| This thesis aims at developing an analysis of the fact that obstruent voicing interacts with tone registers in the Xiangxiang Dialect (hereafter the XXD), a subdialect of the Xiang Dialects of Chinese, within the framework of Optimality Theory (OT).Up till now, the data of the XXD are all auditorily based and focus on the surface segments and tones. Neither phonetic nor phonological analysis has been done. This thesis is to describe the surface segment and tone inventory in the XXD based on the fieldwork done in 2002 and phonetic experiments conducted in 2003; to analyze the underlying representations of the segments and tones, the syllable structure and tone bearing unit in the XXD in the framework of Non-linear Phonology; and to give an OT analysis of the interaction between obstruent voicing and tone registers in the XXD.In the XXD, a word is typically a monosyllabic free morpheme. Obstruents occur only in the onset. Based on the fieldwork done in 2002 and phonetic experiments conducted in 2003, there are three series of stops — voiceless unaspirated, voiceless aspirated and voiced, and seven surface tones — four low register and three high register tones in the XXD. Voiced obstruents cooccur only with low register tones within a syllable.The main issues to be addressed with respect to the interaction between obstruent voicing and tone registers in this thesis include:(1) the phonetic basis of the interaction of obstruent voicing with tone registers;(2) an OT account of the interaction between obstruent voicing and tone registers. Two important findings emerge from this investigation.First, phonetic evidence from production and perception supports" the cross-linguistic preference for the cooccurrence of voiced obstruents with low register tones and voiceless obstruents with high register tones:Previous physiological research leads to the assumption that the cooccurrence of voiced obstruents with low tones and voiceless obstruents with high tones is grounded in articulatory factors: the articulatory gesture is the same for the presence of voicing of the obstruent andthe low pitch of the following vowel, as well as the absence of voicing of the obstruent and the high pitch of the following vowel — the cricothyroid muscle is relaxed and the larynx is in a low position for both voiced obstruents and low tone; the cricothyroid muscle is contracted and the larynx is in a high position for both voiceless obstruents and high tone. This thesis proposes that the cooccurrence of the segmental feature [+voiced] with the tonal feature L, and [-voiced] with H is also due to mutual enhancement: on one hand, a voiced/voiceless obstruent can enhance the perceptual distinctiveness of a low/high tone, i.e. a voiced/voiceless obstruent makes a following low/high tone perceptually more low/high; on the other, a low/high tone can enhance the perceptual distinctiveness of a voiced/voiceless obstruent, i.e. a voiced/voiceless obstruent is more fully realized preceding a low/high tone. Second, in some languages the marked structures voiced/high or voiceless/low are allowed: the former in the Yuanlingxiang Dialect and the latter in the Rucheng Dialect (both are Chinese dialects and are to be presented in this thesis). And the strategies to avoid these marked configurations are different from language to language: changing the tonal specification of the underlying tone (in the XXD) or changing the voicing specification of the underlying obstruent (in the Yabem to be presented).The phonetic evidence suggests that a full account of the phenomenon of obstruent-tone interaction must refer to the articulatory and perceptual factors, which pertain to the phonetics. The idea that phonological markedness has phonetic roots has been expressed in Grounded Phonology by Archangeli & Pulleyblank, which is within the rule-based framework, where one must provide the theory with multiple fixes, all of which address the same phonetic difficulty. If a language allows some marked structure, the characteristic way of explaining it in the rule-based phonology is to set up an intermediate level where this structure is unmarked. This leads to a sharp increase in the abstractness of representation and rule interaction. In OT, the problem (what is phonetically difficult) is separated from the solution (how to fix it): the former is encoded in markedness constraints; the latter is, by relative ranking of the markedness and faithfulness constraints, deciding how and whether a marked structure is avoided. As a result, the idea that phonology is phonetically-based can be expressed directly in OT: a markedness constraint is grounded in phonetics and can embody a particular form of phonetic difficulty, with the issue of how and whether the difficulty is avoided relegated toother parts of the grammar — constraint ranking.Therefore, the central components of the OT analysis developed in this thesis are:(1) The proposal of two markedness constraints ([+vd], L)o and ([-vd, -sg], H])o stating the preference for a voiced (voiceless unaspirated) obstruent to cooccur with a low (high) register tone within a syllable, based on a typological survey and evidence in articulation and perception;(2) The relative ranking of these two grounding constraints with respect to faithfulness constraints and other markedness constraints by which the surface distribution of consonants and tone registers in the XXD can be accounted for;(3) The relative rankings of these two grounding constraints with respect to faithfulness constraints and other markedness constraints by which the surface distribution of consonants and tone registers in the RD and the YLXD, as well as the strategy of changing the voicing specification of the underlying obstruent employed by the Yabem, which goes in the opposite direction to that employed by the XXD, can be accounted for respectively.Theoretically, the significance of the present study is that it shows that the relation between phonology and phonetics can be better handled in the framework of OT.
|
PDF Full Text Request