東亞漢字音之入聲韻變化: 以優選理論探討

Abstract

本篇論文主旨在於以優選理論探討東亞漢字音的入聲韻變化。研究重點有三: (一) 東亞漢字音歷史音韻變化與現代音韻理論的相容性; (二) 東亞漢字音歷史音韻變化與音韻晦澀性 (Opacity); (三) 音節結構差異對東亞漢字音音韻變化的影響。研究資料乃以《方言調查字表》中所列的入聲韻為依據。本研究採用三個層次相近的東亞漢字音 (臺灣閩南語之文讀音、日本漢字音之漢音與韓國漢字音)。而研究方法則以母音種類為主，而非以傳統的攝，並分成 [a]母音、非[a]母音與合口。 首先，在歷史音韻與現代音韻理論的相容性方面，基於過去漢字音歷史音韻研究較少與現代音韻理論結合，因此本文以優選理論，特別以Sanders (2003) 的擴散優選理論 (Faithfulness, Markedness, Dispersion-Optimality Theory) 為主，來探討東亞漢字音入聲韻之歷史音韻變化。結果顯示在入聲韻中，[a]母音、非[a]母音與合口皆能與優選理論相容。而Sanders (2003) 的擴散優選理論更能清楚地將母音變化以擴散限制 (Dispersion Constraints) 表現，不同的限制代表了不同的母音變化。同時本研究亦提出央元音限制以補足先前研究之不足。 本文更進一步探討歷史音韻變化與音韻晦澀性。本文認為歷史音韻變化有一特色，即是連續性(Continuity)。但優選理論強調輸入項與輸出項的直接對應，使得優選理論無法解釋晦澀性。因此本文延伸McCarthy (2007) 的連續候選項 (Candidate Chains) 並納入連續性 (Continuity) 概念，以解釋歷史音韻變化中出現的晦澀性。 最後本文亦探討音節結構差異對東亞漢字音音韻變化的影響，可從二方面來探討: 莫拉 (mora) 於音節內的位置 (Phonotactic Constraints) 與音段交流限制 (Licensing Constraints)。當音節內的莫拉位於同一音節分支時，位於下位的音段則不會有交流現象 (如: 臺灣閩南語)，但當莫拉分屬於不同分支時，則會產生交流現象。其中依據音節中的莫拉是否顯著，再加以分類。若是莫拉顯著者，則音段交流時，莫拉傾向被保留 (如: 日本漢字音)。若是莫拉不顯著者，則音段交流時，莫拉傾向不被保留 (如: 韓國漢字音)。

This thesis investigates the phonological change of entering tone in Sino-Xenic languages based on Optimality Theory. The research issues include (1) How is Optimality Theory applied to historical linguistics on the basis of Sino-Xenic languages? (2) How is Opacity in historical linguistics represented in Optimality Theory by means of Sino-Xenic languages? (3) How does internal syllable structure affect phonological change in Sino-Xenic languages? In this study, the data of entering tone are collected according to Fāngyán Diàochá Zìbiăo (A wordlist for dialectal research). Specifically, the layers, which are pertinent to each other, from three Sino-Xenic languages, Literal Reading (Wéndú) in Southern Min and Kan-on in Sino-Japanese and Modern Sino-Korean, are compared. In addition, instead of the traditional method, Shè, this study concentrates on vowel changes by dividing vowels into [a]-vowel group, non-[a]-vowel group and Hékŏu. First, to investigate the compatibility of historical phonology and modern phonological theories, this study adopts Faithfulness, Markedness, Dispersion-Optimality Theory (FMD-OT) (Sanders 2003), paying specific attention to vowel changes. The results show that Optimality Theory can account for the historical change of entering tone in Sino-Xenic languages and FMD-OT succinctly explains vowel changes in terms of dispersion constraints. Furthermore, this study amends Sanders’ FMD-OT by proposing a new set of constraints for central vowel. Second, this study also discusses opacity in historical phonology. To solve this problem, this study modifies and extends Candidate Chains (McCarthy 2007). It is proposed historical sound change is continual, each stage is evaluated parallelly, and new constraints are activated and reranked with previous constraints. This proposal is supported by Gĕngshè (梗攝) in Southern Min, Division III of Shēngshè (深攝) and Xiánshè (咸攝) in Sino-Japanese and Gĕngshè (梗攝) in Sino-Korean. This study finally explores the influence of syllable structure on sound changes in Sino-Xenic languages. Two factors are suggested, Phonotactic Constraints and Licensing Constraints. When the moras belong to the same branch in syllable node, there is no interaction of the segments, such as Southern Min. On the contrary, when the moras locate in different branches, the segments interact, such as Sino-Japanese and Sino-Korean. What distinguishes Sino-Japanese from Sino-Korean is whether mora is prominent or not. On one hand, if the mora is prominent, the mora is maintained, for example, Sino-Japanese. On the other hand, if the mora is not prominent, the mora is likely to be deleted, i.e. Sino-Korean, when the interaction takes place.