人物专栏 | Éva Katalin Dékány博士访谈录

点击上方蓝字关注我们

编者按

《理论语言学五道口站》（2021年第42期，总第176期）“人物专栏”与大家分享本站采编人员王平对Éva Dékány博士进行采访的访谈录。Éva Dékány，匈牙利布达佩斯匈牙利语研究中心高级研究员，匈牙利青年学会成员。她的主要研究领域是句法与形态句法，尤其是匈牙利语与其他芬兰-乌戈尔语族语言的句法和形态句法。

本期访谈中，Éva Dékány博士首先分享了她选择了语言学专业的原因，随后对于生成语法框架下的汉语研究给出了自己的建议，并详细说明了最简方案与句法制图的关系，最后阐释了一些具体的语言学现象，如DP以及名词性成分。

人物简介

Éva Katalin Dékány博士

Éva Katalin Dékány，匈牙利布达佩斯匈牙利语研究中心高级研究员，匈牙利青年学会成员，于2012年获得挪威特罗姆瑟大学博士学位。研究领域为句法与形态句法，尤其是匈牙利语与其他芬兰-乌戈尔语族语言的句法和形态研究。主要研究兴趣包括名词性短语与附置词短语的功能序列、左缘结构。最近于Springer出版了一部关于匈牙利语DP研究的专著：The Hungarian Nominal Functional Sequence。

Brief Introduction

Éva Katalin Dékány is a senior researcher at the Hungarian Research Institute for Linguistics in Budapest and a member of the Hungarian Young Academy. With a PhD in theoretical linguistics from the University of Tromsoe (2012), she specializes in syntax and morpho-syntax, especially in Hungarian and other Finno-Ugric languages. Her main areas of interest include the functional sequence of nominal and adpositional phrases as well as the structure of the left periphery. She has recently published a monograph（The Hungarian Nominal Functional Sequence） on the Hungarian DP with Springer.

访谈内容

01.

王平：您为什么对转换生成语法感兴趣，为什么选择语言学作为您的专业？

Éva Dékány博士：缘起是我大学的时候选了英语和语言学作为专业。我们的课程要求学生必须学习转换生成语法框架下的短语句法和小句句法。其实一开始我并不感兴趣，但后来继续选修高级句法课程后，我逐渐感兴趣了。我们当时每周都有句法学的作业，这些作业难度适中：颇具挑战性，但认真思考并灵活运用课堂知识，是可以完成的。完成这些作业激起了我对句法学的热情。我被复杂的语言学问题深深地吸引了，于是我选择了理论语言学作为第二学位并且继续攻读该方向的博士学位。

转换生成语法吸引我的原因在于它利用科学的方法研究语言。研究时不能只关注数据，还需要提出可测试和可反驳的假设。但只提出假设还不够，你需要去探索更深层次的原因和结果。换句话说，生成语法也是严谨的科学。

02.

王平：众所周知，与印欧语系的语言相比，汉语的形态变化更少。汉语有很多独特的语言现象。例如，汉语中独特的间接宾语现象。不只是施事、受事等中心语义角色可以做宾语，工具、场所等非中心语义角色也可以做宾语。对于这一现象，中国学者引入轻动词并对其语法操作进行解释，例如，转换、移位。然而，有人认为这是对生成语法的滥用。甚至有人认为不适合在生成语法框架下研究汉语，您如何看待这种现象？

Éva Dékány博士：在我看来，认为一个框架X不适合研究一种语言Y或者现象Z，这种看法是欠妥的。所有语言和现象都可以在任何框架下研究。当然，并不是每个框架都能对每种现象做出同样有洞察力的分析。当一个框架不能一次次地提供有洞察力的分析时，我们则需要修改或者摒弃该框架。

至于轻动词，很多语言用它来形成复杂的谓词。日语和英语中就有许多这样的例子。生成语法非常重视语言对比：在分析语言Y中的现象X的过程中获得的灵感可以用来分析语言W中的现象Z。我认为，只要分析能得出正确的预测，并且避免一个假设解释一种语言的这种偏见，那么可以假设汉语中存在轻动词。

03.

王平：制图研究属于生成语法的一种，然而由于激进制图理论的影响，句法树变得异常复杂，使得许多追随乔姆斯基的语言学家认为制图研究违反了最简方案所提倡的最简主义。在您看来，句法制图是否违反了生成语法所倡导的最简主义？您认为激进制图学与最简主义之间的关系是什么？

Éva Dékány博士：要在普遍生成语法与制图之间划分一个清楚的界限是不可能的，因为存在非乔姆斯基生成语法框架，例如词汇功能语法（LFG）。在乔姆斯基生成语法框架下，我们聚焦于绘制功能投射的层级图（即：功能序列），以及层级结构不同位置间的依存关系。从这个意义上来说，乔姆斯基语法是制图的核心。所谓的“激进制图”学派假设一个核心有且仅有一个特征（Cinque & Rizzi 2010）。这自然而然的就导致了功能结构的激增。然而，即使不支持激进制图理论，也有可能成为优秀的最简主义句法学家。乔姆斯基本人就是一个很好的例子，比如说他从来就没有接受过意大利的制图学者所提出的层级结构。而且有些可行的分析放弃了小句中的FocP 和TopP等与信息结构相关的投射。因此，最简主义很大可能并不相当于激进制图理论。

激进制图的特殊之处在于它认为词库包含大量不同的功能中心（因为一个特征一个核心原则）。然而最简主义主要关注狭义句法而不是词库。最简主义尝试利用最少的句法模型、原则、操作以及表征层面，从而消除句法框架中不必要的假设。因此，最简主义与“激进制图”之间并没有明显的对立。在我看来，激进制图学家与温和/保守制图学家之间的争论是心理词库的内容而非狭义句法操作的属性。

04.

王平：DP假说出现于管约论时期。该假说认为名词性表达受到以限定词D为核心的功能范畴的扩展，名词短语在DP结构中充当核心成分D的补足语。那么，在匈牙利语中DP是如何实现表征的呢？

Éva Dékány博士：Anna Szabolcsi早期针对匈牙利名词短语所做的研究，特别是有关定冠词和领属成分的研究，在很大程度上支持了新兴的DP假说（Szabolcsi 1983, 1994, 等）。引用当时学术界的一句话来说就是，“所有致力于匈牙利语NP研究的人，都受到了Szabolcsi的影响”。Szabolcsi对于匈牙利语DP/NP研究的贡献可见一斑。在匈牙利语中，DP域（DP-zone）内部包含多个不同的投射。其中居于最下层的是DeixP，它在指示词的位置进行合并（Den Dikken & Dékány 2018, Dékány 2021）。当DeixP受到包含定冠词的DP的统治时，短语指示词（Phrasal demonstratives ）从DeixP的Spec位置移到DP的Spec位置；当DP中不包含定冠词时，指示词将从核心Deix移位至核心D。DP的“逃生舱”（‘escape hatch’）位置被带有属格标记的领有者占据：这些领有者可以从名词性投射中被提取出来。根据Farkas & Alberti (2016) 及 Alberti & Farkas (2018)的最新发现，在动词名物化（deverbal nominalizations）的过程中，基本动词的旁格修饰语可以出现在与格领有者之前。至于这些具体是如何在DP中实现的，仍需进一步的研究。

05.

王平：Hauser等(2002) 认为递归是普遍语法的一部分，也就是人类语言官能的一部分。在您看来这种说法合理吗？能谈谈原因吗？

Éva Dékány博士：语法递归指我们可以使用一组有限的基本构件（语素和单词）来构建理论上无限的结构（短语和从句）。这是人类语言的核心，也是人类语言区别于动物交流的核心所在。允许语法递归的操作是合并。目前最大的问题是“合并”究竟是为人类语言所独有的（在这种情况下，它是人类语言官能的一部分），还是其他认知功能也可以使用“合并”这一操作（如果是，那么“合并”是一个独立于某一专门领域的通用操作，而不是语言能力的一部分）。在生成语法看来，合并是语言特有的，因此是人类语言官能的一部分。在我看来，这种观点在某种程度上是基于一种根深蒂固的信念，而不是经验/实验证据。若想对这一问题进行更加深入的研究，语言学家和认知科学家必须齐心协力。若要圆满地解决这一问题，我们需要同时具备语言学和认知科学知识的团队。

范畴递归则是另一回事。范畴本身可以进行嵌套。比如DP可以嵌入DP，CP可以嵌入CP。Hauser等(2002)尚未介绍这种类型的递归，他们只关注了语法递归，也就是“合并”。是否每种语言都存在范畴递归，这一问题大约15年前就引起了争论。比如，Everett (2005)认为皮拉罕语（Pirahã）中不存在范畴递归。但在我看来，Nevins, Pesetsky & Rodrigues (2009)等文献已经很好地反驳了Everett的主张。迄今为止，我们尚未发现令人信服的某种语言中缺乏范畴递归的案例。但正如我前面说到过的，这个问题与Hauser等(2002)的主张并没有多大的联系。

06.

王平：我们都知道句法学是语言学的分支，研究组词造句的规则；而语法是决定语言语法结构的一套规则，那么句法和语法之间的区别又是什么呢？

Éva Dékány博士：“语法”指的是有结构的语言的所有层面，语法包括语音学、形态学、句法学和语义学。这就意味着语法和句法是包含关系。

07.

王平：作为一个语法范畴，名词性成分指的是可以起名词作用的词汇组合。名词性成分中的其他词汇赋予了其名词核心更多细节，并使其更加具体。名词性成分可以充当动词的主语或者宾语。依您之见，名词性短语与名词性从句之间的关系是什么？

Éva Dékány博士：名词性从句具有小句的内部结构（以动词为核心）和名词短语的外部分布。这种情况可以以两种方式出现。一是子句修饰空名词（零名词或隐性代词），作为其补语或附加语。在这种情况下，作为短语核心的空名词成分决定了名词分布。二是“混合投射”，即先从动词功能核心投射小句，然后转向名词功能中心的投射。根据转向发生的位置，投射可变得更具动词性或更具名词性。在“混合投射”的情况下，我们需要处理名物化。但是第一种情况不存在名物化的过程。

08.

王平：现代匈牙利语关系从句除了包含英语中的wh-REL成分和德语中的dem-REL成分外，还包含第三种关系代词。Bacskai-Atkari & dékány(2021)指出:“现代匈牙利语关系算子包含两种标记，因此将被称为DEM-wh-REL”。您能从形态学的角度解释一下匈牙利语中的dem-wh-REL现象吗？

Éva Dékány博士：在我与康斯坦茨大学的Julia Bacskai-Atkari合作的研究中， “wh-REL”这个术语表示与疑问代词形式相同的关系代词（例如，英语中的Who was late? 和the girl who was late）。我们认为“dem-REL”是语法化了的关系代词，因此与指示代词有相同的形式。这是两种最常见的跨语言模式。匈牙利语的关系代词在形态上很复杂。它们由与疑问代词形式相同的词干和由指示代词语法化的前缀组成。然后由这个前缀对关系代词和疑问代词进行正式区分。我们将匈牙利语中的关系代词称为dem-wh-REL，是为了反映匈牙利语中关系代词包括两部分：一部分与wh-词项有关，另一部分与指示词有关。这种模式并不具有普遍性，但它也不是匈牙利语所独有的，在希腊语中也能观察到。

参考文献

[1] Alberti, Gábor, and Judit Farkas. 2018. Modification. In Gábor Alberti and Tibor Laczkó (eds.), Syntax of Hungarian. Nouns and Noun Phrases, 2, 775–896. Amsterdam: Amsterdam University Press.

[2] Bacskai-Atkari, Julia and Éva Dékány. 2021. Cyclic changes in Hungarian relative clauses. In Thórhallur Eythórsson and Jóhannes Gisli Johnsson (eds.),  Syntactic Features and the Limits of Syntactic Change. Oxford: Oxford University Press. 40–63.

[3] Cinque, Guglielmo and Luigi Rizzi. 2010. The cartography of syntactic structures. In Bernd Heine and Heiko Narrog (eds.), The Oxford handbook of linguistic analysis. Oxford: Oxford University Press. 51–65.

[4] Dikken, Marcel den and Éva Dékány. 2018. A restriction on recursion. Syntax, 21(1): 37–71.

[5] Everett, Daniel L. 2005. Cultural constraints on grammar and cognition in Pirahã. Current Anthropology, 46(4): 621–646.

[6] Farkas, Judit, and Gábor Alberti. 2016. Positions for oblique case-marked arguments in Hungarian Noun Phrases. Jezikoslovlje, 17 (1–2): 295–319.

[7] Hauser, Marc D., Noam Chomsky and W. Tecumseh Fitch. 2002. The faculty of language: What is it, who has it, and how did it evolve? Science, 298. 1569–1579.

[8] Nevins, Andrew and David Pesetsky and Cilene Rodrigues. 2009. Pirahã exceptionality: a reassessment. Language, 85(2): 355–404.

[9] Szabolcsi, Anna. 1983. The possessor that ran away from home. The Linguistic Review, 3: 89–102.

[10] Szabolcsi, Anna. 1994. The Noun Phrase. In Ferenc Kiefer and Katalin É. Kiss (eds.), The syntactic structure of Hungarian. New York: Academic Press. 179–275.

English version

01.

Ping Wang: The Transformational Generative grammar proposed by Noam Chomsky marked a major watershed moment in linguistics. Could you tell us why you’re interested in TG grammar and why you chose linguistics as you major?

Dr. Éva Dékány: After high school I stared university as an English Language and Linguistics major. I wanted to learn English and I was interested in literature and perhaps professional translation. As part of the curriculum, we had to take a course in phrasal syntax and another one in clausal syntax in the framework of TG. I wasn’t enchanted at first…but then I went on to take an elective advanced syntax course and things started to get interesting. It was definitely the weekly homework assignments that did it for me. The instructor had a knack for choosing the right level of difficulty for the assignments: they were challenging, but if you thought hard and used creatively what has been said in class, it was possible to solve them. Getting to the bottom of complex linguistic problem sets got me completely hooked. I took up Theoretical Linguistics and a second major and then did my PhD in this area, too.

I am interested in TG because it uses the scientific method in the study of language. You can’t just waffle about the data: you need to have testable and refutable predictions. You can’t just say “it’s a construction”, and be done: you need to ask the deeper “why” questions and explore the consequences as well. In other words, TG plays in the same arena as the hard sciences.

02.

Ping Wang: As we all know, compared with Indo-European languages, Chinese is a language with less morphological changes. And there are many unique linguistic phenomena in Chinese. For example, there is a unique oblique object phenomenon in Chinese, that is, not only the central semantic roles such as agent and patient can serve as objects, but also the non-central semantic roles such as tool and place can serve as objects. In the past, Chinese scholars introduced light verbs and carried out subsequent grammatical operations such as shifting and moving. Therefore, some people think that this research is an abuse of the operational means of generative grammar. Some people even think that it is not suitable to study Chinese under the framework of generative grammar. How do you think about this phenomenon?

Dr. Éva Dékány: In my opinion, saying that framework X is not suitable to study language Y or phenomenon Z is simply silly. Any language and any phenomenon can be studied in any framework. Of course, not every framework will be able to contribute equally insightful analyses about every phenomenon, and when a framework fails to provide insights case after case, then we have a reason to modify it or to just put it aside.

As for light verbs specifically, many languages use them to form complex predicates. We need to go no further than English and Japanese to find well-studied examples. TG is a strongly comparative enterprise; it allows you to draw inspiration from the analysis of phenomenon X is language Y for the analysis of phenomenon Z in language W. I don’t see a problem with postulating light verbs for Chinese, as long as the analysis makes the right predictions and avoids the one-assumption-to-explain-one-language-datum trap.

03.

Ping Wang: Cartography is one of the researches under the frame of generative grammar. However, due to the influence of extreme cartography or hardcore cartography, the syntax tree becomes very complex, which makes many linguists who follow Chomsky think that cartography approach violates the minimalism pursued by the Minimalist Program. In your opinion, does syntactic cartography approach violate the minimalism pursued by generative grammar? And how do you view the relationship between "extreme cartography" and "minimalism"?

Dr. Éva Dékány: It is not possible to put an equation mark between generative grammar in general and cartography: there are also non-Chomskyan generative frameworks such as LFG! Within Chomskyan generative grammar, we are interested in mapping out the hierarchy of functional projections (the so-called functional sequence), and the dependencies that hold between positions in the hierarchy. In this sense, Chomskyan grammar is cartographic to the core. The school called “extreme cartography” operates with the assumption that there is only one feature per head (Cinque and Rizzi 2010). Naturally, this leads to an explosion of functional structure. But one can very well be a Minimalist syntactician without buying into extreme cartography. Chomsky himself is a good example here: he has never adopted the fine-grained hierarchies that Italian cartographers posited, for instance. There are also viable analyses that forgo information-structure related projections such as FocP and TopP in the clause. So Minimalism is most certainly not equal to “extreme cartography”.

What is special about “extreme cartography” is that the lexicon contains a really large number of different functional heads (due to the one-feature-per-head approach). Minimalism has nothing to say about the lexicon. It is concerned with narrow syntax: it tries to operate with the bare minimum of syntactic modules, principles, operations, levels of representation, etc., eliminating unnecessary assumptions from the framework. Therefore there is no inherent tension between Minimalism and “extreme cartography”. In my view, extreme cartographers and moderate/minimal cartographers mostly disagree about the contents of the mental lexicon rather than the nature of narrow syntactic operations.

04.

Ping Wang: The so called DP Hypothesis has gained acceptance in Government and Binding (GB) theory. This hypothesis proposes that a nominal expression is headed by a determiner that takes a noun phrase as its complement. How is DP represented in Hungarian?

Dr. Éva Dékány: Early work by Anna Szabolcsi on the Hungarian Noun Phrase, especially on the definite article and possession, has been very influential in cementing the then-emerging DP-hypothesis (Szabolcsi 1983, 1994, among others). To adapt a phrase from literature studies, everybody who works on the Hungarian NP or on possession in general “came out of Szabolcsi’s overcoat”. In Hungarian, the DP-zone comprises multiple projections. The lowermost of these, which I have called DeixP, is the merge-in site of demonstratives (Den Dikken and Dékány 2018, Dékány 2021). DeixP is dominated by the DP, which has the definite article in its head. Phrasal demonstratives move to spec, DP from spec, DeixP. If there is no definite article, demonstratives which spell out the Deix head move to D. The ‘escape hatch’ position of the DP phase is occupied by dative-marked possessors: these can be extracted from the nominal projection. It is a recent discovery of Farkas and Alberti (2016) and Alberti and Farkas (2018) that in deverbal nominalizations an oblique modifier of the base verb can appear before dative possessors. Exactly how these should be represented in the DP is a matter of ongoing debate.

05.

Ping Wang: Hauser et al. (2002) claim that recursion is part of Universal Grammar, that is, the human language faculty. Do you think it is reasonable and why?

Dr. Éva Dékány: Grammatical recursion means that we use a finite set of basic building blocks (morphemes and words) to build a potentially infinite set of structures (phrases and clauses). This is at the core of human language, which distinguishes it from animal communication. The operation that allows grammatical recursion is Merge. The big question is whether Merge is specific to language (in which case it is part of the human language faculty) or other cognitive functions also use Merge (if so, then Merge is a domain-independent, general operation, not part of the language faculty). The position of generative grammar is that Merge is specific to language, hence, part of the human language faculty. My impression is that in many cases this is based on a deep-seated conviction rather than empirical/experimental evidence. In order to dig deep into this question, linguists and cognitive scientists must work together: we need teams that have a thorough understanding of both linguistics and other cognitive abilities in order to satisfactorily address this question.

Category recursion is a different matter. It means that a category can embed itself. For instance, a DP can embed a DP, a CP can embed a CP, etc. Hauser et al (2002) do not make any claims about this type of recursion; their only concern is grammatical recursion, that is, Merge. Whether category recursion is attested in every language came under debate about 15 years ago, when Everett (2005) claimed that this type of recursion is absent from Pirahã. To my mind, papers like Nevins, Pesetsky and Rodrigues (2009) have successfully rebutted Everett’s claims. To this date, we do not have convincing cases of particular languages lacking category recursion. But as I said, this issue is independent of the claims of Hauser et al (2002).

06.

Ping Wang: As we all know, syntax is a branch of linguistics which includes rules of words and phrases to create coherent sentences while grammar is a system of rules that defines the grammatical structure of a language. What are the differences between syntax and grammar?

Dr. Éva Dékány: ‘Grammar’ refers to all levels of language with a structure. That is, grammar includes phonology, morphology, syntax as well as compositional semantics. This means that the relationship between grammar and syntax is that of proper containment.

07.

Ping Wang: As a grammatical category, nominal describes words or groups of words that function together as a noun. The words in a nominal group give more details about the noun (the headword) and make it specific. Nominal phrases are phrases that can function as the subject or object of a verb. What’s the relationship between nominal phrases and nominal clauses?

Dr. Éva Dékány: A nominal clause is a phrase that has the internal structure of a clause (it’s headed by a verb) but the external distribution of a Noun Phrase. This situation can arise in two ways. We can have a clause that modifies a null nominal (a Kaynean ZERO noun or a covert pronoun), either as a complement or as an adjunct. In this case the nominal distribution is due to the null nouny element that is the head of the phrase. Alternatively, we can have a so-called ‘mixed projection’, where we start projecting a clause with verbal functional heads, and somewhere along the way we switch to nominal functional heads. Such projections will be more verby or more nouny depending on where the switch happened. It’s important that in the case of ‘mixed projections’ we are dealing with a nominalization, while in the first case, with a clause modifying a null nominal, there is no nominalization.

08.

Ping Wang: Modern Hungarian relative clauses demonstrate a third type of relative pronoun in addition to wh-REL elements in English and dem-REL elements in German: “Modern Hungarian relative operators contains both markers, and will hence be referred to as dem-wh-REL” (Bacskai-Atkari & Dékány,2021). Could you explain this dem-wh-REL from the perspective of morphology?

Dr. Éva Dékány: My joint work with Julia Bacskai-Atkari at the University of Konstanz uses the term ‘wh-REL’ for relative pronouns that are form-identical to interrogative pronouns (e.g. English Who was late? vs. the girl who was late). We use ‘dem-REL’ for relative pronouns that have grammaticalized from, and therefore share their form with demonstrative pronouns. These are the two most common patterns cross-linguistically. Relative pronouns in Hungarian are morphologically complex. They consist of a stem that is formally identical to an interrogative pronoun, and a prefix that has grammaticalized from a demonstrative pronoun. This prefix then formally differentiates relative pronouns from interrogative pronouns. We called Hungarian-type relative pronouns dem-wh-REL to reflect the fact that relative pronoun has two parts: one that is related to wh-item and another that is related to demonstratives. It is not a wide-spread pattern, but it can also be observed in Greek.

References

[1] Alberti, Gábor, and Judit Farkas. 2018. Modification. In Gábor Alberti and Tibor Laczkó (eds.), Syntax of Hungarian. Nouns and Noun Phrases, 2, 775–896. Amsterdam: Amsterdam University Press.

[2] Bacskai-Atkari, Julia and Éva Dékány. 2021. Cyclic changes in Hungarian relative clauses. In Thórhallur Eythórsson and Jóhannes Gisli Johnsson (eds.),  Syntactic Features and the Limits of Syntactic Change. Oxford: Oxford University Press. 40–63.

[3] Cinque, Guglielmo and Luigi Rizzi. 2010. The cartography of syntactic structures. In Bernd Heine and Heiko Narrog (eds.), The Oxford handbook of linguistic analysis. Oxford: Oxford University Press. 51–65.

[4] Dikken, Marcel den and Éva Dékány. 2018. A restriction on recursion. Syntax, 21(1): 37–71.

[5] Everett, Daniel L. 2005. Cultural constraints on grammar and cognition in Pirahã. Current Anthropology, 46(4): 621–646.

[6] Farkas, Judit, and Gábor Alberti. 2016. Positions for oblique case-marked arguments in Hungarian Noun Phrases. Jezikoslovlje, 17 (1–2): 295–319.

[7] Hauser, Marc D., Noam Chomsky and W. Tecumseh Fitch. 2002. The faculty of language: What is it, who has it, and how did it evolve? Science, 298. 1569–1579.

[8] Nevins, Andrew and David Pesetsky and Cilene Rodrigues. 2009. Pirahã exceptionality: a reassessment. Language, 85(2): 355–404.

[9] Szabolcsi, Anna. 1983. The possessor that ran away from home. The Linguistic Review, 3: 89–102.

[10] Szabolcsi, Anna. 1994. The Noun Phrase. In Ferenc Kiefer and Katalin É. Kiss (eds.), The syntactic structure of Hungarian. New York: Academic Press. 179–275.

往期推荐

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 闫玉萌 陈金玉 王平

排版：马晓彤 闫玉萌 高洁

审校：王丽媛 李芳芳 陈旭