在有關漢語聲調與英語輕重音的借字研究中，英語重音與漢語聲調的對應並不一致，因此，本篇研究分別從語料庫與優選理論的觀點，重新檢驗漢語借字中漢語聲調與英語輕重音的關係。觀察語料庫發現英語重音傾向與包含高音 (high pitch) 的聲調對應，因此本文提ANCHOR-L(H,σ) 和ANCHOR-R(H, σ) 制約，分別要求英語重音節的H音高要固定在對應音節聲調的左端或右端。在*[+voiced]/[+H]- 和 *[-voiced]/[-H]- 的制約下，輸出值呈現voice enhancement的效果。換句話說，*[+voiced]/[+H]- 制約要求含有聲聲母之英語重音節與低起始音域的聲調對應；*[-voiced]/[-H]- 制約要求含無聲聲母之英語重音節與高起始音域的聲調對應。含有聲聲母之英語重音節可能對應到高平調或中升調，制約層級的改變能預測之。若ANCHOR-L(H,σ)與ANCHOR-R(H,σ)支配*[+voiced]/[+H]- ，則高平調被選出，但是，若 *[+voiced]/[+H]- 支配ANCHOR-L(H,σ) 與 ANCHOR-R(H,σ)，則中升調被選出。在RELCORR制約的要求下，對應重音節聲調的左端調值不低於對應輕音節聲調的左端調值。 / The studies of Mandarin loanwords adapted from English have suggested different tonal preferences in correspondence to English stress. This thesis reexamines the relationship between stress and tone in the English-to-Mandarin loanwords based on a corpus and the Optimality Theory. Two anchoring constraints, ANCHOR-L(H,σ) and ANCHOR-R(H,σ) are proposed to require the realization of the pitch accent H of the English stressed syllables, indicating the fact that English stress prefers a tone with a high pitch. The *[+voiced]/[+H]- and *[-voiced]/[-H]- constraints show a pattern of voice enhancement. *[+voiced]/[+H]- requires an English stressed syllable with a voiced onset to be adapted with an initially low-registered tone; *[-voiced]/[-H]- requires an English stressed syllable with a voiceless onset to be adapted with an initially high-registered tone. In addition, constraint re-ranking predicts the alternatively choices between the H-tone and the MH-tone for an English stressed syllable with a voiced onset. The H-tone is selected if ANCHOR-L(H,σ)and ANCHOR-R(H,σ) outrank *[+voiced]/[+H]-; whereas the MH-tone is chosen if *[+voiced]/[+H]- dominates these two anchoring constraints. The RELCORR constraint requires a left-edge tonal correspondence, where the left-edge pitch of a tone that corresponds to an English stressed syllable is not lower than one that corresponds to an English unstressed syllable.

本文針對甲骨文的文例，運用比較同文例的方法，主要由二方面進行研究：一為通讀一版甲骨所需的基本文例之整理；一為甲骨文書寫法所引起的契刻現象之整理與對特殊現象的討論。 本文以第三章所整理的基本文例為基礎，通過第四章、第五章和第六章，三章的討論，指出甲骨文的特殊契刻例十二種，即誤刻、缺刻、奪文、衍文、補刻、倒文、側書、顛亂、省文、析書、合文及借字例，同時校勘甲骨文本身的刻辭錯誤，並糾正對此學者誤釋之例。 第一章 緒論 第一節 研究動機與目的 第二節 研究方法與範圍 第三節 文獻檢討 第二章 甲骨文形體結構特徵 第一節 字無定格，部件位置不固定 第二節 一字正反書無別 第三節 一字異形，繁簡並存 第四節 形近或義近部件通用 第五節 書寫工具、習慣與字形之關係 第三章 甲骨版面通讀法研究 第一節 刻辭行款類型 壹、一般行款形式 貳、特殊行款形式 第二節 甲骨文內容間的聯繫 壹、卜辭的結構 貳、卜問形式 參、正面卜辭與背面卜辭相承 肆、獸骨相間刻辭 伍、獸骨邊面對應 第三節 一事多卜 壹、同文卜辭 貳、成套卜辭 參、同對卜辭 第四章 甲骨文契刻特殊例（上） 第一節 誤刻例 壹、誤增或誤穿筆劃 貳、受它字的影響而誤刻 參、形近或義近而誤 肆、音近而誤 伍、干支之誤 第二節 缺刻例 壹、個字漏刻 貳、整條卜辭之缺刻 參、缺刻而致誤字、誤釋之例 肆、缺刻字比較表 第三節 奪文例 壹、卜辭所見奪文例及說明 貳、奪文之形態 參、奪文之內容 肆、時代別情況 第四節 衍文例 壹、卜辭所見之衍文例 貳、衍文的內容及時代別特徵 第五章 甲骨文契刻特殊例（中） 第一節 補刻例 壹、空間不足，刪字而重刻 貳、適當空間補刻 參、甲骨文補刻原因與補法特點 第二節 倒文例 壹、文字結構之倒 貳、倒語 參、倒行 第三節 側書例 壹、文字側書 貳、數字側書 第四節 顛亂例 壹、補刻而顛亂 貳、部分內容刻於它處而顛亂 參、行款方向轉變而顛亂 肆、不遵守辭序而顛亂 第六章 甲骨文契刻特殊例（下） 第一節 省文例 壹、前辭內容之省略 貳、干支之省略 參、身分、人名、稱謂等之省略 第二節 析書例 壹、析書的主要偏旁 貳、析書者的主要類別與形式 參、析書而致誤者 第三節 合文例 壹、合文類別 貳、合文結合方式 參、甲骨文合文與金文、戰國文字合文異同 第四節 借字例 壹、一字兩用 貳、重文符號的運用 第七章 結論 參考書目 引用著錄簡稱表 附錄：特殊契刻例表 / In order to develop an understanding of lines of characters in oracle bone inscriptions, using a comparison of similar lines of characters, this thesis has two major aims. The first is to develop an understanding of the basic understanding of the lines of characters on an individual bone. The second is to organize and study the phenomena of the transition between the initial written characters and their subsequent carved versions. Chapter Three provides a fundamental organization of the basic lines. Then, following the discussions of Chapters Four Five and Six, there are presented twelve characteristics to analyze the carved characters: Wu ke(wrong words), Que ke(missing carving), Duo wen(incomplete words), Yan wen(repeated words), Bu ke(supplementary carving), Dao wen(upside down characters), Ce shu(sideways characters), Dian luan(confused), Sheng wen(omitted characters), Xishu(divided characters), He wen(combined characters), Jie zi(loan words). From this overall analysis, it is possible to correct mistaken carvings and mistaken readings of previous scholars.

本篇論文從語料庫及優選理論，分析台語中的日語借字，探討兩個語言之間的音韻調整現象。 本研究建立一個日語借字語料庫，以數據資料佐證音段替代的趨向，釐清音段替代的主要性和次要性。統計發現，單一日語音段，雖然可能被不同的台語音段取代，但由於比例差距的懸殊，日語和台語音段的對應情形，大多可視為一對一的關係。至於一對二的對應情形，則可歸因於台語音韻限制，或是對於借字和來源字之間音徵保留的要求。 基於語料庫的數據，本篇從優選理論分析日語和台語間的音韻調整現象。音段替代方面，主要歸因於IDENT[feature]，MAX，以及結構性制約(markedness constraint)之間的排序部署。另外，文中提出四個並存音韻，藉由調整MAX和IDENT[MR]之間，以及SHARE[back]和IDENT[anterior]之間的排序，解釋日語借字中的自由變異(free variation)現象。最後，受到IDENT[μ]和*[μμμ]的控制，相鄰兩個音節不會合併為一個，如果合併後發生音拍(mora)數量刪減或增加，或是產生一個含有二個音拍以上的音節。 / This thesis investigates the phonological adjustment of Japanese loanwords in Taiwanese. A corpus is established, and the statistics are provided to observe the substitute tendency of each of the Japanese sources. The use of the statistics properly explains whether a corresponding Taiwanese segment is either a regular substitute or merely an exception. Most of the Japanese segments are in one-to-one mapping when they are borrowed into Taiwanese. Some Japanese segments are in one-to-two mapping, due to either Taiwanese phonotatics or feature preservation between the sources and the loans, such as the phenomenon of reconfiguration. Besides, two adjacent syllables will combine to one, if the vowels are syllabified as a VG sequence after combination; in that event, the second syllable does not end with a nasal coda. Moreover, a glottal stop will insert to a word-final position if the penultimate syllable in a loanword ends with a vowel or a non-sonorant consonant. This analysis, based on the corpus, is conducted under the framework of Optimality Theory (Prince and Smolensky 1993/2004). For segmental substitutions, a set of IDENT[feature] is adopted to require the identity of certain features. The markedness constraints, regarding the segmental and place markedness and the Taiwanese phonotatics, are employed. High ranking of markedness constraints are responsible for adjustment of segments. Besides, re-rankings of SHARE[back] and IDENT[anterior] and MAX and IDENT[MR], are suggested. They are responsible for the substitutes that are in free variation. As for the contraction of syllables, it allows the loans and the sources to have the same numbers of moras. However, contraction is suspended, to prevent a syllable to have three or more moras.