点击上方蓝字关注我们

《理论语言学五道口站》（2020年第34期，总第98期）“人物专栏”与大家分享本站采编人员王竹叶对Yoshio Endo教授进行采访的访谈录。Yoshio Endo，神田外语大学研究生院教授、北京语言大学语言学系国际教授委员会成员，曾在麻省理工学院担任访问学者。

本期访谈中，Yoshio Endo教授分别阐述了区分中心状语从句和边缘状语从句的决定因素、主句现象对状语从句的影响以及全称量词和阻断效应之间的关系等问题，同时，就句末语气词对句法制图的影响、局域性原则的作用和准据方法的本质等问题提出了自己的看法，对于我们研究汉语中相应的语言现象指明了思路，非常具有启发意义。

人物简介

Yoshio Endo教授

远藤喜雄，神田外语大学研究生院教授，曾在麻省理工学院担任访问学者。2006年，远藤喜雄教授在日内瓦大学获得博士学位（委员会成员：Luigi Rizzi（法兰西学院）、Ur Shlonsky（日内瓦大学）、Shigeru Miyagawa（麻省理工学院）），其博士论文于2007年由John Benjamins公司出版。目前，他的研究方向集中于句法制图框架下的语言变异，包括比较语言学、说话人之间的语言变化、方言变异、历时变异以及自闭症患者的语言研究等。

Brief Introduction

Yoshio Endo, a graduate school professor at Kanda University of International Studies. He was a visiting scholar at MIT in 2005. He received his Ph.D. from University of Geneva in 2006 (committee members: Luigi Rizzi (Collège de France, supervisor), Ur Shlonsky (U of Geneva), Shigeru Miyagawa (MIT)). The refined version of his Ph. D. thesis was published by John Benjamins in 2007. His current interest is linguistic variations in the framework of the cartography of syntactic structures, including comparative linguistics, inter-speaker variations, dialectal variations, diachronic variations, autism, etc.

访谈内容

01.

王竹叶：您曾讨论过中心状语从句和边缘状语从句的区别，要区分这两类从句，您认为有哪些特征可以起决定性作用呢？这些特征是否具有普遍性？

远藤喜雄教授：中心状语从句具有以下特点（i）表达前景信息，（ii）可能没有像话题、焦点这样与CP层相关的语篇表达；（iii）与主句的低层级成分结合在一起。而边缘状语从句（i）表达背景信息，（ii）可能包含像话题这样与CP层相关的语篇表达，并且读降调（Rizzi（1997）），（iii）与主句的高层级成分相关（参见Endo（2020b）和Endo & Haegeman（2019））。要想知道这些特征是否适用于汉语，你可能需要关注语篇成分的普遍性，比如话题成分。“A或B”是反话题的表达，而“A和B”在不同语言中都可以作话题。因此，在日语中，“A或B”这种表达后面不可以出现话题标记wa，但“A和B”后面可以（Endo（2014：176-177））。这一特性可用于考察汉语中的中心状语从句和边缘状语从句。要明白这一点，我们来看下面一个小插曲：我的一个研究生在分析翻译自日语小说的中文句子时，碰到了下面的句子，她不确定划线的主语应该位于括号内的状语从句中还是括号外的状语从句中，即主句中：

根据前面的论述，我们可以知道括号内的状语从句是中心状语从句，因为它不表达背景信息。所以，我让她用“A或B”和“A和B”这样的表达来代替划线的主语。她认为“A和B”听起来比“A或B”好。因为“A和B”可以作为话题表达，而“A或B”不可以。这说明划线的成分是话题。此外，由于中心状语从句中不允许出现话题成分，因此我推测划线成分应该是中心状语从句之外的话题成分。为了证实这一点，我进一步问她划线的主语是否读降调。她的回答是肯定的。从这些事实中，我们可以得出以下结论：她对上述句子的最初分析是错误的，划线主语其实是带括号的中心状语从句之外的话题成分。如果要想出一个汉语边缘状语从句，可以试着考虑具备以下特征的状语从句：表达背景信息、可以用“A和B”作主语、读降调。Kenichin Mihara曾研究过日语中一个与此相关的缺少时态的状语结构（参见Mihara（2015）)。Endo和Haegeman有关中心状语从句和边缘状语从句的研究借用了许多传统语法学家的观点，如Fujio Minami（1974）和Hisashi Noda。在研究汉语状语从句时，你们也可以借鉴传统语言学家的观点。 

02.

王竹叶：主句现象这一新的研究视野汇集了既有学者和后来学者的研究成果，学者们利用各种语言对主句现象（MCP）进行了大量分析，尤其关注了助词和一致性标记词、标句词和V2、以及允许出现主句现象的不同句型等问题，您能简要谈一谈主句现象对于分析状语从句的影响吗？

远藤喜雄教授：这是个很好的问题。想要弄清这一点，我们可以看一些具体的例子：如表现出主句现象（MCP）的边缘状语从句（peripheral adverbial clauses）。Shigeru Miyagawa（2012）发现，日语中masu/desu这样的敬语成分只出现在主句中，这正是主句现象的一个例子。事实上，边缘状语从句中可能出现masu/desu这样的敬语成分，而且总是出现于中心状语从句之前，这是因为边缘状语从句会和主句中一个更高层级的成分相关联，而且这一成分要高于中心状语从句（Endo and Haegeman（2019））。之前许多有关状语从句和主句现象的研究都是从语义的角度进行解释的。上述我们所看到的中心状语从句和边缘状语从句的先后关系可以强有力地说明，状语从句和主句现象的区别本质上是句法层面的。如果想了解汉语中的这种句法效果，可以先去找找从日语小说中翻译成中文的句子，可以是带有敬语成分masu/desu的状语从句，也可以是不带敬语成分的句子。然后试着把这些状语从句用到一个句子中，你会发现，边缘状语从句在中心状语从句之前而非之后。除此之外，还可以利用中心状语从句和边缘状语从句分析约束、辖域等句法现象，然而这些方面还有待探究。Nobuko Hasegawa（2007）研究了日语中的主句现象。

03. 

王竹叶：日语和汉语都有丰富的句末语气词。有时候句末语气词可以显性表现为特定的功能中心语。从理论和实证两方面来看，研究句末语气词对句法制图有何贡献呢？

远藤喜雄教授：从实证方面来看，还有很多研究话题有待发掘。比如，日语句末语气词和德语的语篇助词之间的关系非常有趣。言外之意的语用强度是根据德语情态助词和日语句末语气词的意义来进行调整和排列的（Endo(2020a)）。此外，Josef Bayer和Hans-Georg Obenauer（2011）研究表明，德语中各类反问句的构成都有语篇助词。同样，日语句末语气词也有类似的现象（Endo（2020a）），你可以核查汉语中是否有类似的现象，进而探究其普遍性。非限制性关系从句也与此相关，和边缘状语从句一样，日语中特殊的非限制性关系从句也可能在其CP层中出现类似话题的成分。更准确地说，日语中由to ‘that’ iu ‘say’引导的同位语类型的非限制性关系从句中可能出现句末语气词，但边缘状语从句中不会出现，因为句末语气词是主句现象，你或许想问，主句现象在非限制性关系从句是否具有普遍性。事实上，德语的非限制性关系从句中可以容纳语篇助词这种和CP层相关的成分，这一点和句末语气词非常相似（Endo（2020a））。接下来你可能会思考，汉语的非限制性关系从句中是也可以出现句末语气词。从理论层面上来看，句末语气词对绘制详细的语气成分句法图有很大帮助。具体而言，通过思考不同类型句末语气词在Cinque的层级结构中的位置，我们可以判断Cinque的层级结构是否具备普遍性。如果你发现汉语句末语气词并不适用于Cinque的层级结构，你可以绘制出新的句法图来说明这其中参数化的问题，这对理论语言学将会是一项新的贡献。除此之外，你也可以就句末语气词和其他语法操作的相互作用提出假设。例如，当带有[+N]特征的句末语气词出现时，这个名词性的句末语气词就满足了主语准据，语法上的主语为了能够进入到vP之上的否定辖域范围内，就保留在了VP内。在缺乏名词性句末语气词的时候，语法上的主语为了满足主语准据，会位移到主语位置上，这样主语就可以在否定辖域之外得到解读（2007, 2012）。Yukiko Ueda（2007）研究提出日语中的句末语气词有决定句子类型（Rizzi（1997）所说的语力）的作用。期待在不久的将来可以看到汉语句型的句法制图研究成果。

04.

王竹叶：日语全称量词daremo ‘everyone’会阻碍wh移位，但同样作为全称量词的subeta-no gakusei ‘all students’却不会阻碍这一操作，全称量词和阻断效应之间存在着怎样的关系呢？

远藤喜雄教授：这是个有趣的问题。Endo（2007，第三章）指出，量词类型的本质和语篇特点之间存在密切的关系。具体来说，有些量词可以是话题性的，而有些本质上则是反话题的，也就是说，是否能够以话题助词wa作为后缀是它们的不同之处。daremo ‘everyone’后面不能以wa作为后缀，它是一个真量词。Rizzi（2004）将wh成分也划分为量词，而根据相对化最简条件（Relativized minimality），wh移位会受到daremo ‘everyone’的阻碍。与此相反，subeta-no gakusei ‘all students’可以用话题助词wa作为后缀，或许是因为这个表达的语义内容更容易从前文中获得理解，即话题。把subeta-no gakusei ‘all students’这一表达划分到话题类型中，这使其不同于量词类型。由于量词型和话题型属于不同的范畴（Rizzi 2004），所以不存在RM冲突。我认为这种由语篇为导向的RM在汉语中也值得研究，我前面提到的鉴别话题成分的手段或许会有帮助。Masatoshi Honda（2019）研究的是日语中的焦点/量化成分，我期待能有汉语量词类型成分的相关研究成果，尤其是涉及RM相关的研究。关于这个话题，我们或许会注意到还有两个不同的主语位置，一个位置由指称或话题性名词占据，另一个由非指称性或非话题性名词占据。Anna Cardinaletti（2004）的罗曼语研究和Kazuma Fujimaki（2011）的日语研究中都有探讨这一点。我也很期待在句法制图框架下，不同主语位置的汉语句法研究能有新成果。 

05.

王竹叶：根据乔姆斯基的语段理论及后来的研究成果，我们可以认为，句子结构建立于包括CP和及物vP在内的语段上。局域性在不同的语段中是如何发挥作用的？它在制图研究中扮演着什么样的角色？

远藤喜雄教授：这是一个有关技术的问题。关于局域性至少有两种方法：派生法和表征法（derivational approach vs. representational approach）。如果采用Rizzi式的表示法，可以利用跨语段的语链核查局域性效应，通过这种方法，我们可以假设局域性语链形成于语段之中（见Rizzi（1986）有关局域性语链的研究）。采用派生法，可以在推导过程中鉴别局域性，在这个过程中，每一次移位都会检查是否遵循了局域性原则。Endo（2007：105-106）曾讨论过局域性的派生法和表示法的区别。Masako Maedo（2014）系统地采用了派生法分析局域性。希望汉语语言学家能在这方面研究有所建树。

06.

王竹叶：有主语准据，却没有宾语准据，这决定了有的主语可以移位至主句，有的却不可以。您能简要解释下准据方法的本质是什么吗？

远藤喜雄教授：准据位置的其中一个特性是在语篇特征中发现的，如话题、相关性、焦点等，这些都是在CP层发现的。虽然主语位置是在IP层，但它位于IP层的最高位置，所以我们可以这样假设：有些语篇特性也可以在主语位置上找到。与此相比，宾语在IP层中处于很低的位置，所以宾语不具有语篇特性，也就不存在宾语准据。然而，Adriana Belletti的研究表明，在vP区域周围存在一个较低的焦点位置，我们或许可以预测存在带有低位焦点成分的宾语准据。有关准据冻结的新的研究方向非常令人期待。Koichiro Nakamura（2020）在有关日语中的低位焦点成分研究中，还没有发现准据的相关特性。希望研究人员可以继续在汉语中探索这种可能性。

Interview

01.

Zhuye Wang: As you have discussed about the distinction between central adverbial clauses and peripheral adverbial clauses, in your opinion, what characteristics can decide the adverbial clauses as central adverbial clause or peripheral adverbial clauses? And do the characteristics display universality?

Prof. Yoshio Endo: Good question. Core adverbial clauses (i) express foreground information, (ii) may not contain CP-related discourse items like topic, focus, etc. and (iii) are integrated into and are associated with low elements of the matrix clauses. In contrast, peripheral adverbial clauses (i) express background information, (ii) may contain CP related discourse items like topic, which is pronounced with falling intonation (Rizzi (1997)), and (iii) are associated with high elements of the matrix clauses (cf. Endo (2020b) and Endo and Haegeman (2019)). If you want to know how these points fit into Chinese, you might pay attention to universal characterizations of discourse elements, like topic elements. For instance, the expression “A or B” is an anti-topic item, in contrast to the expression “A and B”, which can be a topic item cross-linguistically. Thus, the expression “A or B” cannot be suffixed by the topic particle wa in Japanese, which is possible with “A and B” (Endo (2014: 176-177)). This property can be utilized in exploring central vs. peripheral adverbial clauses in Chinese. To see this, look at the following episode: when one of my graduate students was analyzing Chinese sentences translated from a Japanese novel, she faced the following sentence in Chinese and was not sure whether the underlined subject would be located within the bracketed adverbial clause or found outside adverbial clause, i.e. in the matrix clause:

From the previous discourse, I understand that the bracketed adverbial clause is a central adverbial clause because it does not express background information. So, I asked her to replace the underlined subject with the expression “A or B” and “A and B”. Her reaction was that “A and B” sounds better than “A or B”. Because “A and B” can be a topic item while “A or B” is an anti-topic item, it suggests that the underlined element would be a topic item. In addition, because topic elements do not appear in central adverbial clauses, my conjecture was that the underlined element would be a topic element appearing outside the bracketed central adverbial clause. To confirm this, I further asked her whether or not the underlined subject would be pronounced with falling intonation. Her answer was “yes”. From these facts, we reached the conclusion that her first analysis of the Chinese sentence above was wrong and the underlined subject would be a topic element appearing outside the bracketed central adverbial clause. If you want to come up with a peripheral adverbial clause in Chinese, you can try to come up with a context where the above adverbial clause expresses background information with the subject “A and B” pronounced with falling intonation. A related tenseless adverbial construction in Japanese is pursued by Kenichin Mihara (cf. Mihara (2015)). Incidentally, Endo and Haegeman’s central vs. peripheral adverbial clauses borrows many insights from traditional grammarians like Fujio Minami (1974) and Hisashi Noda (1989). In exploring adverbial clauses, you might also work together with traditional grammarians in China.

02.

Zhuye Wang：As Main Clause Phenomena is the new horizon which brings together work by both established and up-and-coming scholars, who present analyses for a wide range of MCP, from a variety of languages, with a particular focus on particles and agreement markers, complementizers and verb second, and the licensing of MCP in different types of clauses, could you briefly talk about the influence of main clause phenomena on the analysis of adverbial clauses?

Prof. Yoshio Endo: Nice question. To see this point, let us consider some concrete cases where (i) peripheral adverbial clauses show MCP. Shigeru Miyagawa (2012) found that the Japanese honorific elements masu/desu may only appear in the main clause, an instance of MCP. In fact, peripheral adverbial clauses may contain the honorific elements masu/desu and always precedes rather than follow central adverbial clauses because peripheral adverbial clauses are associated with a higher element in the matrix clause than central adverbial clauses (Endo and Haegeman (2019)). Many of the previous studies of adverbial clauses and MCP are accounted for in semantic terms. The precedence relation of central vs. peripheral adverbial clauses we saw above is a very strong piece of evidence showing that the distinction is syntactic in nature. If you want to know this syntactic effect in Chinese, you might try to find Chinese sentences translated from Japanese novels with and without the honorific elements masu/desu in adverbial clauses, and then try to use these adverbial clauses at the same time in a single clause. Then, you might realize that the peripheral adverbial clause precedes rather than follow the central adverbial clause. In addition, you might also make use of central adverbial clause and peripheral adverbial clause in analyzing syntactic phenomena like binding, scope, etc., which has not been explored yet. MCP in Japanese is explored by Nobuko Hasegawa (2007).

03.

Zhuye Wang: Japanese and Chinese have a fruitful inventory of sentence final particles. Sometimes SFPs are taken to be the overt realization of certain functional heads. So, how can studies of SFPs contribute to Cartography, both on the theoretical and empirical sides?

Prof. Yoshio Endo: Excellent question. On the empirical side, there are many topics that remain unexplored. For instance, there is an interesting connection between SFPs in Japanese and discourse particles in German, where the pragmatic strength of the illocutionary force is modified and ordered according to the meaning of modal particles in German and SFPs in Japanese (Endo (2020a)). In addition, as Josef Bayer and Hans-Georg Obenauer (2011) shows that various types of rhetorical questions are formed with discourse particles in German and the same effects are detected with Japanese SFPs (Endo (2020a)). You can check whether the same holds for Chinese to see the universality. Furthermore, non-restrictive relative clauses are also relevant here. Just like peripheral adverbial clauses, special non-restrictive relative clauses in Japanese may host elements in the CP zone like a topic element. To be more specific, SFPs may appear in non-restrictive relative clauses of the appositive type in Japanese headed by to ‘that’ iu ‘say’ may house a SFP, which is impossible with peripheral adverbial clauses may not. Because SFPs are MCP, you might want to ask whether the MCP in non-restrictive relative clauses is universal. In fact, non-restrictive relative clauses in German may also house CP related elements like discourse particles, which is very similar to SFPs. (Endo (2020a)). Then, it might be interesting to ask yourself whether nonrestrictive relative clauses of any kind allow for SFPs in Chinese as well. On the theoretical side, SFPs can make a great contribution to drawing a detailed syntactic map of mood-elements. To be more specific, by asking where various types of SFPs fit into Cinque’s hierarchy, we can check how far Cinque’s hierarchy is universal. In case you find SFPs that do not fit into Cinque’s hierarchy in Chinese, you can create a new map to show that some parametrization is required, which would be a new contribution to the theoretical linguistics. In addition, you can also expect some interactions between SFPs and other grammatical operations. For instance, in the presence of a SFP with the feature [+N], the subject criterion is satisfied by this nominal SFP, and then the grammatical subject remains within VP to be in the scope of negation above vP. In the absence of a nominal SFP, the grammatical subject has to move into the subject position to satisfy the subject criterion, and then the subejct is interpreted outside the scope of negation (2007, 2012). In addition, SFPs in Japanese plays a role in deciding clause types or Rizzi’s (1997) Force, which is explored by Yukiko Ueda ((2007). I am expecting works on clause-typing in the cartography of Chinese syntactic structures in the near future.

04.

Zhuye Wang: Japanese universal quantifier daremo ‘everyone’ blocks wh-movement, while the universal quantifier subeta-no gakusei ‘all students’ does not. What’s the relationship between the universal quantifier and blocking effects? 

Prof. Yoshio Endo: Interesting question. Endo (2007, Chapter 3) shows that there is a close relation between the nature of quantifier types and the discourse properties. To be more specific, some quantifiers are topic and others are anti-topic in nature, i.e. they differ with respect to whether they can be suffixed by the topic particle wa or not. Because daremo ‘everyone’ cannot be suffixed by the topic particle wa, it is a genuine quantifier. Because wh-elements are also classified into a quantifier (cf. Rizzi 2004), wh-movement is blocked by daremo ‘everyone’ by relativized minimality (RM). In contrast, subeta-no gakusei ‘all students’ can be suffixed by the topic particle wa perhaps because a rich semantic content of this expression can make it easier to be interpreted to appear in the previous discourse, i.e. topic. Then, the expression subeta-no gakusei ‘all students’ is classified into the topic type, which is distinct from the quantifier type. Because the quantifier type and the topic type belong to different class (cf. Rizzi (2004)), there is no RM violation. I believe that this type of discourse-oriented RM is worth exploring in Chinese as well, where the techniques to identify a topic element that I mentioned above might be of some help. Masatoshi Honda (2019) is exploring focus/quantificational elements in Japanese and I am expecting works coming out in Chinese studies on elements of the quantifier type, especially in relation to RM. Related to this topic, it might be interesting to notice that there are two different subject positions as well, one of which is occupied by referential or topical nouns and the other of which is occupied by non-referential or non-topical nouns. This point is explored by Anna Cardinaletti (2004) in Romance and Kazuma Fujimaki (2011) in Japanese. I am expecting works coming out in the cartography of Chinese syntactic structures of distinct subject positions. 

05.

Zhuye Wang: According to Chomsky’s phase theory and the subsequent work, it is claimed that clause structure is built up in phases including CP and transitive vP. How does locality work in different phases? What roles does locality play in Cartography?

Prof. Yoshio Endo: Nice technical question. There are at least two approaches to locality: derivational approach vs. representational approach. If you choose Rizzi-style representational approach, you might make use of chain across phases to check locality effect, where you can assume a local chain is formed within a phase (see Rizzi (1986) on local chain.) If you choose a derivational approach, you can calculate locality in the course of derivation, where each movement step checks whether locality principle is obeyed or not. The comparison between derivational vs. representational locality is discussed in Endo (2007: 105-106). Masako Maeda (2014) systematically pursues derivational approach to locality. I expect works on this topic to come out by by Chinese linguists.

06. 

Zhuye Wang: There is subject criterion but no object criterion, which determines that what kind of movement of the subject into the matrix clause is possible and what kind of movement is impossible. What’s the nature of the criterial approach?

Prof. Yoshio Endo: Good point. One of the properties of the criterial positions is found in discourse features like topic, aboutness, focus, etc, all of which are typically found in the CP zone. Although the subject position is in the IP zone, it is the highest position in the IP zone and thus we can expect some discourse properties is also found in the subject position as well. In contrast, the object is very low in the IP zone and thus we might not expect discourse properties with the object, i.e. no object criterion. However, Adriana Belletti shows that there is a low focus position around the vP area. Then, we might expect object criterion with low focus elements. This is an exciting domain where new aspects of criterial freezing are expected. Koichiro Nakamura (2020) exploring low focus elements in Japanese, where no criterial properties has not been explored yet. I expect researchers to explore this possibility in Chinese. 

参考文献

本文版权归“理论语言学五道口站”所有，转载请联系本平台。

编辑：马晓彤 王竹叶 訾姝瑶 

排版：马晓彤 安镜儒

审校：陈旭 王丽媛