【作者】 贾光茂；

【导师】 丁言仁；

【作者基本信息】 南京大学 ， 外国语言学及应用语言学， 2011， 博士

【摘要】 岛屿限制现象指的是某些句子结构如主语、复杂名词短语(如关系从句和名词补语从句)和状语等结构中的成分不能移位。生成语言学主要运用自足的句法原则即邻接原则来描述并解释这一语言现象,该原则规定移位不能超越两个以上界限节点。然而,句法中有许多邻接原则无法解释的反例：目的状语从句和呈现型关系从句中的成分似乎可以移位,并非所有补语从句中的成分都可移位。还有的岛屿现象普遍语法没有涉及,如双宾构式中接受者不能移位,有的岛屿限制与语篇有关。这些现象却似乎可以用背景构式中的成分不能移位这一理论结合加工难度来解释。背景化程度低的目的状语从句和呈现型关系从句中的成分移位要比背景化程度高的时间状语从句和关系从句中成分移位所形成的疑问句可接受程度高,补语从句是否构成岛屿也由其背景化程度高低决定。双宾构式中的接受者不能移位也是由于其背景化程度高。该理论符合涌现论的基本观点,即语言现象来源于大脑中非语言的因素及其互动,对于岛屿限制现象似乎比邻接原则具有更强的描写和解释能力,在二语习得领域,探讨邻接原则对二语学习者是否可及一直是研究的热点。因为这种隐性的句法知识不可能来自学习者的经验,违反岛屿限制的句子在实际的语言输入中极其少见；并且对于二语学习者来说,如果他们的母语不涉及移位现象,那么这种知识就只能借助先天的普遍语法来解释。然而,迄今为止还未有二语习得研究以涌现理论为指导。涌现论和先天论关于岛屿限制现象提出了不同的假设。受此争辩启发,本次研究通过实证方法来调查中国英语学习者的岛屿限制知识,以检验哪种理论假设能构更好的描述和解释这种语法知识。此次研究的调查对象共有四组：1)中低水平组,受试均为来自南京某高校的一年级67名非英语专业大学生；2)中高水平组,成员均为来自同一高校的54名三年级英语专业大学生；3)高水平组,成员为南京某高校英语专业文学文化方向的16博士生；4)本族语组,成员为11名南京某高校海外教育学院的美国留学生。所有受试均非语言学专业。所使用的研究工具包括语法判断测试卷、否定测试和对划线部分提问任务。语法判断题用来调查受试的岛屿限制知识,由36句实验句和28句控制句组成。实验句包括四种类型的岛屿限制句,分别为状语从句、关系从句、补语从句和双宾语句。其中状语从句又可分为目的、时间、和原因三类；关系从句分为呈现型和非呈现型两类；补语从句根据主句的动词分为三类,双宾句涉及接受者、主题、目标移位三类。部分句子根据普遍语法理论和涌现论都不可接受,还有部分句子普遍语法理论认为不可接受,但涌现论却认为可以接受。否定测试根据上述实验句修改而成,用来调查学习者的句子信息结构知识并探讨岛屿限制知识和信息结构特点之间的关系。对划线部分提问任务也由状语从句、关系从句、补语从句组成,以映证语法判断结果。本研究获得了以下发现：1.中国英语学习者认为状语从句、关系从句和部分补语从句中的成分移位会使句子不可接受,并且知道不同类型的岛屿限制句可接受程度不同。他们认为涉及目的状语从句的岛屿限制句比涉及其他状语的更可接受,涉及呈现型关系从句的岛屿限制句比涉及其他关系从句的更可接受,涉及补语从句中成分移位的句子可接受程度因主句的动词不同而不同,涉及接受者移位的双宾句不可接受。2.英语本族语者同样知道状语从句、关系从句和部分补语从句,并且知道不同类型的岛屿限制句可接受程度不同。他们也认为涉及目的状语从句的岛屿限制句比涉及其他状语的更可接受,涉及呈现型关系从句的岛屿限制句比涉及其他关系从句的更可接受,涉及补语从句中成分移位的句子可接受程度因主句的动词不同而不同,涉及接受者移位的双宾句不可接受。3.随着英语水平的提高,受试对岛屿限制现象更加敏感。对于中低英语水平受试,高英语水平受试认为以上各类岛屿限制句的可接受程度更低。4.中国英语学习者岛屿限制知识和句子信息结构特点知识之间具有显著的相关性,说明信息结构特征至少可以部分解释岛屿限制知识。因为邻接原则并不预测受试会认为不同类型的状语从句、关系从句和补语从句可接受程度不同,所以本次研究结论不支持邻接原则而支持背景构式中成分不能移位的假设。本研究为涌现论和构式语法提供了来自二语学习者的实证依据,对语言学界解决形式与功能两派的争论有重要意义。从二语习得角度来说,传统的关于岛屿限制现象的研究大都以普遍语法为指导,本研究受BCI理论启发,为二语习得研究提供了新的视角,可以深化我们对中介语特征和本质的认识,促进二语习得理论发展。从外语教学角度来说,本研究结果将为中国的英语教师和学习者探索和认清外语学习规律提供参考,他们可以借鉴本次研究的某些成果以提高教学效率。更多还原

【Abstract】 The term "island constraints" refers to the grammatical phenomena that elements in a construction such as a subject clause, an adjunct, or a relative clause (RC) cannot move out of the construction. In generative linguistics, the phenomena are accounted for by the subjacency principle, which means that no constituent can move across more than one bounding node at a time. The subjacency principle is claimed to be autonomous, domain-specific and innate.Recently, the subjacency principle has been challenged by an emergentist theory: back-grounded constructions are islands (BCI). According to BCI, island constraints are said to be determined by information structure properties of constructions. For instance, non-presentational RCs, and time adjuncts are predicted to be islands because they are back-grounded in a sentence; presentational RCs, and purpose or reason adjuncts are sometimes not islands because they are fore-grounded.In SLA, a large number of studies have been conducted to test second language learners’knowledge of island constraints within the generative framework to see whether they have access to Universal Grammar (UG). However, no SLA studies have been based on BCI.Inspired by the debate between the nativist subjacency principle and the emergentist BCI hypothesis, this study examined the Chinese EFL learners’knowledge of island constraints in order to see which theory best accounts for such knowledge.The subjects included both Chinese EFL (English as a foreign language) learners and English natives. They were asked to take a grammaticality judgment test, which consisted of some sentences whose island status UG and BCI do not agree upon. They also took a negation test and completed a question formulation task.The study yielded the following results:1) Chinese EFL learners know that adjuncts, RCs, and some complement clauses are islands. More importantly, they treat the subtypes of the islands differently. They think that extractions from purpose clauses are more acceptable than those from time and reason adjuncts, those from presentational RCs are more acceptable than those from non-presentational RCs, those from factive or manner-of-speaking verb complements are more acceptable than those from bridge verb complements, and those of the recipient argument in active ditransitives are unacceptable.2) Native English speakers consider adjuncts, RCs, and some complement clauses as islands and treat the subtypes of the islands differently. Similar to Chinese EFL learners, they rendered extractions from purpose clauses, presentational RCs, and factive or manner-of-speaking verb complements more acceptable than those from reason and time adjuncts, non-presentational RCs and bridge verb complements respectively. They also consider extractions of the recipient argument in active ditransitives unacceptable.3) The Chinese EFL learners’sensitivity to the island-hood of adjunct clauses, RCs, sentential complements, and recipient arguments increases as their language proficiency rises, implying that L2learners can acquire the intricate knowledge if their English proficiency is high enough.4) The Chinese EFL learners’ knowledge of island constraints is related to their knowledge of information structure of sentences, suggesting that the grammatical knowledge is derived from non-grammatical factors.Although UG predicts that language learners should have knowledge of island constraints, it does not predict that they treat different types of adjuncts, RCs, and complements differently. The fact that Chinese EFL learners and native English speakers tend to accept extractions ruled out by UG but predicted by the BCI hypothesis lends support to BCI, and thus adds weight to BCI or emergentist theories but pose threat to UG and UG-based SLA studies. The findings are of value in improving the description and explanation of L2learners’ linguistic competence.更多还原