葛成章

（男，2016级博士，专业：语言学及应用语言学，研究方向：句法学，导师：胡建华）

　　摘 要

　　本文以最简方案为理论框架，考察现代汉语无定NP的句法特性以及儿童语言获得情况。在理论构建阶段，本文系统梳理了与汉语无定NP分布密切相关的两类现象：有定限制和无定限制，并从句法结构和信息结构视角做出深入分析。根据胡建华(2008)对“句法-信息结构接口条件”的考察，本文提出汉语的信息结构已经融入到狭义句法之中，而英语等形态语言的信息结构则不具有这种特性。在此基础上，本文进一步假设汉语在进行句法推导时，会在词项阵列里呈现不可解读的语篇构型特征[u d]。[u d]需要被语篇中的旧信息成分核查并删除，否则表达式就会因违反“句法-信息结构接口条件”而崩溃。因此，与英语等印欧语不同，汉语的无定限制表现更为强烈。此外，信息结构还可以解释有定限制的消解问题。本文证明，当定指NP出现在支点位置时，或者其自身被用作对比焦点，或者整个存现句的信息结构已经改变。在对无定NP内部结构的考察中，本文区分了“一量名”和“数量名”两种形式。根据Grimshaw（1990）的“扩展投射”理论，本文假设当数词短语核心负载[-plural]特征时，会触发扩展投射，并以此解释“一量名”的分布与解读特性。最后，本文还详细讨论了汉语带挂尾成分存现句的句法结构问题。与以往研究不同，本文基于“新描写主义”（胡建华2018）对语言进行“细颗粒度描写”的思想，从存现动词的种类、挂尾成分的谓词属性以及支点名词的形式等几个不同的角度讨论挂尾存现句。本文以主-谓结构描述支点名词同个体谓词挂尾的句法关系，以增价结构描述支点名词同阶段谓词挂尾的句法关系，以此解释存现动词的子语类化问题。

　　在对儿童语言的考察中，本文首先对比描写了汉语儿童和英语儿童获得各类名词性成分的基本过程。通过考察CASS-Child和CHILDES两个语料库，发现汉语儿童与英语儿童对各类名词性单位的获得顺序基本相同：先获得普通名词和指示词，再获得二者的直接组合形式“指+名”，最后获得限定词或数量成分同普通名词的组合。本文认为这种获得状态是对儿童语言获得的“双向成熟假说”（胡建华2016a，2016b；彭鹭鹭2016；郭秀丽2016）的一种印证，显示儿童对特定语言结构的获得也遵循“从虚实两端同时向屈折层生长”的双向发育态势。其次，本文发现汉语儿童对“一量名”和“数量名”两种无定NP结构，在获得时间和产出数量都表现出明显的差别，这说明两种结构具有不同的句法属性。最后，本文对CASS-Child和CASS-Infants语料库的考察还发现，在汉语儿童产出的无定主语句里，无定NP基本都作直指或回指解读。而少数与成人产出存在差异的地方，大多源于儿童对“当下”场景的错误认知，进而造成无定NP的直指或回指用法错误，但是这类错误不会违反“句法-信息结构接口条件”。

　　关键词：有定限制；无定限制；句法-信息结构接口条件；d-特征；挂尾；“一量名”；“数量名”；双向成熟模式

　　Indefinite NP in Chinese: Its Syntax and Acquisition

　　Abstract

　　This dissertation investigates the syntactic properties of indefinite NP in Chinese as well as its acquisition by the Chinese-speaking children.

　　The distribution of indefinite NPs is regulated by two conditions. One is the Definiteness Restriction, which bans the occurrence of definite NP in the pivot position in existential constructions, and the other is the Indefiniteness Restriction, which bans the occurrence of indefinite NP in the sentence-initial position. It is argued in this dissertation that the distribution of indefinite NPs accounted for by the above-mentioned two conditions are actually determined by the Information Structure/IS in Chinese. Following an analysis of “syntax-information structure interface” proposed by Hu (2008), it is claimed in this dissertation that IS is part of Narrow Syntax in Chinese, though it is not in English. We hypothesized that IS in Chinese is associated with a discourse configurational feature, i.e., “d-feature” (cf. Miyagawa 2017), in the Lexical Array, which is assigned to the head of CP in derivation. The d-feature remains uninterpretable until it is checked by the old information in the discourse domain, otherwise the derivation will crash. Since indefinite NP conveys new information, it is assumed that if a sentence contains an indefinite subject, the d-feature must be checked by overt or covert elements. The Indefiniteness Restriction is then explained as a by-product of the checking mechanism of the d-feature. IS can also explain the violation of the Definiteness Restriction. It is noticed in the literature that the definite pivot in the existential sentence can either work as a focus bearing new information (e.g., Leonetti 2016), or to present old information if there are other focal elements in the sentence (e.g., Hu & Pan 2007). This dissertation offers a unified analysis for these two functions. We demonstrate that the information status of the existential sentence would be partitioned into a “Given-New” pattern when the definite pivot provides new information, while a “Focus-Tail” pattern (cf. Vallduví & Engdahl 1996) when it provides old information. As for the internal structure of the indefinite NP in Chinese, two forms are identified according to the different numeral elements (the number ‘one/yī’ versus other numbers) in the “numeral-classifier-noun” construction. Following Grimshaw (1990), we propose that the [-plural] feature associated with the Num0 position will trigger an extended projection, and such a structural analysis can explain the special syntactic and semantic properties of the number “one/yī” in Chinese. Furthermore, we have also investigated the coda construction in existential sentences, which was tangled with the subcategorization problem, since the locative verbs (such as zhàn, ‘stand’) and the appearance-disappearance verbs (such as lái, ‘come’), according to Huang (1987), lack the ability of taking complements that are structurally bigger than DP. Following the spirit of New Descriptivism (Hu 2018), this dissertation offers a fine-grained analysis of the classification of existential verbs (see Huang 1987 for more discussion), the form of the pivot indefinites, the predication type as well as the internal structure of the coda. Two findings are noticed according to the status of coda: if the coda is an individual-level predicate, the “yī-classifier-noun” pivot and the coda will form a “subject-predicate” structure under an empty DP-shell, which was licensed by the referential property from the extended projection of yī. In contrast, the stage-level predicate coda, is argued to be embedded in an applicative structure, which supply an abstract anchor function to the event expresses by the main sentence.

　　The second part of the dissertation focuses on the Chinese children’s acquisition of the indefinite NP as well as their usage of the related constructions. The goal of the research is twofold: it not only offers a fine-grained analysis of the acquisitional facts, but also justifies the syntactic assumptions which are proposed in the former chapter. Firstly, we have investigated the data from the CASS-Child and the CHILDES Database, which are composed by the longitudinal data of Beijing Mandarin-speaking children and the longitudinal data of the English-speaking children, respectively. The result shows that Chinese children and English children share a similar acquisitional path of the nominal elements (i.e., bare nouns and demonstratives are acquired individually at the beginning stage, which are followed by the direct conjunction of demonstratives and bare nouns, while the last stage is assigned to the “determiner-noun” form in English and the “numeral-classifier-noun” form in Chinese). Importantly, the current study draws special attention to the optional form “dem-noun”, and serves as a crucial argument in support of the Bidirectional Maturation Model proposed by Hu (2016). It is argued that previous acquisition theories involving unidirectional/bottom-up maturation hypothesis can hardly explain the relevant facts. Secondly, we have examined the timing of the acquisition as well as the usage frequency of the following constructions: the “numeral-classifier” construction without nouns, the “yī-classifier-noun” construction, and the normal “numeral-classifier-noun” construction. The different acquisition timings and usage frequencies of the three constructions not only justify the structural analysis as argued in the previous chapter, since Chinese children acquire those constructions in a rigid order which can never be reversed, but also count as an evidence supporting the left-branch analysis of the “numeral-classifier-noun” construction as proposed by He (2016), since the data from our research never exhibit a particular stage, in which Chinese children only produce “classifier-noun” construction without any numerals. Finally, we have also analyzed the data involving an indefinite subject produced by the Chinese-speaking children from both the CASS-Child and CASS-Infants Database. The result confirms Lee & Wu’s claim that Chinese children rarely produce ungrammatical sentences with an indefinite subject. Based on Wexler (1998), Chinese children’s acquisition of the indefinite NP at a very early age implies their correct setting and checking of the d-feature in the syntactic domain. Furthermore, in all the ungrammatical productions (such as *Yí-gè hóngsè ne?, ‘Where is a red (ball)?’), the indefinite subject is only used as an anaphoric element referring to an object which is absent in the current context. In other words, the anaphoric element stands for an old information. As a result, the indefinite subject still satisfies the “Given-New” information partition requirement of IS. The observed ungrammatical productions are then considered to be a result of the limited development of children’s cognitive ability.

　　Key Words: Definiteness Restriction; Indefiniteness Restriction; Syntax-Information Structure Interface Condition; d-feature; Coda; “yī-classifier-noun”; “numeral-classifier-noun”; Bidirectional Maturation Model