Consonant-tone interaction in Thai

This dissertation establishes that phonological consonant-tone interaction occurs in a non-local configuration in Thai, in contrast to current theories where this interaction is only thought to occur locally. A phonological account posits that in bimoraic unchecked syllables, tones associated with the second mora, and not the first mora, interact with onset consonants in Thai. This is implemented via an Optimality-theoretic markedness constraint, *[+CG]-[H]μ2. Evidence for this position comes from a quantitative lexical gap study and a pair of judgment experiments, which show that both rising tone and high tone are ungrammatical in unchecked syllables with laryngealized obstruent onsets. In Thai, these tones share a common high-tone target at the end of the syllable. These facts suggest the presence of a phonological constraint violated by onset-tone sequences, where the tone on the second mora is referenced, despite the fact that it is the more distant of the two. This constraint is grounded since it is phonologically less marked for tones to be realized late in syllables – the second mora is treated as the “head” mora of the syllable in Thai. This phonological analysis is supported via evidence from an acoustic study of voiced and unaspirated obstruent onsets in Thai. It is discovered that these obstruents are articulated with laryngeal constriction, and that they form a natural class under the feature [+constricted glottis] A second important contribution of the acoustic study was that [ʔ] onsets are phonetically distinct (they raise F0 and spectral tilt), even though they are part of the same phonological [+constricted glottis] class. Two judgment experiments confirm the psychological reality of the constraint *[+CG]-[H]μ2 among native Thai words. However, three of four onset-tone sequences that violate this constraint are considered grammatical under interpretation as English loans. This result is consistent with findings in other languages, where loan strata are more permissive than native strata. In addition, participants exhibited preferences for [+constricted glottis] onsets and low tone that cannot be explained via language experience. Therefore, it is argued that the relative ranking of markedness constraints can distinguish between grammatical forms, a finding that replicates previous experiments in English and Hebrew.Ph. D.Includes bibliographical referencesby Jeremy M Perkin