|Title:||Against an ONSET Analsyis of Hiatus Resolution|
|Comment:||2003 LSA Paper, Document uses SIL IPA fonts|
|Abstract:||In Yatzachi Zapotec (YZ), spoken in southeastern Mexico, onsetless syllables are allowed initially, but not word internally. In derived vowel sequences, hiatus is repaired through coalescence or diphthongization. Interestingly, sequences of vowel-glottal stop-vowel behave much the same as VV sequences, as shown in (1). The similar patterning of Zapotec VV and V?V sequences suggests that they be subsumed under the same analysis.
(1) a. zetSa + o? --> zetSao? -diphthongization
b. tSSagna? + o? --> tSSagnao- -diphthongization
c. zetSa + a? --> zetSa? -coalescence
d. tSSagna? + a? --> tSSagnaa- -coalescence
This data is problematic for analyses claiming that hiatus resolution is motivated by the need for every syllable to have an onset. While an ONSET approach can handle the VV sequences, it cannot motivate repair of V?V, which does not violate ONSET.
I propose that hiatus resolution may occur independently of ONSET. The approach taken here is inspired by the framework of gestural phonology. In this framework, the phonological specifications of a segment are seen as instructions to the articulators regarding a specific target gesture associated with that segment. Speech involves the temporal coordination of the physical gestures comprising the utterance. Crucial is Gafos (2001)'s notion of the coordination of gestures, and the claim that this temporal coordination between gestures is constrained by gestural coordination constraints, which state the optimal coordination of gestural landmarks.
We can state the problem thus; faithful realization of input hiatus results in the gestures of two vowels being adjacent or overlapping. During the transition from one vowel to the next, gestures are made that do not reflect any underlying phonological specification; they are merely consequences of the physical system. During this period, acoustic cues are produced that the hearer must ignore to correctly perceive the speech stream; if he does not, he may incorrectly interpret the cues as being the result of an underlying gestural target. This situation is surely undesirable; by allowing gestures that are not part of the phonological specification of the word to be perceived, we increase ambiguity. I propose that hiatus resolution is the result of a gestural coordination constraint demanding that the gestural targets of every input sequence of vowels be aligned with a consonantal target such that the consonant overlaps the transitional gestures, henceforth VCV-COORD.
This approach handles the YZ data better than ONSET analyses because it not only applies to VV contexts, but can also extend to V?V contexts; glottal stop is not specified for an oral place target, and cannot satisfy VCV-COORD. Moreover, I show that this analysis also explains puzzling data from Yucatec Maya (YM) and Axeninca Campa (AC), which exhibit vowel spreading across glottals and position dependent repair of ONSET violations, respectively. In the case of YM, this data is easily explained; spreading results in a single gesture across the glottal closure, consequently no longer violating VCV-COORD. Positional differences in repair of ONSET violations are handled by teasing apart two motivations for repair; ONSET and VCV-COORD. When VCV-COORD outranks ONSET only word medial onsetless syllables will be repaired.
(For Abstract, S= voiceless alveo-palatal fricative, ?= glottal stop. For document SIL IPA fonts are used.)