checked tones) from open and sonorant-closed syllables. The checked syllables/tones in Modern Chinese have developed from the *Rù syllables/tones in Middle Chinese.
The second part of the dissertation explores the acoustics of checked syllables and tones in Xiapu Min, a Chinese language. I find that, in the citation forms, checked syllables/tones in Xiapu Min have different F0, shorter duration, and more glottalized quality compared with unchecked tones. In sandhi forms where checked tones neutralize with unchecked tones, checked tones nonetheless remain shorter than unchecked tones.
In a series of experiments, I describe the cues the listeners use to identify checked syllables and tones in Xiapu Min. I show that F0, duration, and voice quality each has an independent effect on eliciting checked tone responses, and that listeners can distinguish checked tones from unchecked tones in sandhi forms by the shorter duration of checked tones.
This dissertation provides criteria for future research to determine whether a language has checked constituents. Also, it contributes novel data from an under-documented language to the phonetics of checked syllables and tones.