X标杆理论视阈下的汉语音节结构研究
|
摘要
本文以生成音系学有关音节结构表征的X标杆理论为基础，参考van deWeijer&Zhang (2008)提出的“多指示语的X标杆模型”，综合运用其他相关理论，提出了表征汉语音节结构的“广义多指示语的X标杆模型”，然后在对四种主要汉语方言进行深入分析研究的基础上证明，该模型适用于表征主要汉语方言的音节结构。
音系学中有关音节结构表征的X标杆理论由J.列文（1983,1985）提出，该理论的创新之处在于借用句法学中的X标杆理论对音节结构进行抽象化表征。Van de Weijer&Zhang（2008）通过研究汉语标准语（即普通话）音节中的介音归属，参考乔姆斯基等(1995)提出的关于句子结构的“多指示语的X标杆模型”，对音系学的X标杆理论进行修改，提出了针对汉语普通话音节结构的“多指示语的X标杆模型”（简称“多指示语模型”）。该模型认为音节是核心元音V的最大投射Nmax，介音是中间投射N”的左向分支。本文借鉴生成音系学中的特征几何理论、特征标记性理论、不充分赋值理论以及莫拉理论等对“多指示语的X标杆模型”进行改进，提出了“广义多指示语的X标杆模型”（简称“广义模型”），
“广义模型”与“多指示语模型”的不同之处在于：1、“多指示语模型”认为最大投射为Nmax，而中间投射分别为N”和N’，最多容纳四级投射；而“广义模型”认为最大投射为N”，中间投射均为N’，因此可以容纳更多的中间投射，能够灵活表征更加复杂的音节结构；2、“多指示语模型”认为在最大投射和终结点之间还有CV层，而“广义模型”取消了C、V等节点，明确提出，各层次投射的终节点均为代表某一音段的区别特征集，这些特征集中的区别性特征按照特征几何理论、标记性理论和不充分赋值理论排列，从而避免了底层特征的冗余。
此后，本文提出假设：“广义多指示语的X标杆模型”可以表征主要汉语方言的音节结构。为了证明该假设，本文以北京话、上海话、广州话和厦门话等四种主要汉语方言为例对汉语音节结构进行研究。具体研究步骤为：首先应用“广义模型”对方言的音节结构进行总体表征，然后针对音节中首音、介音、音节核和音节尾音等成分提出具体假设，最后通过列举丰富的证据逐一证明这些具体假设的合理性，从而证明总体表征的合理性。这些证据主要来自如下5个方面：1、前人研究成果；2、体现在方言民谣、流行歌曲、反切语、诗歌、熟语等文学艺术形式中的汉语押韵规则；3、针对某一方言音系现象的心理语言学实验；4、方言声调系统；5、音节成分之间的共现制约。本文在搜集和分析证据的过程中主要使用了6种具体的研究方法：文献回顾法，心理语言学实验法，反切系联法，分布分析法，语言比较法和区别特征分析法。前两者主要用于搜集证据，后面四种主要用于分析证据。
值得一提的是，本文对生成音系学框架下的汉语声调结构研究进行了细致的梳理，并在此基础上提出，汉语声调与音节之间通过莫拉（即音节重）层相互联系，最后提出了针对汉语的“音节结构-音节重-声调相关性模型”。本文从端木三（1993,1994）的研究中为该模型找到了直接的语音学数据支持。该模型是本论文用以确定汉语方言韵母结构的基础。
本文的最后结论是：“广义多指示语的X标杆模型”能够简单明了直观的表征汉语方言的音节结构，这说明汉语各方言的音节结构有很大的共性，这些共性主要表现为：1、没有复杂首音；2、存在核前介音；3、存在辅音节尾。但研究中我们也发现各种方言的音节成分有特殊之处，具体而言，1、在音节首音方面，北京话和广东话的音节可以没有首音，而上海话的音节必须有首音；2、在核前介音方面，北京话有明显的[j]、[w]介音，广东话中的[j]介音已经消失，[w]介音只存在于软腭辅音[k]和[kh]之后；上海话中的介音与首音关系相对紧密，而厦门话中介音与韵的关系相对紧密；3、在核后介音方面，上海话没有核后介音，而其他三种方言均有明显核后介音；4、在核后鼻辅音方面，北京话有两种核后鼻辅音[n]和[]，上海话只有一个鼻音位[N]，广东话和厦门话都有三种核后鼻音[m],[n]和[]，厦门话中还存在鼻化元音；5、在核后辅音方面，北京话没有核后辅音，上海话中有喉塞辅音节尾，广东话中有塞音尾[p/t/k]，厦门话中有塞音尾和喉音尾；6、在声调方面，北京话没有入声调，而其他三种方言均存在入声调。这些不同之处对具体方言的音节结构在投射层次、终结点的表征和音节重等方面产生影响，但并不影响总体表征的合理性。
This dissertation proposes a Generalized Multi-spec X-bar Model for therepresentation of Chinese syllable structure based on the Multi-spec X-bar Modelproposed by van de Weijer&Zhang (2008) and through a comprehensive applicationof syllable-related theories within the framework of Generative Phonology (GP),especially the X-bar Theory. After an in-depth study on the syllable structure of fourmajor Chinese dialects, this dissertation concludes that the Generalized Multi-specX-bar Model adequately captures the syllable structure of major Chinese dialects.
The X-bar Model in phonology was proposed by Levin (1983,1985), and dealsspecifically with syllable structure. It draws upon the idea of the X-bar representationof syntactic structure and applies it to the representation of syllable structure inphonology. Based on the X-bar Model and enlightened by the Multiple-Specconstruction proposed by Chomsky (1995) for sentence structure, van de Weijer&Zhang (2008) proposed a Multiple-Spec X-bar Model for Mandarin Chinese (MC)syllable structure through an in-depth study of the status of the pre-nuclear glide inMC syllables. In this model, the syllable node Nmaxis the maximal projection, and thepre-nuclear glide is the terminal node of the left-ward branch of the middle projectionN”. Based on the Multiple-Spec X-bar Model, this dissertation proposes aGeneralized Multi-spec X-bar Model, by drawing upon the theories of FeatureGeometry, Markedness Theory, Underspecification Theory and Mora Theory,
The Generalized Multi-spec X-bar Model is different from the Multi-spec X-barModel in that:(i) the latter can accommodate at most four levels of projection, whilethe former is more flexible in accommodating different syllable types since it allows an indefinite number of intermediate N’ projections;(ii) the latter proposes that thereis a CV tier between the projection nodes and the terminal nodes, while the formerproposes that the projection nodes are directly associated with the terminal nodes,which stand for clusters of distinctive features in the lexicon, organized and specifiedaccording to the theory of Feature Geometry, Markedness Theory andUnderspecification Theory, and thus avoiding redundant information in the lexicon.
This dissertation then makes the claim that the Generalized Multi-spec X-barModel can capture the syllable structure of major Chinese dialects. In order to justifythis claim, this dissertation presents an in-depth study of the syllable structure of fourmajor Chinese dialects: MC, Shanghai dialect, Cantonese and Amoy. To be morespecific, this dissertation first makes the general claim that the overall syllablestructure of a specific dialect can be captured by the Generalized Multi-spec X-barModel, and then translates this general claim into specific hypotheses about specificconstituents of the syllable, including the onset, pre-nuclear glide, rhyme, nucleus andcoda. These hypotheses require justification, which mainly comes from the followingfive sources of evidence:(i) research results presented in previous related studies;(ii)rhyming systems as embodied in folk songs, popular songs, local Fanqie languagegames, children’s lyrics and idioms;(iii) psycholinguistic experiments designed toanswer specific phonological questions;(iv) evidence from the tone patterns; and (v)co-occurrence restrictions (CRs) on syllable constituents. In the process of obtainingthe evidence, this dissertation uses two major methods: literature review andpsycholinguistic experiments. In the process of analyzing the evidence, thisdissertation uses four major methods: association through Fanqie syllables, analysisof phonological distribution, language comparison, and distinctive feature analysis.
In view of the close relationship between the tone and the syllable in Chineselanguages, this dissertation presents a thorough review of major theories concerningChinese tones, and claims that the relationship between Chinese syllables and tone ismediated by syllable weight. The Syllable Structure-Weight-Tone Correlation Modelis proposed to capture this relationship. This model shows that the segmental tier islinked to the tone tier through the mora tier. Phonological evidence for this model is adduced partly on the basis of Duanmu’s (1993,1994) studies, which also providefirst-hand data from phonetic experiments that support this model. This model thusserves as the argument for the rhyme structure of Chinese languages.
This dissertation concludes that the Generalized Multi-spec X-bar Modeladequately captures the syllable structures of all the four major Chinese dialectsstudied in this dissertation, showing that there are many commonalities betweenChinese dialects in terms of syllable structure, which include the following points:(i)complex onsets are not allowed;(ii) there is a pre-nuclear glide;(iii) consonantalcodas exist. This dissertation also discovers some major distinctions in syllableconstituents between the dialects under study:(i) in terms of the onset, MC andCantonese have onsetless syllables while Shanghainese syllables must have onsets;(ii)in terms of the pre-nuclear glide, MC syllables obviously have pre-nuclear glides [j]and [w], while in Cantonese, the pre-nuclear glide [j] has disappeared, and [w] canonly co-occur with the velar stops [k] and [kh]. The pre-nuclear glides inShanghainese are closer to the onset while those in Amoy are closer to the rhyme;(iii)in terms of the post-nuclear glide, Shanghainese does not have syllables withpost-nuclear glides, while MC, Cantonese and Amoy do;(iv) in terms of thepost-nuclear nasals, MC syllables have two nasal codas, i.e.[n] and []; Shanghainesesyllables only have one nasal phoneme [N]; Cantonese and Amoy syllables have threekinds of nasal codas [m],[n] and []; there are nasalized vowels in Amoy;(v) in termsof the consonantal codas, MC syllables do not allow any consonantal codas;Shanghainese syllables allow the glottal coda; Cantonese syllables allow the stopconsonants [p/t/k]; and Amoy syllables allow both stop consonant codas and theglottal coda;(vi) in terms of the tone, MC does not have the entering tone, whileShanghainese, Cantonese and Amoy have. All these differences have impacts on thesyllable structure in terms of the number of projection levels and the representation ofthe terminal nodes, but do not influence the validity of the X-bar representation of theoverall syllable structure.
音系学中有关音节结构表征的X标杆理论由J.列文（1983,1985）提出，该理论的创新之处在于借用句法学中的X标杆理论对音节结构进行抽象化表征。Van de Weijer&Zhang（2008）通过研究汉语标准语（即普通话）音节中的介音归属，参考乔姆斯基等(1995)提出的关于句子结构的“多指示语的X标杆模型”，对音系学的X标杆理论进行修改，提出了针对汉语普通话音节结构的“多指示语的X标杆模型”（简称“多指示语模型”）。该模型认为音节是核心元音V的最大投射Nmax，介音是中间投射N”的左向分支。本文借鉴生成音系学中的特征几何理论、特征标记性理论、不充分赋值理论以及莫拉理论等对“多指示语的X标杆模型”进行改进，提出了“广义多指示语的X标杆模型”（简称“广义模型”），
“广义模型”与“多指示语模型”的不同之处在于：1、“多指示语模型”认为最大投射为Nmax，而中间投射分别为N”和N’，最多容纳四级投射；而“广义模型”认为最大投射为N”，中间投射均为N’，因此可以容纳更多的中间投射，能够灵活表征更加复杂的音节结构；2、“多指示语模型”认为在最大投射和终结点之间还有CV层，而“广义模型”取消了C、V等节点，明确提出，各层次投射的终节点均为代表某一音段的区别特征集，这些特征集中的区别性特征按照特征几何理论、标记性理论和不充分赋值理论排列，从而避免了底层特征的冗余。
此后，本文提出假设：“广义多指示语的X标杆模型”可以表征主要汉语方言的音节结构。为了证明该假设，本文以北京话、上海话、广州话和厦门话等四种主要汉语方言为例对汉语音节结构进行研究。具体研究步骤为：首先应用“广义模型”对方言的音节结构进行总体表征，然后针对音节中首音、介音、音节核和音节尾音等成分提出具体假设，最后通过列举丰富的证据逐一证明这些具体假设的合理性，从而证明总体表征的合理性。这些证据主要来自如下5个方面：1、前人研究成果；2、体现在方言民谣、流行歌曲、反切语、诗歌、熟语等文学艺术形式中的汉语押韵规则；3、针对某一方言音系现象的心理语言学实验；4、方言声调系统；5、音节成分之间的共现制约。本文在搜集和分析证据的过程中主要使用了6种具体的研究方法：文献回顾法，心理语言学实验法，反切系联法，分布分析法，语言比较法和区别特征分析法。前两者主要用于搜集证据，后面四种主要用于分析证据。
值得一提的是，本文对生成音系学框架下的汉语声调结构研究进行了细致的梳理，并在此基础上提出，汉语声调与音节之间通过莫拉（即音节重）层相互联系，最后提出了针对汉语的“音节结构-音节重-声调相关性模型”。本文从端木三（1993,1994）的研究中为该模型找到了直接的语音学数据支持。该模型是本论文用以确定汉语方言韵母结构的基础。
本文的最后结论是：“广义多指示语的X标杆模型”能够简单明了直观的表征汉语方言的音节结构，这说明汉语各方言的音节结构有很大的共性，这些共性主要表现为：1、没有复杂首音；2、存在核前介音；3、存在辅音节尾。但研究中我们也发现各种方言的音节成分有特殊之处，具体而言，1、在音节首音方面，北京话和广东话的音节可以没有首音，而上海话的音节必须有首音；2、在核前介音方面，北京话有明显的[j]、[w]介音，广东话中的[j]介音已经消失，[w]介音只存在于软腭辅音[k]和[kh]之后；上海话中的介音与首音关系相对紧密，而厦门话中介音与韵的关系相对紧密；3、在核后介音方面，上海话没有核后介音，而其他三种方言均有明显核后介音；4、在核后鼻辅音方面，北京话有两种核后鼻辅音[n]和[]，上海话只有一个鼻音位[N]，广东话和厦门话都有三种核后鼻音[m],[n]和[]，厦门话中还存在鼻化元音；5、在核后辅音方面，北京话没有核后辅音，上海话中有喉塞辅音节尾，广东话中有塞音尾[p/t/k]，厦门话中有塞音尾和喉音尾；6、在声调方面，北京话没有入声调，而其他三种方言均存在入声调。这些不同之处对具体方言的音节结构在投射层次、终结点的表征和音节重等方面产生影响，但并不影响总体表征的合理性。
This dissertation proposes a Generalized Multi-spec X-bar Model for therepresentation of Chinese syllable structure based on the Multi-spec X-bar Modelproposed by van de Weijer&Zhang (2008) and through a comprehensive applicationof syllable-related theories within the framework of Generative Phonology (GP),especially the X-bar Theory. After an in-depth study on the syllable structure of fourmajor Chinese dialects, this dissertation concludes that the Generalized Multi-specX-bar Model adequately captures the syllable structure of major Chinese dialects.
The X-bar Model in phonology was proposed by Levin (1983,1985), and dealsspecifically with syllable structure. It draws upon the idea of the X-bar representationof syntactic structure and applies it to the representation of syllable structure inphonology. Based on the X-bar Model and enlightened by the Multiple-Specconstruction proposed by Chomsky (1995) for sentence structure, van de Weijer&Zhang (2008) proposed a Multiple-Spec X-bar Model for Mandarin Chinese (MC)syllable structure through an in-depth study of the status of the pre-nuclear glide inMC syllables. In this model, the syllable node Nmaxis the maximal projection, and thepre-nuclear glide is the terminal node of the left-ward branch of the middle projectionN”. Based on the Multiple-Spec X-bar Model, this dissertation proposes aGeneralized Multi-spec X-bar Model, by drawing upon the theories of FeatureGeometry, Markedness Theory, Underspecification Theory and Mora Theory,
The Generalized Multi-spec X-bar Model is different from the Multi-spec X-barModel in that:(i) the latter can accommodate at most four levels of projection, whilethe former is more flexible in accommodating different syllable types since it allows an indefinite number of intermediate N’ projections;(ii) the latter proposes that thereis a CV tier between the projection nodes and the terminal nodes, while the formerproposes that the projection nodes are directly associated with the terminal nodes,which stand for clusters of distinctive features in the lexicon, organized and specifiedaccording to the theory of Feature Geometry, Markedness Theory andUnderspecification Theory, and thus avoiding redundant information in the lexicon.
This dissertation then makes the claim that the Generalized Multi-spec X-barModel can capture the syllable structure of major Chinese dialects. In order to justifythis claim, this dissertation presents an in-depth study of the syllable structure of fourmajor Chinese dialects: MC, Shanghai dialect, Cantonese and Amoy. To be morespecific, this dissertation first makes the general claim that the overall syllablestructure of a specific dialect can be captured by the Generalized Multi-spec X-barModel, and then translates this general claim into specific hypotheses about specificconstituents of the syllable, including the onset, pre-nuclear glide, rhyme, nucleus andcoda. These hypotheses require justification, which mainly comes from the followingfive sources of evidence:(i) research results presented in previous related studies;(ii)rhyming systems as embodied in folk songs, popular songs, local Fanqie languagegames, children’s lyrics and idioms;(iii) psycholinguistic experiments designed toanswer specific phonological questions;(iv) evidence from the tone patterns; and (v)co-occurrence restrictions (CRs) on syllable constituents. In the process of obtainingthe evidence, this dissertation uses two major methods: literature review andpsycholinguistic experiments. In the process of analyzing the evidence, thisdissertation uses four major methods: association through Fanqie syllables, analysisof phonological distribution, language comparison, and distinctive feature analysis.
In view of the close relationship between the tone and the syllable in Chineselanguages, this dissertation presents a thorough review of major theories concerningChinese tones, and claims that the relationship between Chinese syllables and tone ismediated by syllable weight. The Syllable Structure-Weight-Tone Correlation Modelis proposed to capture this relationship. This model shows that the segmental tier islinked to the tone tier through the mora tier. Phonological evidence for this model is adduced partly on the basis of Duanmu’s (1993,1994) studies, which also providefirst-hand data from phonetic experiments that support this model. This model thusserves as the argument for the rhyme structure of Chinese languages.
This dissertation concludes that the Generalized Multi-spec X-bar Modeladequately captures the syllable structures of all the four major Chinese dialectsstudied in this dissertation, showing that there are many commonalities betweenChinese dialects in terms of syllable structure, which include the following points:(i)complex onsets are not allowed;(ii) there is a pre-nuclear glide;(iii) consonantalcodas exist. This dissertation also discovers some major distinctions in syllableconstituents between the dialects under study:(i) in terms of the onset, MC andCantonese have onsetless syllables while Shanghainese syllables must have onsets;(ii)in terms of the pre-nuclear glide, MC syllables obviously have pre-nuclear glides [j]and [w], while in Cantonese, the pre-nuclear glide [j] has disappeared, and [w] canonly co-occur with the velar stops [k] and [kh]. The pre-nuclear glides inShanghainese are closer to the onset while those in Amoy are closer to the rhyme;(iii)in terms of the post-nuclear glide, Shanghainese does not have syllables withpost-nuclear glides, while MC, Cantonese and Amoy do;(iv) in terms of thepost-nuclear nasals, MC syllables have two nasal codas, i.e.[n] and []; Shanghainesesyllables only have one nasal phoneme [N]; Cantonese and Amoy syllables have threekinds of nasal codas [m],[n] and []; there are nasalized vowels in Amoy;(v) in termsof the consonantal codas, MC syllables do not allow any consonantal codas;Shanghainese syllables allow the glottal coda; Cantonese syllables allow the stopconsonants [p/t/k]; and Amoy syllables allow both stop consonant codas and theglottal coda;(vi) in terms of the tone, MC does not have the entering tone, whileShanghainese, Cantonese and Amoy have. All these differences have impacts on thesyllable structure in terms of the number of projection levels and the representation ofthe terminal nodes, but do not influence the validity of the X-bar representation of theoverall syllable structure.
引文
1Strict rhyme requires that words that rhyme with each other have the same nucleus plus coda, while lax rhymerequires that words that rhyme with each other have a similar nucleus. In practice, rhyming falls within the twoextremes.
2In these examples, whether the prenuclear glide belongs to the Initial or the Final is not clear. In the example of“双”, the prenuclear glide belongs to the Initial, but in the example of “腔”, the prenuclear glide belongs to theFinal. This thesis will discuss this topic in greater detail in Chapter IV.
3The mismatch between the represented syllable and the representational syllables is caused by the differencebetween Middle Chinese and modern Chinese in phonology.
4In these examples, we use the Middle Chinese syllables in order to explain the sequential equivalence of thesesyllables.
5It should be noted that sounds that are in complementary distribution may still be different phonemes. Otheraspects, such as phonetic similarity, should also be taken into consideration.
6These rules are different from phonological rules in Generative studies. The latter refer to the rules that state thechanges in phonological derivation, while the former refers to the rules that state the phonological facts.
7GP is a general term for all theories that follow the Government tradition established by Kaye, Lowenstamm&Vergnaud (1990), while GC is GP in strict sense, which also includes the Charm Theory which mainly addressesthe issue of syllabification within the GP framework.
8An onset with skeletal material but without melodic material suggests a possible way to explain lengthening orreduction, while an onset without skeletal material suggests that such phonological processes could not take place.
9A segmentally empty nucleus may be filled by the “cold vowel”[] in GP.
10These are not the illustrations in Zhang (2008), but the adaptation of this dissertation according to the X-barmodel.
11This fact proves that syllable-final nasals have the same status in phonology as the syllable-final vowel glides,according to a well-known rule in phonology which says that two segments which are in complementarydistribution in the syllable have the same phonological status.
12For example, English has syllabic nasals such as the final syllable in the word button.
13For greater detail, refer to the section about MC in Chapter3.
14This thesis uses GP to refer to Generative Phonology, while it uses GC to refer to Government and CharmPhonology, also known as Government Phonology.
15Except for the so-called “light” tone syllables, which is the subject of study by many Chinese phonologists.
16Yinchuan is the capital of northwest China’s Ningxia Province.
17Spoken in north China’s Tianjin municipality.
18The reason why nasals can be [+syllabic] will be explained in the next chapter.
19For detailed reviews of these theories, please refer to Chapter II.
20But every feature is on an independent tier, so that a feature can act separately to result in such phenomena asfeature spreading and feature assimilation.
21Tone is a very important part of Chinese syllables, but it is mainly features attached to the nucleus. This thesiswill leave it out when illustrating the syllable structure of different Chinese dialects, but will discuss itsrelationship with the syllable structure in a separate section (see Section3.2).
22The same is true for syntax. Although the determiner phrase is generated in time before the complement, thecomplement has a closer relationship with the head.
23Here “closer relationship” involves three aspects: i) In terms of the interaction of features, the features of thecoda have a greater influence on the features of the nucleus than those of the onsets do; ii) in rhyming systems,“nucleus+coda” is considered to be a more typical rhyme than “onset+nucleus”; iii) Psychologically speaking,people have a stronger tendency of considering the coda as a more natural rhyming part than they do with theonset (REF to psycholinguistics evidence for this?).
24Bao (1990) imposes a constraint on the number of tone autosegments so that there should be no more than3pitches on one single TBU.
25Here “q” indicates “checked”.
26Duanmu (1990) regards these as downsteps.
27These examples are used to illustrated the possible contour tones predicted by Duanmu’s (1990) model, and donot necessarily exist in actual languages.
28According to Chen (2001), Bao (1990) claims that the TBU is the syllable, but in fact Bao explicitly claims thatthe TBU is the rhyme (Bao1990: abstract).
30See Bao (1990) for the original version of the tone feature geometry.
29Logically speaking, there can be such contour tones as hhl, hll, llh and lhh, but on second thought thesecontours are the same as hl and lh, and therefore are redundant. A formal way of ruling out structures like hhl, ll,etc. is by invoking the Obligatory Contour Principle (OCP).
30This corresponds to their claim that tone is an autosegment. Recall that although Duanmu claims that tone is asegmental feature, he still use a “Root” in the tone feature geometry, which is very confusing. Therefore, I put a (?)after the “yes” in the “root node” column.
31Contour spread refers to the process that the contour of one syllable of a multi-syllable word (or a syllable inanother word) spreads to the other syllables whose contour features have been deleted while keeping the register(s)of all the syllables so that the contour of the whole word is the same as the spread contour. Contour shift refers tothe process that the contour of one syllable replaces that of another within a multi-syllable (usually two) word.
32Chen (2001) maintains that Yip’s (1989) model cannot predict contour spread/shift, but as far as this author isconcerned, since Yip recognizes contour tones, her model should be able to predict contour spread or shift.
33Register spread refers to the process that the register of one syllable becomes the register for all other syllableswithin a multi-syllable (usually two) word. Register shift refers to the process that the register of one syllablereplaces that of another syllable within a a multi-syllable (usually two) word.
34Whole contour tone spread refers to the process that the feature representation of one syllable is divided acrossthe whole multi-syllable word. Whole contour tone shift refers to the process that the feature representation of onesyllable replaces that of another syllable within a multi-syllable (usually two) word.
35Terminal node spread refers to the process that a level tone node under the contour node spreads by itself to anyterminal node of another syllable within a multi-syllable (usually two) word. Terminal node shift refers to theprocess that the level tone node under the contour node replaces the terminal node of another syllable within amulti-syllable (usually two) word.
36This claim is different from that of Bao’s (1990).
37For more details, see Sections4.2-4.4.
38M-languages refer to Chinese languages whose tone patterns and tone sandhi patterns are similar to that of MC.
39S-languages refer to Chinese languages whose tone patterns and tone sandhi patterns are similar to that of theShanghai dialect.
40This thesis has a major difference in opinion on this point, see Section4.2. But it doesn’t influence the analysisthat follows in Duanmu’s (1994) study.
41Duanmu (1994) cites Zhu (1986) and Xu et al (1988) to show that in S-languages all rhymes are either nasalizedvowels or glottalized vowels, and that there is no coda at all. In M-languages, on the other hand, there are differentcodas, such as [n] and [] in Mandarin, and [p],[t] and [k] in Cantonese. This thesis doesn’t agree with Duanmu’s
(1994) analysis of the syllable codas in Chinese, but agrees with it on its analysis of the tone-weight relationship inChinese. For more details, see Chapter4.
42These factors can be implemented as constraints as in OT.
43They are the glide correspondents of the vowel/i/,/u/and/y/. In this dissertation, we use these symbolsinterchangeably for the glides except when it is necessary to discriminate between them.
44The corresponding Chinese character is not found in the Pinyin Input software.
45Although this form is permitted in some Chinese dialects, it is not permitted in MC.
46This is a possible word in Chinese, but not valid as a liaison form for “fan an”.
47These four variants of the zero onset are found in different dialects in China, but these dialects all belong toMC.
48These variants are not all valid in Beijing MC, but they occur in other varieties of MC, such as the dialects inHubei and Sichuan Provinces, which belongs to the MC dialect group in general.
49Here consonants such as/m/and/n/are concerned. But according to Duanmu (1990,2000,2008) otherconsonants such as/z/can also be the nucleus, which are referred to as syllabic consonants. This thesis does notagree with this claim either, and will elaborate on this point later.
50For further details about the two poems, please refer to http://baike.baidu.com/view/47169.htm, andhttp://baike.baidu.com/view/155622.htm. Although the nasal/n/is the only common segment of the rhyme, itiimportant to realize that the rhyme consists of the nasalized vowels.
51We don’t give all the examples in order to save space.
52Note that the actual pronunciation is changed from/n/into/an/, which is due to a low-level phonetic rule.These two representations are underlyingly the same.
53Here I just use phrases instead of sentences for simple illustration.
54The name of a fairy lady who lives in the moon; also the name of the Chinese spacecraft that landed on themoon in2014.
55CNstands for the nasal codas.
56Tone III is a special case. We will argue in this section that it is composed of two feature values underlyingly,but gains the extra tone value if it occurs in isolation.
57The complexity of T3sandhi is far greater than suggested by these three rules. Duammu (2000,2007) has aspecial chapter discussing T3sandhi patterns.
58Different dialects of MC treat the rhyme of the unstressed syllable differently. In the northern part of China, thesimplification of the rhyme together with the loss of the tone is quite common, while in the southern part of China,the simplification is not very common.
59There were voiced onsets in ancient Chinese. For details, please refer to the next section.
60The data comes fromhttp://www.baike.com/wiki/%E4%B8%8A%E6%B5%B7%E8%AF%9D&prd=button_doc_jinru.
61According to Qian (2007), there are28initial consonants in New Shanghai, including a zero onset “”.
62For a detailed discussion of the changes taking place in Shanghai, see Qian (1987).
63This illustration will be revised after the argumentation in this chapter.
64See http://zh.wikipedia.org/wiki/%E4%B8%8A%E6%B5%B7%E8%AF%9D.
65According to Duanmu (2008), the CVX structure represents both the feature geometry and the timing slots ofthe syllable.“C” represents the onset consonant, and the VX represents two possible combinations of rhymes: VVand VC. Every syllable has three timing slots.
66This procedure can be omitted here because the second onset is the same as the first one. We list it here forclarity.
67In this example, the second onset is/t/instead of/p/, because of the co-occurrence restriction (CR) between thebilabial onset and the nasalized coda. The consequence of this CR is that there is no syllable like*/pia/inShanghai.
68But Duanmu (1990) argues otherwise. For greater detail, refer to Section4.4about Amoy.
69This lyric is sung everywhere in China, and the following transcription is the version in Shanghai dialect.
70Refer to Note2for details about this song.
71Although the three varieties of Shanghai dialect took shape in sequence, they co-exist in Shanghai today, withOld Shanghai being spoken by senior locals, Middle Shanghai by middle-aged local people, and New Shanghai byyoung people in Shanghai.
72Qian (2007) uses [] to represent the voiced glottal.
73See http://www.baike.com/wiki/%E4%B8%8A%E6%B5%B7%E8%AF%9D&prd=button_doc_jinru.
74The underlining signifies the syllabic feature.
75See Duanmu (1994).
76The underlining in the table signifies the short version of the tone.
77“T” refers to any tone.
78The rule says that Shanghai syllables in connected speech lose their stress and only have one mora.
79Qian (2007) lists four kinds, because he distinguishes between and. Now they merge into one phoneme//.
80//represents a nasalized vowel. But the [+palatal] feature of//seems to spread to the preceding vowel, sothat V also has the features [+nasal,+palatal]. It is also reasonable to group it into the vowel-final syllables. Herein this thesis we treat it as the nasal-final rhyme.
81Duanmu (2008) uses/a/to represent this phoneme. The difference doesn’t influence our argumentation heresince we only pay attention to the nasal feature of this phoneme. In this respect, Qian (2007) and Duanmu (2008)share the same opinion.
82Some scholars categorize/ts/,/tsh/, and/s/together with/t/,/t h/and//because these groups of sounds are incomplementary distribution.
83These segments are referred to as labiovelars by Cheung (1986). Their status is controversial. This thesis claimsthat these segments involve combinations of the onsets/k/and/kh/with the pre-nuclear glide/w/.
84stands for the zero onset. This thesis claims, however, that syllables with the zero onset are in fact onsetlesssyllables. For more details, refer to Section4.1.
85There are other versions of the number of rhymes in Cantonese, e.g. according to Yuan (2006), there are53rhymes in Cantonese. However, this discrepancy is a result of the difference as to whether the phonemic variantsshould be counted in. Therefore, the difference in number is negligible.
86Cheung (1986) translates the four contours as even, ascend, depart, and enter, which are the literal translationsof the four Chinese words for the contours.
87See also Killingley (1993:6).
88The reason why some scholars think that there are11tones in Cantonese is that they count the two varieties ofT1and T4as two independent tones, while scholars who think that there are6tones in Cantonese categorize thechecked tones T7, T8and T9together with their smooth counterparts T1, T6and T3, respectively.
89In this thesis, the vowel glides/i/and/u/are often transcribed as/j/and/w/. In ancient Chinese,/j/and/w/werepart of the complex onsets.
90Here,/i/is long and part of the nucleus, and/u/is the post-nuclear vowel glide.
91For more details about the songs and lyrics, please refer to the appendix.
92The name of this song is Nan’er Dang Ziqiang, which means that a man should be self-reliant.
93These pairs are not actual words in Cantonese, but possible rhymes (item1-3) or alliterations (item4-6) inCantonese rhyming literature, such as Cantonese popular songs and children’s lyrics.
94Here C stands for the consonantal codas/p/,/t/and/k/.
95Coda gemination adds one mora to the syllable (see Roca1990).
96Here this is equivalent to ‘no coda’, and which will not play a role in the following argumentation.
97In contrast to right-headed rhymes, i.e. rhymes in which the right-hand vowel of a diphthong is the major vowel,and the left one a glide, such as/ia/and/ua/.
98Also known as Huiji Yasutong Shiwuyin, or ‘ACollection ofAmoy Phonology’.
99This number comes from Yuan (1959,1980,2000). Other opinions include57(Duanmu1990; Luo1956),89
(Wang2008), and99(Ma2002). The difference can be attributed to whether phonemes or allophones are listed.
100Here/i/and/u/denotes the pre-nuclear glides.
101Some scholars (Yuan1959,1980,2000) use the nasalization diacritic “~” on all segments, others (Ma2002; Li2011) use it only on the nucleus, and still others (Duanmu1990) use the diacritic on both the nucleus and thepost-nuclear glide. This thesis agrees with Duanmu (1990) for two reasons: firstly, it is the whole rhyme, whichcomprises the nuclear vowel and the post-nuclear glide, which is nasalized; secondly, this makes it easier for us tocompare Amoy with MC.
102Please see the website http://zh.wikipedia.org/wiki/%E5%BB%88%E9%96%80%E8%A9%B1.
103According to the website in the previous note, there are in fact8tones in Amoy. But Yangshang in modernAmoy has merged with Yangqu, thus reducing the number of Amoy tones to7.
104There are other opinions concerning the exact values. We follow the Wikipedia source mentioned in the noteabove.
105The underlining indicates an occlusion in the coda.
106Here “C” refers only to the consonantal codas/p/,/t/, and/k/. Although the glottal stop//can be aconsonantal coda, however, its special status leads us to use//to refer specifically to the glottal coda.
107For more details, please refer to Section1of Chapter4.
108The last four sources are quoted from Duanmu (1990).
109This means that this rule is not operative in this language.
110Note that in this Fanqie language, the onsets of both syllables have the same place of articulation, so that in thefollowing example,/l/is changed into/t/, instead of/p/. This is not the case in Taiwanese La-mi and Ma-sa.
111Since Taiwanese is a variety of the southern Min dialects, we hereby consider it the same as Amoy inphonology in relevant respects.
112Duanmu (1990) refers to this language game as La-pi. That is to say, the onset of the second syllable in thisFanqie language is a bilabial. Here we adopt Zhong’s (1997) name.
113The “N” in these structures corresponds to R.
114These words are vernacular in Amoy and corresponding characters in written form are not found.
115Strictly speaking, Chinese syllable finals comprise the pre-nuclear glide and the rhyme. This thesis uses theterm ‘rhyme’ to refer to both, for convenience. The term “final” is used only when it is necessary to differentiatebetween rhymes and finals.
116See the rhyme table in Section4.4.1.
117Refer to http://zh.wikipedia.org/wiki/%E5%BB%88%E9%96%80%E8%A9%B1.
118The list is adapted by this author taking into account the status of the pre-nuclear glide.
119Note that the nasalization diacritic is only used on the nucleus instead of on every segment of the rhyme.
120Vp refers to rhymes ending with the stop consonants/p, t, k/.
121Here C{p/t/k/} stands for the consonantal coda p/t/k and the glottal consonant.
122The nucleus is nasalized as a result of the spreading of the [+nasal] feature of the post-nuclear glide.
123Here this refers to absence of a coda.
124The starred rhyme types are not found in actual Amoy syllables.
125There are differences between the four lists of rhymes in this thesis. In Luo’s (1956) and Ma’s (2002) lists,
there are no VG rhymes, but in Yuan’s (2002) and Wikipedia’s lists, there are VG rhymes. We adopt the latter for
the convenience of comparison.
126The sign “≠” means “must be different from.”
127The sign “≈” means “may be the same as.”
128For more details about the sonority hierarchy, refer to Chapter2.
129The reason why and how the register of the tones remains despite the loss of the contrast of the onsets on the[voice] feature is seldom discussed and studied. This thesis leaves this question open.
130In actual analysis of the syllables in Chinese languages, this thesis still uses C/V to represent the terminalnodes for clarity.
131Please see Chapter3for more details about this principle.
132The following representations are adapted from those in Section4.1. For more details, refer to Section4.1.
133“Gl” here refers to the glottal//.
134For more details, see Chapter1.
Bao, Z.1990. On the Nature of Tone. PhD dissertation. MIT.
Broselow, E. etc.1997. Syllable Weight: Convergence of Phonology and Phonetics.Phonology,14,47-82.
Bybee, J.2004. Phonology and Language Use. New York: Cambridge UniversityPress.
Carlisle, R.2001. Syllable Structure Universals and Second Language Acquisition.International Journal of English Studies,1(1):1-19.
Chen, M. Y.2000. Tone Sandhi: Patterns across Chinese Dialects. New York:Cambridge University Press.
Cheung, G.1986. The Phonology of Present-day Cantonese. PhD dissertation.University College.
Chomsky, N.&M. Halle.1968. The Sound Pattern of English. New York: Harper&Row.
Chomsky, N.2008. The Minimalist Program. Beijing: Foreign Language Teachingand Research Press.
Clark, J.&Yallop, C.1990,1995. An Introduction to Phonetics and Phonology.Blackwell Publishers Ltd (online sources).
Clayton, M.1976. The Redundancy of Morpheme Structure Redundancies. Language,52:292-313.
Clements, N.&Keyser, S.1983. CV Phonology: Generative Theories of the Syllable.Cambridge: MIT Press.
Coleman, J.1998. Phonological Representations: Their Names, Forms and Powers.New York: Cambridge University Press.
Cohn, A. C.1993. Nasalization in English: Phonology or Phonetics. Phonology,10(1):43-81.
Davenport, M.&Hannahs, S. J.2005. Introducing Phonetics and Phonology.Berlin/New York: Mouton de Gruyter.
Duanmu, S.1990. A Formal Study of Syllable Tone, Stress&Domain in ChineseLanguages. PhD Dissertation. MIT.
Duanmu, S.1994. Syllabic Weight and Syllabic Duration: A Correlation betweenPhonology and Phonetics. Phonology,11(1):1-24.
Duanmu, S.2000,2007. The Phonology of Standard Chinese. New York: CambridgeUniversity Press.
Duanmu, S.2008. Syllable Structure: the Limits of Variation. New York: OxfordUniversity Press.
Edmondson, J.A&Esling, J.H.2006. The Valves of the Throat and Their Functioningin Tone, Vocal Register and Stress: Laryngoscopic Case Studies. Phonology23:157-191.
Fery, C.&Van de Vijver, R.(eds.).2003. The syllable in Optimality Theory. NewYork: Cambridge University Press.
Fudge, E.1969. Syllables. Journal of Linguistics,5:253-286.
Goldsmith, J. A.1996. The Handbook of Phonological Theory. Hoboken: BlackwellPublishing.
Gordon, M.2002. A Phonetic Driven Account of Syllable Weight. Language,78(1):51-80.
Gordon, M.2006. Syllable Weight: Phonetics, Phonology, Typology. NewYork/London:Routledge.
Hayes, B.1989. Compensatory Lengthening in Moraic Phonology. Linguistic Inquiry,20:253-306.
Hammond, M.1997. Vowel Quantity and Syllabification in English. Language,73(1):1-17.
Hoard, J.1971. Asperation, Tenseness and Syllabification in English. Language,47:133-140.
Hyman, L.1985. A Theory of Phonological Weight. Dodrecht: Foris Publications.
Hsu, H.2003. A Particle Analysis of ü in Chinese. The Tsing Hua Journal of ChineseStudies,33(1):173-197.
Inkelas, S.1994. The Consequence of Optimization for Underspecification.http://roa.rutgers.edu/files/40-1294/40-1294-INKELAS-0-0.PDF.
Jensen, J.2004. Principles of Generative Phonology. Amsterdam/Philadelphia: JohnBenjamins Publishing Company.
Kager, R.1999,2001. Optimality Theory. New York: Cambridge University Press.Beijing: Foreign Language Teaching and Research Press.
Kager, R., Pater, J.&Zonneveld, W.(eds.).2004. Constraints in PhonologicalAcquisition. New York: Cambridge University Press.
Kahn, D.1976. Syllable-based generalizations in English phonology. Ph.Ddissertation. MIT.
Keating, P.1994. Phonological Structure and Phonetic Form. New York: CambridgeUniversity Press.
Kaisse, E.&Shaw, P.1985. On the Theory of Lexical Phonology. PhonologyYearbook,2:1-30.
Kügler, F., Féry, C.&Van de Vijver, R.(eds).2009. Variation and Gradience inPhonetics and Phonology. Berlin/New York: Mouton de Gruyter.
Ladd, R.2008. Intonational Phonology (2ndedn.). New York: Cambridge UniversityPress.
Ladefoged, P.2009. A Course in Phonetics. Singapore: Cengage Learning. Beijing:Foreign Language Teaching and Research Press.
Leben, W.1973. Segmental Phonology. PhD Dissertation. MIT.
Levin, J.1985. A Metrical Theory of Syllabicity. PhD Dissertation. MIT.
Levelt, C., Schiller, N.&Levelt, W.1999/2000. The Acquisition of Syllable Types.Language Acquisition,8(3):237-264.
Lin, Y.2001. A Constraint-based Approach to Chinese Speakers’ of EnglishConsonant Clusters. Concentric: Studies in English Literature and Linguistics,27(2):1-33.
Li, A.2002. Chinese Prosody and Prosodic Labeling of Spontaneous Speech. SpeechProsody Conference, Paris.
McCarthy, J.1988. Feature geometry and dependency: A review. Phonetica,45:84-108.
McCarthy, J.&Prince, A.1990. Prosodic Morphology and Templatic Morphology.Current Issues in Linguistic Theory, Vol.72:1-54.
Mohanan, K.1981. Lexical Phonology. PhD dissertation. MIT.
McCawley, J.1968. The Phonological Component of a Grammar of Japanese. TheHague: Mouton.
Nespor, M.&Vogel, I.2007. Prosodic Phonology. Berlin/New York: Mouton deGruyer.
Odden, D.2008. Introducing Phonology. Beijing: Foreign Language Teaching andResearch Press.
Packard, J.2001. The Morphology of Chinese: A Linguistic and Cognitive Approach.Beijing: Foreign Language Teaching and Research Press.
Pellegrino, F., Marsico, E., Chitoran, I.&Coupé, C.(ed.)2009. Approaches toPhonological Complexity. Berlin/New York: Mouton de Gruyer.
Prince, A.&Smolensky, P.2004. Optimality Theory. Malden: Blackwell Publishing.
Roca, I.1994. Generative Phonology. New York/London: Routledge.
Roach, P.1983,1991. English Phonetics and Phonology: A Practical Course. NewYork: Cambridge University Press.
Sohn, H.S.1994. Syllabification in Tangale: A Prosodic Analysis. Grammars.org.
Van de Hulst, H.&. Ritter, N.1999. The Syllable: Views and Facts. Berlin/New York:Mouton de Grutyer.
Van de Weijer, J.&Zhang, J.2008. An X-bar approach to the syllable structure ofMandarin. Lingua, doi:10.1016/j.lingua.2007.09.006
Yip, M.1980. A Tonal Phonology of Chinese. PhD Dissertation. MIT.
Yip, M.1989. Feature Geometry and Co-occurrence Restrictions. Phonology,6:349-374.
Yip, M.1996. Tone in East Asian Languages. The Handbook of Phonological Theory:374-389. Blackwell, reference online.
Zhang, J.2006. The Phonology of Shaoxing Chinese. Utrecht: LOT.
Zhang, J.&van de Weijer, J..2012. An OT-CC Analysis of Phonological Opacity inShaoxing Tri-syllabic Tone Sandhi. Journal of Chinese Linguistics,40(1):139-154.
包智明、侍建国、许德宝，1997.生成音系学理论及其应用[M].北京：中国社会科学出版社.
北大中文系语言学教研室（编），2003.汉语方音字汇[Z].北京：语文出版社.
曹晓燕，2012.方言与普通话的接触研究[D].苏州大学博士学位论文.
陈洁（2003）,昆明话音系系统初探[J].云南师范大学学报（哲学社会科学版），35（1）：98-101.
杜伟东，1999.普通话朗诵教程[M].北京：警官教育出版社.
傅灵，2010.方言与普通话的接触研究---以长沙、上海、武汉为背景[D].苏州大学博士学位论文.
耿振生.2004.20世纪汉语音韵学方法论[M].北京：北京大学出版社.
国玮秋，1999.元音后响辅音在音节内的归属[J].外语学刊，3:57-63.
宫齐,&范俊军，2003.优选论及其在汉语音系中的应用[J].暨南学报（哲学社会科学版）,25(3):88-85.
---,2005.现代音系学音节理论面面观：《音节论：理论与事实》述要[J].暨南学报（哲学社会科学版），4:88-93.
---，2011.汉语方言塞音韵尾保留情况的优选论分析[J].暨南学报（哲学社会科学版）,6:104-110.
顾钦，2007.语言接触对上海市区方言语音演变的影响[D].上海师范大学博士学位论文.
贺俊杰,2010a.管辖音系学的声调理论[J].当代语言学，12(3):252-261.
——,2010b.“天津话连续变调之谜”再探[J].南开语言学刊，2:23-36.
——,2011a.跨方言声调生成及变调处理[J].清华大学学报，51(9):1258-1260.
——,2011b.镇江方言声调音系过程的管辖音系学解释[J].方言，1:37-43.
胡开宝,2006.汉外语言接触近百年:回顾与展望[J].外语与外语教学,206(5):53-57.
姜美玉，2001.汉语口误研究[D].中国社会科学院博士学位论文.
李兵，1990.音系标记理论的产生与发展[J].新疆师范大学学报（哲社版）.3:46-52.
---，1998.优选论的产生、基本原理与应用[J].现代外语,3:71-91.
---，2001.当代音系学的方法论特征[J].现代外语，24（1）：1-22.
李越,2008.试论管辖音系学质量分离表征的合理性[J].四川外语学院学报,24(4):66-62.
——,2010. sC音丛的音节结构：空韵核、偶分韵基和首音[J].外语与外语教学,1:36-41.
李志强，1997.生成音系学的音节理论[J].外语教学与研究，4:4-11.
李新魁，1990.粤方言语音特点探论[J].广东社会科学，1：149-160.
林焘、耿振生，2008.音韵学概要[M].北京：商务出版社.
林秋茗,2002，广州话中的U音[J].语言研究,1：114-120.
刘蕾，2008.粤语童谣的修辞语用特色[J].东莞理工学院学报，6:59-63.
刘娇蛟,2006，汉语普通话的管辖音系学特征及提取方法[J].电子学报，10：1917-1919.
刘晓娟，2005.不充分赋值理论概述[J].吕梁教育学院学报，22（3）：37-39.
刘纶鑫,2001.音韵学基础教程[M].北京：中国社会科学出版社.
刘倩，2010.香港粤语塞擦音、擦音声母的音变考察---以三十年来香港粤语歌词为例[J].暨南学报（哲学社会科学版），149（6）：123-129.
陆继伦、王嘉龄,2012.汉语轻声的优选论分析[M].天津：天津大学出版社.
罗常培,1956.厦门音韵[M].北京：科学出版社.
马秋武，2000a.管辖音系学：一种以制约为基础的音系学理论[J].解放军外国语学院学报,23(1):15-20.
---,2000b.管辖音系学的基本理论及最新发展[J].当代语言学,2(4):218-226.
---，2001.后SPE音系学理论的发展取向[J].外国语，3：15-22.
---,2004.普通话二合元音韵母的组构与优选论分析[J].当代语言学,6(1):19-25.
---,2005a.“天津话连续变调之谜”的优选论解释[J].中国语文，6：561-569.
---,2005b.再论“天津话连续变调之谜”[J].当代语言学，7:97-106.
---&吴力菡，2012.三论“天津话连续变调之谜”[J].当代语言学,14(1):12-23.
马重奇,1994.闽南漳州方言中的反切语[J].福建师范大学学报（哲学社会科学版），1:70-81.
马重奇,2002.闽台闽南话与普通话韵母系统比较研究[J].福建论坛（人文社会科学版），4：99-107.
麦耘，2009.从粤语的产生和发展看汉语方言形成的模式[J].方言，(3):219-232.
钱乃荣，1987.上海方言音变的微观[J].语言研究，2:104-115.
---、许宝华、汤珍珠，2007.上海话大辞典[Z].上海辞书出版社.
---，2007.上海方言[M].上海：文汇出版社.
---，2006.新世纪上海话流行语2500条[M].上海：汉语大辞典出版社.
---，2002.沪语盘点：上海话文化[M].上海：上海文艺出版社.
曲彦斌(主编)，1995.中国隐语行话大辞典续编---隐语行话研究事典[Z].沈阳：辽宁教育出版社.
邵慧君，2010.论粤方言i、u介音韵母——由粤西方言说起[J].暨南学报（哲学社会科学版），6:116-122.
沈同，1985.新派上海话声调的底层形式[J].语言研究，9(2):85-101.
施俊,2009.关于义乌方言“n”化元音时长的讨论[J].中国语文，第6期。
石峰，1990.再论天津话声调及其变化[J].语言研究，2：15-24.
孙景涛,2007.从儿化看音节的重量[J].中国语文，4:298-301.
王志洁，1997.英汉音节鼻韵尾的不同性质[J].现代外语，4:17-29.
王茂林,2005.音系学的时长理论[J].当代语言学,7(2):113-121.
王轶之，2011.吴语塞音声母的声学和感知研究---以上海话为例[D].浙江大学博士学位论文.
王晓梅，2009.天津方言三字组连读变调的优选论分析[J].语言研究，29(1):33-39.
王嘉龄，2002.优选论及天津话的连读变调及轻声[J].中国语文，4:363-373.
王力，2013.汉语音韵[M].北京：中华书局.
王红梅，2008.生成音系学在中国20年来的发展历程[J].四川外语学院学报，24（2）:98-101.
徐越、朱晓农,2011.喉塞尾入声是怎么舒化的——孝丰个案研究[J].中国语文，3:263-270.
闫小斌，2007.中国优选论研究十年[J].外国语言文学，2:83-86.
---,2013.汉语方向性连读变调研究述评[J].山东理工大学学报(社会科学版)，29（4）：76-79.
杨剑桥，2005.汉语音韵学讲义[M].上海：复旦大学出版社。
---，2012.汉语现代音韵学[M].上海:复旦大学出版社.
杨蓓，2002.上海儿童的上海话音系习得[D].上海复旦大学博士学位论文.
杨春生，2006.近年来海内外汉语介音归属研究综述[J].当代语言学，8（2）：144-155.
杨军，2012.汉语母语者ESL/EFL音节划分案例的再分析[J].南京邮电大学学报（哲学社会科学版），14(1):75-82.
袁家骅等,1960,1983,2006.汉语方言概要[M].北京：北京文字改革出版社.
詹伯慧，1999.粤语研究的回顾与展望[J].暨南学报（哲学社会科学版），6:11-21.
赵元任,1931.反切语八种[A].中研院史语所集刊，2：320-354.
张洪明&尹玉霞，2012.优选论的是与非:现代音系学研究的若干反思[J].中国语文,6:483-499.
张洪年，2002.21世纪的香港粤语:一个新语音系统的形成[J].暨南学报（哲学社会科学版），24（2）：25-40.
张吉生,1993.英汉音节结构对比[J].宁波大学学报,7(1):56-62.
---，2003.优选论与汉语外来语的音节可接受性[J].解放军外国语学院学报，26（4）：7-12.
---，2006.再论汉语外来语音节可接受性的优选分析[J].外国语，2：25-32.
---,2007.汉语韵尾辅音演变的音系理据[J].中国语文，4：291-298.
---，2008a.汉语音节结构再分析[J].外国语，31(1):45-52.
---，2008b.绍兴方言连续变调的特征几何分析[A].第八届中国语音学学术会议、语音科学问题国际研讨会论文集.
张天堡，2001.中国民间反切语简论[J].淮北煤师院学报（哲学社会科学版），22:（1）：116-117,133.
张光宇,2012,汉语方言的鼻化运动[J].语言研究，32（2）：17-28.
钟荣富,1997，台湾闽南语的介音[J].中国语文,4:289-294.---,1995.优选论与汉语的音系[J].国外语言学,3:1-14.
周长楫,1999.重读《厦门音系》[J].方言,3：176-180.
周长楫，2001.厦门方言熟语歌谣[M].福建人民出版社.
朱晓农，2007.说鼻音[J].语言研究，27（3）：1-13.
2In these examples, whether the prenuclear glide belongs to the Initial or the Final is not clear. In the example of“双”, the prenuclear glide belongs to the Initial, but in the example of “腔”, the prenuclear glide belongs to theFinal. This thesis will discuss this topic in greater detail in Chapter IV.
3The mismatch between the represented syllable and the representational syllables is caused by the differencebetween Middle Chinese and modern Chinese in phonology.
4In these examples, we use the Middle Chinese syllables in order to explain the sequential equivalence of thesesyllables.
5It should be noted that sounds that are in complementary distribution may still be different phonemes. Otheraspects, such as phonetic similarity, should also be taken into consideration.
6These rules are different from phonological rules in Generative studies. The latter refer to the rules that state thechanges in phonological derivation, while the former refers to the rules that state the phonological facts.
7GP is a general term for all theories that follow the Government tradition established by Kaye, Lowenstamm&Vergnaud (1990), while GC is GP in strict sense, which also includes the Charm Theory which mainly addressesthe issue of syllabification within the GP framework.
8An onset with skeletal material but without melodic material suggests a possible way to explain lengthening orreduction, while an onset without skeletal material suggests that such phonological processes could not take place.
9A segmentally empty nucleus may be filled by the “cold vowel”[] in GP.
10These are not the illustrations in Zhang (2008), but the adaptation of this dissertation according to the X-barmodel.
11This fact proves that syllable-final nasals have the same status in phonology as the syllable-final vowel glides,according to a well-known rule in phonology which says that two segments which are in complementarydistribution in the syllable have the same phonological status.
12For example, English has syllabic nasals such as the final syllable in the word button.
13For greater detail, refer to the section about MC in Chapter3.
14This thesis uses GP to refer to Generative Phonology, while it uses GC to refer to Government and CharmPhonology, also known as Government Phonology.
15Except for the so-called “light” tone syllables, which is the subject of study by many Chinese phonologists.
16Yinchuan is the capital of northwest China’s Ningxia Province.
17Spoken in north China’s Tianjin municipality.
18The reason why nasals can be [+syllabic] will be explained in the next chapter.
19For detailed reviews of these theories, please refer to Chapter II.
20But every feature is on an independent tier, so that a feature can act separately to result in such phenomena asfeature spreading and feature assimilation.
21Tone is a very important part of Chinese syllables, but it is mainly features attached to the nucleus. This thesiswill leave it out when illustrating the syllable structure of different Chinese dialects, but will discuss itsrelationship with the syllable structure in a separate section (see Section3.2).
22The same is true for syntax. Although the determiner phrase is generated in time before the complement, thecomplement has a closer relationship with the head.
23Here “closer relationship” involves three aspects: i) In terms of the interaction of features, the features of thecoda have a greater influence on the features of the nucleus than those of the onsets do; ii) in rhyming systems,“nucleus+coda” is considered to be a more typical rhyme than “onset+nucleus”; iii) Psychologically speaking,people have a stronger tendency of considering the coda as a more natural rhyming part than they do with theonset (REF to psycholinguistics evidence for this?).
24Bao (1990) imposes a constraint on the number of tone autosegments so that there should be no more than3pitches on one single TBU.
25Here “q” indicates “checked”.
26Duanmu (1990) regards these as downsteps.
27These examples are used to illustrated the possible contour tones predicted by Duanmu’s (1990) model, and donot necessarily exist in actual languages.
28According to Chen (2001), Bao (1990) claims that the TBU is the syllable, but in fact Bao explicitly claims thatthe TBU is the rhyme (Bao1990: abstract).
30See Bao (1990) for the original version of the tone feature geometry.
29Logically speaking, there can be such contour tones as hhl, hll, llh and lhh, but on second thought thesecontours are the same as hl and lh, and therefore are redundant. A formal way of ruling out structures like hhl, ll,etc. is by invoking the Obligatory Contour Principle (OCP).
30This corresponds to their claim that tone is an autosegment. Recall that although Duanmu claims that tone is asegmental feature, he still use a “Root” in the tone feature geometry, which is very confusing. Therefore, I put a (?)after the “yes” in the “root node” column.
31Contour spread refers to the process that the contour of one syllable of a multi-syllable word (or a syllable inanother word) spreads to the other syllables whose contour features have been deleted while keeping the register(s)of all the syllables so that the contour of the whole word is the same as the spread contour. Contour shift refers tothe process that the contour of one syllable replaces that of another within a multi-syllable (usually two) word.
32Chen (2001) maintains that Yip’s (1989) model cannot predict contour spread/shift, but as far as this author isconcerned, since Yip recognizes contour tones, her model should be able to predict contour spread or shift.
33Register spread refers to the process that the register of one syllable becomes the register for all other syllableswithin a multi-syllable (usually two) word. Register shift refers to the process that the register of one syllablereplaces that of another syllable within a a multi-syllable (usually two) word.
34Whole contour tone spread refers to the process that the feature representation of one syllable is divided acrossthe whole multi-syllable word. Whole contour tone shift refers to the process that the feature representation of onesyllable replaces that of another syllable within a multi-syllable (usually two) word.
35Terminal node spread refers to the process that a level tone node under the contour node spreads by itself to anyterminal node of another syllable within a multi-syllable (usually two) word. Terminal node shift refers to theprocess that the level tone node under the contour node replaces the terminal node of another syllable within amulti-syllable (usually two) word.
36This claim is different from that of Bao’s (1990).
37For more details, see Sections4.2-4.4.
38M-languages refer to Chinese languages whose tone patterns and tone sandhi patterns are similar to that of MC.
39S-languages refer to Chinese languages whose tone patterns and tone sandhi patterns are similar to that of theShanghai dialect.
40This thesis has a major difference in opinion on this point, see Section4.2. But it doesn’t influence the analysisthat follows in Duanmu’s (1994) study.
41Duanmu (1994) cites Zhu (1986) and Xu et al (1988) to show that in S-languages all rhymes are either nasalizedvowels or glottalized vowels, and that there is no coda at all. In M-languages, on the other hand, there are differentcodas, such as [n] and [] in Mandarin, and [p],[t] and [k] in Cantonese. This thesis doesn’t agree with Duanmu’s
(1994) analysis of the syllable codas in Chinese, but agrees with it on its analysis of the tone-weight relationship inChinese. For more details, see Chapter4.
42These factors can be implemented as constraints as in OT.
43They are the glide correspondents of the vowel/i/,/u/and/y/. In this dissertation, we use these symbolsinterchangeably for the glides except when it is necessary to discriminate between them.
44The corresponding Chinese character is not found in the Pinyin Input software.
45Although this form is permitted in some Chinese dialects, it is not permitted in MC.
46This is a possible word in Chinese, but not valid as a liaison form for “fan an”.
47These four variants of the zero onset are found in different dialects in China, but these dialects all belong toMC.
48These variants are not all valid in Beijing MC, but they occur in other varieties of MC, such as the dialects inHubei and Sichuan Provinces, which belongs to the MC dialect group in general.
49Here consonants such as/m/and/n/are concerned. But according to Duanmu (1990,2000,2008) otherconsonants such as/z/can also be the nucleus, which are referred to as syllabic consonants. This thesis does notagree with this claim either, and will elaborate on this point later.
50For further details about the two poems, please refer to http://baike.baidu.com/view/47169.htm, andhttp://baike.baidu.com/view/155622.htm. Although the nasal/n/is the only common segment of the rhyme, itiimportant to realize that the rhyme consists of the nasalized vowels.
51We don’t give all the examples in order to save space.
52Note that the actual pronunciation is changed from/n/into/an/, which is due to a low-level phonetic rule.These two representations are underlyingly the same.
53Here I just use phrases instead of sentences for simple illustration.
54The name of a fairy lady who lives in the moon; also the name of the Chinese spacecraft that landed on themoon in2014.
55CNstands for the nasal codas.
56Tone III is a special case. We will argue in this section that it is composed of two feature values underlyingly,but gains the extra tone value if it occurs in isolation.
57The complexity of T3sandhi is far greater than suggested by these three rules. Duammu (2000,2007) has aspecial chapter discussing T3sandhi patterns.
58Different dialects of MC treat the rhyme of the unstressed syllable differently. In the northern part of China, thesimplification of the rhyme together with the loss of the tone is quite common, while in the southern part of China,the simplification is not very common.
59There were voiced onsets in ancient Chinese. For details, please refer to the next section.
60The data comes fromhttp://www.baike.com/wiki/%E4%B8%8A%E6%B5%B7%E8%AF%9D&prd=button_doc_jinru.
61According to Qian (2007), there are28initial consonants in New Shanghai, including a zero onset “”.
62For a detailed discussion of the changes taking place in Shanghai, see Qian (1987).
63This illustration will be revised after the argumentation in this chapter.
64See http://zh.wikipedia.org/wiki/%E4%B8%8A%E6%B5%B7%E8%AF%9D.
65According to Duanmu (2008), the CVX structure represents both the feature geometry and the timing slots ofthe syllable.“C” represents the onset consonant, and the VX represents two possible combinations of rhymes: VVand VC. Every syllable has three timing slots.
66This procedure can be omitted here because the second onset is the same as the first one. We list it here forclarity.
67In this example, the second onset is/t/instead of/p/, because of the co-occurrence restriction (CR) between thebilabial onset and the nasalized coda. The consequence of this CR is that there is no syllable like*/pia/inShanghai.
68But Duanmu (1990) argues otherwise. For greater detail, refer to Section4.4about Amoy.
69This lyric is sung everywhere in China, and the following transcription is the version in Shanghai dialect.
70Refer to Note2for details about this song.
71Although the three varieties of Shanghai dialect took shape in sequence, they co-exist in Shanghai today, withOld Shanghai being spoken by senior locals, Middle Shanghai by middle-aged local people, and New Shanghai byyoung people in Shanghai.
72Qian (2007) uses [] to represent the voiced glottal.
73See http://www.baike.com/wiki/%E4%B8%8A%E6%B5%B7%E8%AF%9D&prd=button_doc_jinru.
74The underlining signifies the syllabic feature.
75See Duanmu (1994).
76The underlining in the table signifies the short version of the tone.
77“T” refers to any tone.
78The rule says that Shanghai syllables in connected speech lose their stress and only have one mora.
79Qian (2007) lists four kinds, because he distinguishes between and. Now they merge into one phoneme//.
80//represents a nasalized vowel. But the [+palatal] feature of//seems to spread to the preceding vowel, sothat V also has the features [+nasal,+palatal]. It is also reasonable to group it into the vowel-final syllables. Herein this thesis we treat it as the nasal-final rhyme.
81Duanmu (2008) uses/a/to represent this phoneme. The difference doesn’t influence our argumentation heresince we only pay attention to the nasal feature of this phoneme. In this respect, Qian (2007) and Duanmu (2008)share the same opinion.
82Some scholars categorize/ts/,/tsh/, and/s/together with/t/,/t h/and//because these groups of sounds are incomplementary distribution.
83These segments are referred to as labiovelars by Cheung (1986). Their status is controversial. This thesis claimsthat these segments involve combinations of the onsets/k/and/kh/with the pre-nuclear glide/w/.
84stands for the zero onset. This thesis claims, however, that syllables with the zero onset are in fact onsetlesssyllables. For more details, refer to Section4.1.
85There are other versions of the number of rhymes in Cantonese, e.g. according to Yuan (2006), there are53rhymes in Cantonese. However, this discrepancy is a result of the difference as to whether the phonemic variantsshould be counted in. Therefore, the difference in number is negligible.
86Cheung (1986) translates the four contours as even, ascend, depart, and enter, which are the literal translationsof the four Chinese words for the contours.
87See also Killingley (1993:6).
88The reason why some scholars think that there are11tones in Cantonese is that they count the two varieties ofT1and T4as two independent tones, while scholars who think that there are6tones in Cantonese categorize thechecked tones T7, T8and T9together with their smooth counterparts T1, T6and T3, respectively.
89In this thesis, the vowel glides/i/and/u/are often transcribed as/j/and/w/. In ancient Chinese,/j/and/w/werepart of the complex onsets.
90Here,/i/is long and part of the nucleus, and/u/is the post-nuclear vowel glide.
91For more details about the songs and lyrics, please refer to the appendix.
92The name of this song is Nan’er Dang Ziqiang, which means that a man should be self-reliant.
93These pairs are not actual words in Cantonese, but possible rhymes (item1-3) or alliterations (item4-6) inCantonese rhyming literature, such as Cantonese popular songs and children’s lyrics.
94Here C stands for the consonantal codas/p/,/t/and/k/.
95Coda gemination adds one mora to the syllable (see Roca1990).
96Here this is equivalent to ‘no coda’, and which will not play a role in the following argumentation.
97In contrast to right-headed rhymes, i.e. rhymes in which the right-hand vowel of a diphthong is the major vowel,and the left one a glide, such as/ia/and/ua/.
98Also known as Huiji Yasutong Shiwuyin, or ‘ACollection ofAmoy Phonology’.
99This number comes from Yuan (1959,1980,2000). Other opinions include57(Duanmu1990; Luo1956),89
(Wang2008), and99(Ma2002). The difference can be attributed to whether phonemes or allophones are listed.
100Here/i/and/u/denotes the pre-nuclear glides.
101Some scholars (Yuan1959,1980,2000) use the nasalization diacritic “~” on all segments, others (Ma2002; Li2011) use it only on the nucleus, and still others (Duanmu1990) use the diacritic on both the nucleus and thepost-nuclear glide. This thesis agrees with Duanmu (1990) for two reasons: firstly, it is the whole rhyme, whichcomprises the nuclear vowel and the post-nuclear glide, which is nasalized; secondly, this makes it easier for us tocompare Amoy with MC.
102Please see the website http://zh.wikipedia.org/wiki/%E5%BB%88%E9%96%80%E8%A9%B1.
103According to the website in the previous note, there are in fact8tones in Amoy. But Yangshang in modernAmoy has merged with Yangqu, thus reducing the number of Amoy tones to7.
104There are other opinions concerning the exact values. We follow the Wikipedia source mentioned in the noteabove.
105The underlining indicates an occlusion in the coda.
106Here “C” refers only to the consonantal codas/p/,/t/, and/k/. Although the glottal stop//can be aconsonantal coda, however, its special status leads us to use//to refer specifically to the glottal coda.
107For more details, please refer to Section1of Chapter4.
108The last four sources are quoted from Duanmu (1990).
109This means that this rule is not operative in this language.
110Note that in this Fanqie language, the onsets of both syllables have the same place of articulation, so that in thefollowing example,/l/is changed into/t/, instead of/p/. This is not the case in Taiwanese La-mi and Ma-sa.
111Since Taiwanese is a variety of the southern Min dialects, we hereby consider it the same as Amoy inphonology in relevant respects.
112Duanmu (1990) refers to this language game as La-pi. That is to say, the onset of the second syllable in thisFanqie language is a bilabial. Here we adopt Zhong’s (1997) name.
113The “N” in these structures corresponds to R.
114These words are vernacular in Amoy and corresponding characters in written form are not found.
115Strictly speaking, Chinese syllable finals comprise the pre-nuclear glide and the rhyme. This thesis uses theterm ‘rhyme’ to refer to both, for convenience. The term “final” is used only when it is necessary to differentiatebetween rhymes and finals.
116See the rhyme table in Section4.4.1.
117Refer to http://zh.wikipedia.org/wiki/%E5%BB%88%E9%96%80%E8%A9%B1.
118The list is adapted by this author taking into account the status of the pre-nuclear glide.
119Note that the nasalization diacritic is only used on the nucleus instead of on every segment of the rhyme.
120Vp refers to rhymes ending with the stop consonants/p, t, k/.
121Here C{p/t/k/} stands for the consonantal coda p/t/k and the glottal consonant.
122The nucleus is nasalized as a result of the spreading of the [+nasal] feature of the post-nuclear glide.
123Here this refers to absence of a coda.
124The starred rhyme types are not found in actual Amoy syllables.
125There are differences between the four lists of rhymes in this thesis. In Luo’s (1956) and Ma’s (2002) lists,
there are no VG rhymes, but in Yuan’s (2002) and Wikipedia’s lists, there are VG rhymes. We adopt the latter for
the convenience of comparison.
126The sign “≠” means “must be different from.”
127The sign “≈” means “may be the same as.”
128For more details about the sonority hierarchy, refer to Chapter2.
129The reason why and how the register of the tones remains despite the loss of the contrast of the onsets on the[voice] feature is seldom discussed and studied. This thesis leaves this question open.
130In actual analysis of the syllables in Chinese languages, this thesis still uses C/V to represent the terminalnodes for clarity.
131Please see Chapter3for more details about this principle.
132The following representations are adapted from those in Section4.1. For more details, refer to Section4.1.
133“Gl” here refers to the glottal//.
134For more details, see Chapter1.
Bao, Z.1990. On the Nature of Tone. PhD dissertation. MIT.
Broselow, E. etc.1997. Syllable Weight: Convergence of Phonology and Phonetics.Phonology,14,47-82.
Bybee, J.2004. Phonology and Language Use. New York: Cambridge UniversityPress.
Carlisle, R.2001. Syllable Structure Universals and Second Language Acquisition.International Journal of English Studies,1(1):1-19.
Chen, M. Y.2000. Tone Sandhi: Patterns across Chinese Dialects. New York:Cambridge University Press.
Cheung, G.1986. The Phonology of Present-day Cantonese. PhD dissertation.University College.
Chomsky, N.&M. Halle.1968. The Sound Pattern of English. New York: Harper&Row.
Chomsky, N.2008. The Minimalist Program. Beijing: Foreign Language Teachingand Research Press.
Clark, J.&Yallop, C.1990,1995. An Introduction to Phonetics and Phonology.Blackwell Publishers Ltd (online sources).
Clayton, M.1976. The Redundancy of Morpheme Structure Redundancies. Language,52:292-313.
Clements, N.&Keyser, S.1983. CV Phonology: Generative Theories of the Syllable.Cambridge: MIT Press.
Coleman, J.1998. Phonological Representations: Their Names, Forms and Powers.New York: Cambridge University Press.
Cohn, A. C.1993. Nasalization in English: Phonology or Phonetics. Phonology,10(1):43-81.
Davenport, M.&Hannahs, S. J.2005. Introducing Phonetics and Phonology.Berlin/New York: Mouton de Gruyter.
Duanmu, S.1990. A Formal Study of Syllable Tone, Stress&Domain in ChineseLanguages. PhD Dissertation. MIT.
Duanmu, S.1994. Syllabic Weight and Syllabic Duration: A Correlation betweenPhonology and Phonetics. Phonology,11(1):1-24.
Duanmu, S.2000,2007. The Phonology of Standard Chinese. New York: CambridgeUniversity Press.
Duanmu, S.2008. Syllable Structure: the Limits of Variation. New York: OxfordUniversity Press.
Edmondson, J.A&Esling, J.H.2006. The Valves of the Throat and Their Functioningin Tone, Vocal Register and Stress: Laryngoscopic Case Studies. Phonology23:157-191.
Fery, C.&Van de Vijver, R.(eds.).2003. The syllable in Optimality Theory. NewYork: Cambridge University Press.
Fudge, E.1969. Syllables. Journal of Linguistics,5:253-286.
Goldsmith, J. A.1996. The Handbook of Phonological Theory. Hoboken: BlackwellPublishing.
Gordon, M.2002. A Phonetic Driven Account of Syllable Weight. Language,78(1):51-80.
Gordon, M.2006. Syllable Weight: Phonetics, Phonology, Typology. NewYork/London:Routledge.
Hayes, B.1989. Compensatory Lengthening in Moraic Phonology. Linguistic Inquiry,20:253-306.
Hammond, M.1997. Vowel Quantity and Syllabification in English. Language,73(1):1-17.
Hoard, J.1971. Asperation, Tenseness and Syllabification in English. Language,47:133-140.
Hyman, L.1985. A Theory of Phonological Weight. Dodrecht: Foris Publications.
Hsu, H.2003. A Particle Analysis of ü in Chinese. The Tsing Hua Journal of ChineseStudies,33(1):173-197.
Inkelas, S.1994. The Consequence of Optimization for Underspecification.http://roa.rutgers.edu/files/40-1294/40-1294-INKELAS-0-0.PDF.
Jensen, J.2004. Principles of Generative Phonology. Amsterdam/Philadelphia: JohnBenjamins Publishing Company.
Kager, R.1999,2001. Optimality Theory. New York: Cambridge University Press.Beijing: Foreign Language Teaching and Research Press.
Kager, R., Pater, J.&Zonneveld, W.(eds.).2004. Constraints in PhonologicalAcquisition. New York: Cambridge University Press.
Kahn, D.1976. Syllable-based generalizations in English phonology. Ph.Ddissertation. MIT.
Keating, P.1994. Phonological Structure and Phonetic Form. New York: CambridgeUniversity Press.
Kaisse, E.&Shaw, P.1985. On the Theory of Lexical Phonology. PhonologyYearbook,2:1-30.
Kügler, F., Féry, C.&Van de Vijver, R.(eds).2009. Variation and Gradience inPhonetics and Phonology. Berlin/New York: Mouton de Gruyter.
Ladd, R.2008. Intonational Phonology (2ndedn.). New York: Cambridge UniversityPress.
Ladefoged, P.2009. A Course in Phonetics. Singapore: Cengage Learning. Beijing:Foreign Language Teaching and Research Press.
Leben, W.1973. Segmental Phonology. PhD Dissertation. MIT.
Levin, J.1985. A Metrical Theory of Syllabicity. PhD Dissertation. MIT.
Levelt, C., Schiller, N.&Levelt, W.1999/2000. The Acquisition of Syllable Types.Language Acquisition,8(3):237-264.
Lin, Y.2001. A Constraint-based Approach to Chinese Speakers’ of EnglishConsonant Clusters. Concentric: Studies in English Literature and Linguistics,27(2):1-33.
Li, A.2002. Chinese Prosody and Prosodic Labeling of Spontaneous Speech. SpeechProsody Conference, Paris.
McCarthy, J.1988. Feature geometry and dependency: A review. Phonetica,45:84-108.
McCarthy, J.&Prince, A.1990. Prosodic Morphology and Templatic Morphology.Current Issues in Linguistic Theory, Vol.72:1-54.
Mohanan, K.1981. Lexical Phonology. PhD dissertation. MIT.
McCawley, J.1968. The Phonological Component of a Grammar of Japanese. TheHague: Mouton.
Nespor, M.&Vogel, I.2007. Prosodic Phonology. Berlin/New York: Mouton deGruyer.
Odden, D.2008. Introducing Phonology. Beijing: Foreign Language Teaching andResearch Press.
Packard, J.2001. The Morphology of Chinese: A Linguistic and Cognitive Approach.Beijing: Foreign Language Teaching and Research Press.
Pellegrino, F., Marsico, E., Chitoran, I.&Coupé, C.(ed.)2009. Approaches toPhonological Complexity. Berlin/New York: Mouton de Gruyer.
Prince, A.&Smolensky, P.2004. Optimality Theory. Malden: Blackwell Publishing.
Roca, I.1994. Generative Phonology. New York/London: Routledge.
Roach, P.1983,1991. English Phonetics and Phonology: A Practical Course. NewYork: Cambridge University Press.
Sohn, H.S.1994. Syllabification in Tangale: A Prosodic Analysis. Grammars.org.
Van de Hulst, H.&. Ritter, N.1999. The Syllable: Views and Facts. Berlin/New York:Mouton de Grutyer.
Van de Weijer, J.&Zhang, J.2008. An X-bar approach to the syllable structure ofMandarin. Lingua, doi:10.1016/j.lingua.2007.09.006
Yip, M.1980. A Tonal Phonology of Chinese. PhD Dissertation. MIT.
Yip, M.1989. Feature Geometry and Co-occurrence Restrictions. Phonology,6:349-374.
Yip, M.1996. Tone in East Asian Languages. The Handbook of Phonological Theory:374-389. Blackwell, reference online.
Zhang, J.2006. The Phonology of Shaoxing Chinese. Utrecht: LOT.
Zhang, J.&van de Weijer, J..2012. An OT-CC Analysis of Phonological Opacity inShaoxing Tri-syllabic Tone Sandhi. Journal of Chinese Linguistics,40(1):139-154.
包智明、侍建国、许德宝，1997.生成音系学理论及其应用[M].北京：中国社会科学出版社.
北大中文系语言学教研室（编），2003.汉语方音字汇[Z].北京：语文出版社.
曹晓燕，2012.方言与普通话的接触研究[D].苏州大学博士学位论文.
陈洁（2003）,昆明话音系系统初探[J].云南师范大学学报（哲学社会科学版），35（1）：98-101.
杜伟东，1999.普通话朗诵教程[M].北京：警官教育出版社.
傅灵，2010.方言与普通话的接触研究---以长沙、上海、武汉为背景[D].苏州大学博士学位论文.
耿振生.2004.20世纪汉语音韵学方法论[M].北京：北京大学出版社.
国玮秋，1999.元音后响辅音在音节内的归属[J].外语学刊，3:57-63.
宫齐,&范俊军，2003.优选论及其在汉语音系中的应用[J].暨南学报（哲学社会科学版）,25(3):88-85.
---,2005.现代音系学音节理论面面观：《音节论：理论与事实》述要[J].暨南学报（哲学社会科学版），4:88-93.
---，2011.汉语方言塞音韵尾保留情况的优选论分析[J].暨南学报（哲学社会科学版）,6:104-110.
顾钦，2007.语言接触对上海市区方言语音演变的影响[D].上海师范大学博士学位论文.
贺俊杰,2010a.管辖音系学的声调理论[J].当代语言学，12(3):252-261.
——,2010b.“天津话连续变调之谜”再探[J].南开语言学刊，2:23-36.
——,2011a.跨方言声调生成及变调处理[J].清华大学学报，51(9):1258-1260.
——,2011b.镇江方言声调音系过程的管辖音系学解释[J].方言，1:37-43.
胡开宝,2006.汉外语言接触近百年:回顾与展望[J].外语与外语教学,206(5):53-57.
姜美玉，2001.汉语口误研究[D].中国社会科学院博士学位论文.
李兵，1990.音系标记理论的产生与发展[J].新疆师范大学学报（哲社版）.3:46-52.
---，1998.优选论的产生、基本原理与应用[J].现代外语,3:71-91.
---，2001.当代音系学的方法论特征[J].现代外语，24（1）：1-22.
李越,2008.试论管辖音系学质量分离表征的合理性[J].四川外语学院学报,24(4):66-62.
——,2010. sC音丛的音节结构：空韵核、偶分韵基和首音[J].外语与外语教学,1:36-41.
李志强，1997.生成音系学的音节理论[J].外语教学与研究，4:4-11.
李新魁，1990.粤方言语音特点探论[J].广东社会科学，1：149-160.
林焘、耿振生，2008.音韵学概要[M].北京：商务出版社.
林秋茗,2002，广州话中的U音[J].语言研究,1：114-120.
刘蕾，2008.粤语童谣的修辞语用特色[J].东莞理工学院学报，6:59-63.
刘娇蛟,2006，汉语普通话的管辖音系学特征及提取方法[J].电子学报，10：1917-1919.
刘晓娟，2005.不充分赋值理论概述[J].吕梁教育学院学报，22（3）：37-39.
刘纶鑫,2001.音韵学基础教程[M].北京：中国社会科学出版社.
刘倩，2010.香港粤语塞擦音、擦音声母的音变考察---以三十年来香港粤语歌词为例[J].暨南学报（哲学社会科学版），149（6）：123-129.
陆继伦、王嘉龄,2012.汉语轻声的优选论分析[M].天津：天津大学出版社.
罗常培,1956.厦门音韵[M].北京：科学出版社.
马秋武，2000a.管辖音系学：一种以制约为基础的音系学理论[J].解放军外国语学院学报,23(1):15-20.
---,2000b.管辖音系学的基本理论及最新发展[J].当代语言学,2(4):218-226.
---，2001.后SPE音系学理论的发展取向[J].外国语，3：15-22.
---,2004.普通话二合元音韵母的组构与优选论分析[J].当代语言学,6(1):19-25.
---,2005a.“天津话连续变调之谜”的优选论解释[J].中国语文，6：561-569.
---,2005b.再论“天津话连续变调之谜”[J].当代语言学，7:97-106.
---&吴力菡，2012.三论“天津话连续变调之谜”[J].当代语言学,14(1):12-23.
马重奇,1994.闽南漳州方言中的反切语[J].福建师范大学学报（哲学社会科学版），1:70-81.
马重奇,2002.闽台闽南话与普通话韵母系统比较研究[J].福建论坛（人文社会科学版），4：99-107.
麦耘，2009.从粤语的产生和发展看汉语方言形成的模式[J].方言，(3):219-232.
钱乃荣，1987.上海方言音变的微观[J].语言研究，2:104-115.
---、许宝华、汤珍珠，2007.上海话大辞典[Z].上海辞书出版社.
---，2007.上海方言[M].上海：文汇出版社.
---，2006.新世纪上海话流行语2500条[M].上海：汉语大辞典出版社.
---，2002.沪语盘点：上海话文化[M].上海：上海文艺出版社.
曲彦斌(主编)，1995.中国隐语行话大辞典续编---隐语行话研究事典[Z].沈阳：辽宁教育出版社.
邵慧君，2010.论粤方言i、u介音韵母——由粤西方言说起[J].暨南学报（哲学社会科学版），6:116-122.
沈同，1985.新派上海话声调的底层形式[J].语言研究，9(2):85-101.
施俊,2009.关于义乌方言“n”化元音时长的讨论[J].中国语文，第6期。
石峰，1990.再论天津话声调及其变化[J].语言研究，2：15-24.
孙景涛,2007.从儿化看音节的重量[J].中国语文，4:298-301.
王志洁，1997.英汉音节鼻韵尾的不同性质[J].现代外语，4:17-29.
王茂林,2005.音系学的时长理论[J].当代语言学,7(2):113-121.
王轶之，2011.吴语塞音声母的声学和感知研究---以上海话为例[D].浙江大学博士学位论文.
王晓梅，2009.天津方言三字组连读变调的优选论分析[J].语言研究，29(1):33-39.
王嘉龄，2002.优选论及天津话的连读变调及轻声[J].中国语文，4:363-373.
王力，2013.汉语音韵[M].北京：中华书局.
王红梅，2008.生成音系学在中国20年来的发展历程[J].四川外语学院学报，24（2）:98-101.
徐越、朱晓农,2011.喉塞尾入声是怎么舒化的——孝丰个案研究[J].中国语文，3:263-270.
闫小斌，2007.中国优选论研究十年[J].外国语言文学，2:83-86.
---,2013.汉语方向性连读变调研究述评[J].山东理工大学学报(社会科学版)，29（4）：76-79.
杨剑桥，2005.汉语音韵学讲义[M].上海：复旦大学出版社。
---，2012.汉语现代音韵学[M].上海:复旦大学出版社.
杨蓓，2002.上海儿童的上海话音系习得[D].上海复旦大学博士学位论文.
杨春生，2006.近年来海内外汉语介音归属研究综述[J].当代语言学，8（2）：144-155.
杨军，2012.汉语母语者ESL/EFL音节划分案例的再分析[J].南京邮电大学学报（哲学社会科学版），14(1):75-82.
袁家骅等,1960,1983,2006.汉语方言概要[M].北京：北京文字改革出版社.
詹伯慧，1999.粤语研究的回顾与展望[J].暨南学报（哲学社会科学版），6:11-21.
赵元任,1931.反切语八种[A].中研院史语所集刊，2：320-354.
张洪明&尹玉霞，2012.优选论的是与非:现代音系学研究的若干反思[J].中国语文,6:483-499.
张洪年，2002.21世纪的香港粤语:一个新语音系统的形成[J].暨南学报（哲学社会科学版），24（2）：25-40.
张吉生,1993.英汉音节结构对比[J].宁波大学学报,7(1):56-62.
---，2003.优选论与汉语外来语的音节可接受性[J].解放军外国语学院学报，26（4）：7-12.
---，2006.再论汉语外来语音节可接受性的优选分析[J].外国语，2：25-32.
---,2007.汉语韵尾辅音演变的音系理据[J].中国语文，4：291-298.
---，2008a.汉语音节结构再分析[J].外国语，31(1):45-52.
---，2008b.绍兴方言连续变调的特征几何分析[A].第八届中国语音学学术会议、语音科学问题国际研讨会论文集.
张天堡，2001.中国民间反切语简论[J].淮北煤师院学报（哲学社会科学版），22:（1）：116-117,133.
张光宇,2012,汉语方言的鼻化运动[J].语言研究，32（2）：17-28.
钟荣富,1997，台湾闽南语的介音[J].中国语文,4:289-294.---,1995.优选论与汉语的音系[J].国外语言学,3:1-14.
周长楫,1999.重读《厦门音系》[J].方言,3：176-180.
周长楫，2001.厦门方言熟语歌谣[M].福建人民出版社.
朱晓农，2007.说鼻音[J].语言研究，27（3）：1-13.