人物专栏 | Adriana Belletti 教授访谈（上）

点击上方蓝字关注我们

编者按

《理论语言学五道口站》（2022年第03期，总第206期）“人物专栏”与大家分享近期Simone Guesser（巴西罗赖马联邦大学教授）与Núbia Ferreira Rech教授（巴西圣卡塔琳娜州联邦大学方言语言文学系教授）对Adriana Belletti教授的访谈。此次访谈以“句法制图：轻动词边缘，语言习得及其他”为主题，由本站成员赵欣宇、雷晨、聂简荻、郭思源、丁子意进行翻译。Adriana Belletti教授，意大利锡耶纳大学语言学教授，研究领域为理论比较句法和不同形式下的语言习得比较研究。

本期访谈中，Adriana Belletti教授阐述了句法制图在生成语法研究中的重要性，对轻动词词组边缘域的发展进行了回顾与展望，表达了自己对理论中加入实验研究的支持以及对语言学界不同理论的看法。

本次访谈内容转自Fórum Linguístico (Simone Guesser & Núbia Ferreira Rech, 2019)，共分为上下两期，后续内容将在下一期人物专栏中继续与大家分享，敬请期待。

采访人物简介

Adriana Belletti 教授

Adriana Belletti， 意大利锡耶纳大学语言学教授，美国麻省理工学院语言学系访问学者和研究员。她曾任瑞士日内瓦大学教授，并在多个语言学系和研究机构担任教职。她的主要研究领域为理论比较句法和不同形式下的语言习得比较研究。

Brief Introduction of Interviewee

Adriana Belletti, professor of linguistics at the University of Siena, is a visiting scholar and a research affiliate at the Department of Linguistics of MIT. She has been professor at the University of Geneva and has taught in various linguistics departments and research institutes. Her fields of research are theoretical comparative syntax, comparative studies of different modes of language acquisition.

采访者简介

Simone Guesser 教授

Simone Guesser，巴西罗赖马联邦大学本科和硕士学位课程教授，硕士毕业于意大利锡耶纳大学语言学专业，博士毕业于计算机科学、数理逻辑和认知科学专业。她的研究兴趣包括语法理论（主要研究巴西葡萄牙语和其他罗曼语系的语言）、语言习得以及语法理论和教学之间的关系。

Núbia Ferreira Rech 教授

Núbia Ferreira Rech，巴西圣卡塔琳娜州联邦大学语言学研究生课程和方言语言文学系教授，硕士毕业于巴西南大河州联邦大学理论与语言分析专业，博士毕业于巴西圣卡塔琳娜州联邦大学同专业，并在巴西巴拉那联邦大学进行了同一领域的博士后研究。她的研究范围包括功能中心语以及与语法教学有关的主题。

Brief Introduction

Simone Guesser is a professor of undergraduate and master's degree programs at the Federal University of Roraima. She holds a postgraduate degree in Linguistics and PhD in Computer Science, Mathematical Logic and Cognitive Sciences from the University of Siena. Her research interests include grammatical theory - focusing on the syntax of Brazilian Portuguese and other Romance languages, language acquisition and the relationship between grammatical theory and teaching. 

Núbia Ferreira Rech is a professor of the Postgraduate Program in Linguistics and the Vernacular Language and Literature Department of the Federal University of Santa Catarina. She holds a Master's degree in Theory and Linguistic Analysis from the Federal University of Rio Grande do Sul and a PhD in Theory and Linguistic Analysis from the Federal University of Santa Catarina. She developed a postdoctoral research in the same area at the Federal University of Paraná. The scope of her researches includes functional heads as well as topics related to grammar teaching. 

访谈内容

01.

Simone Guesser教授（S.G.）及Núbia Ferreira Rech教授（N.F.R.）：对于生成语法的发展来说，句法制图有何重要性？

Adriana Belletti教授：80年代和90年代初GB（管辖与约束）框架下的原则和参数方法促使了形式生成语法学界研究成果的空前繁荣。学者们发现了不同的语言和语系（罗曼语、日耳曼语、东亚语、美洲印第安语、南岛语、班图语、乌格罗-芬尼语......）的无数相似和相异之处。我认为，句法制图研究可以再现这一盛况。原因在于，句法制图对语言进行精细分析，为句内各成分绘制信息精准、内容丰富的句法图，句法结构构建则通过简单递归及合并操作完成，这与最简方案相辅相成。Rizzi (1997) 对句子左缘结构的分析、Cinque（1999）通过状语对功能中心的讨论，以及动词短语与句子内部语篇的研究等（见后文），都为研究不同语言的层级结构及绘制句法图带来了新角度和新挑战。从句法制图框架下不同语系的大量研究中，我们已经可以一睹该成果的爆发性繁荣（详见RIZZI  & CINQUE2016, BOCCI&RIZZI2017）。

句法制图不仅丰富了句法学、形态学、形态句法学（包括纳米句法）等领域的研究（STARKE，2010；PAVEL，2009...），而且对于句法与语篇语用及句法与韵律学的接口研究也贡献颇多（BIANCHI；BOCCI；CRUSCHINA，2016）。句法制图研究对理论假设不断完善，催生了更多研究问题，对于接口问题也有了创新性进展。让人感到欣慰的是，还有无尽未知等待我们探索，无数事实亟需挖掘和解释。同时，句法制图也推动了语言习得研究的发展：(1)建立健全的习得语料库 (2)对语料提供合理可信的解释，（实际上，后者是过去五年在欧洲运行的一个大型研究项目的核心问题，该项目由欧洲研究理事会（European research Council）资助，由Luigi Rizzi主持，题目为“从映射到原则：成人语法和语言习得中的句法制图和局域性理论研究”）。

02.

S.G. 教授和N.F.R. 教授：您对语言学研究贡献良多，其中一项是轻动词词组边缘(vP periphery)研究。您根据意大利语中动词后主语可被解释为新信息焦点这一语言现象，指出轻动词词组边缘的存在。您的经典著作“Aspects of the Low IP Area”发表已有14年，您认为目前轻动词词组边缘的研究有哪些主要进展呢？

Adriana Belletti教授：句子低层确实牵涉到不同语言的各种与话语相关的过程。在某种程度上，对句子低层的存在和位置的识别不如句子左缘结构直接，因为这一结构域很少有明确的语素能够作为依据（最近一些阿拉伯方言的研究可能与之相关）。尽管如此，通过假设句中存在一个包含焦点、话题位置的与话语相关的低层域，很多语言现象得到了有趣又具有原则性的解释。这一低层域存在于众多语言中，如马拉雅拉姆语（JAYASEELAN，2001）、汉语（Tsai，2015）、意大利语（此外，还包括分裂结构、兼类现象等常见结构Belletti 2009，2015）、巴西葡萄牙语（包括分裂结构，GUESSER，2007，和相关wh-原位现象，Kato，2003）、古意大利语（POLETTO，2006）、西西里语和其他方言（CRUSCHINA，2012）等。此外，当下一些研究利用低位轻动词词组边缘探讨意大利方言中的wh-原位现象（例如BONAN，2018）。我认为，这一结构域还有许多内容有待发现，也可以从句子低层域和左缘域（高层域）的共性和特性入手展开研究。它们的共性和特性都有原则性的理由作为支撑（比如句子中需要取宽域的成分，只与左缘有关，与低层结构无关）。关于轻动词词组边缘，我们提出如下新问题：句子低层是否如左缘一样会出现副词性修饰语？换言之，句子低层是否存在容纳修饰语的的位置？出现在这一位置的成分是同类型的吗？句子焦点（小句内部和小句外部）的性质和信息结构在不同的语言中是如何体现的？焦点位置的解释与特定韵律有多大关联？这种韵律解释与相关解释的标记词的存在与否有什么关系？这些都是研究进程中遇到的众多问题之一，一些问题正指导着现有研究工作，还有一些是我们即将面临的。这些新的方向还会催发出新的研究问题。我认为轻动词词组边缘领域中能有新的问题提出并且未来能源源不断地有新的问题产生，这是一个好现象，这证明它仍然是一个有生命力的话题。

03.

S.G. 教授和N.F.R. 教授：您如何看待在理论语言学研究中引入实验研究呢？

Adriana Belletti教授：在我的语言学导读课上，我告诉学生的第一件事是，（形式、生成）语言学家通常利用“语法性判断”描述所研究的语言属性，然后才尝试寻找明确的理论解释。如果缺乏完善和精确的描述，就很难对语言现象进行解释，因此语法性判断是十分关键的一步。生成语言学研究的终极目标是对人类的语言能力（CHOMSKY, 2016 非技术讨论）在不同语言层面（句法、语义、语音、形态、会话层面......）的各种表征进行恰当的描述；语法性判断是特殊的“镜子”，它会使原本难以理解的认知能力变得更加容易理解。

在此背景下，为了得出可靠的结论，（形式、生成）语言学家必须非常谨慎地向说话者提出合理的问题。问题不能随机收集，而是要有控制条件。为了揭示说话者内在语法的相关属性，我们首先需要提出一个明确的研究问题（比如需要测试的理论指导假设），然后仔细地设计能进行完美的语法性判断的比较的问题例如，给出成对句子，而不是单个句子。这样，被测试的句子间的差异应该最小的。理想情况下，我们需要测试的是唯一的变量属性。在我看来，这很像我们从科学方法中学到的实验方法，这种方法最早起源于伽利略时期。事实上，我也经常告诉我的学生，从说话者那里得到语法性判断的过程实际上就像是实验过程:一个不需要实验室的实验，但和任何实验一样，都需要提出明确的假设来测试和控制实验材料。因此，自乔姆斯基的《句法结构》之后，生成语法研究都是实验性的。比如，所谓的“新比较语法”（HAEGEMAN, 1997）可以进行微观对比，也是经典的实验方法之一。（Kayne在这一领域取得的创新性成果就是很好的例子，可参阅Kayne, 2000, 2005以及其他相关的工作）。 

总之，实验材料越可控，实验结果就越准确。因此，我很倡导在理论语言学研究中使用精确的实验方法得出研究结果。只要我们能提出清晰明确的研究问题，运用精确的统计量化措施，比如应用正确的科学方法，并准确控制变量、分析结果，我都非常支持。传统实验方法是我前面所讲的理论语言学中的主要方法，而这实际上是对传统实验方法的改进，在生成语法的传统中，语言学是经验科学，因此经验科学的实验方法也自然适用于语言学。

04.

S.G. 教授和N.F.R. 教授：您认为目前语言学界的不同的理论与观点在未来会实现整合吗？

Adriana Belletti教授：如果我们的共同目标是提出一个足够接近人类语言能力的特征，那么整合是很有可能的，并且会很受欢迎。但是这一目标在一开始就必须足够明确。如果有不同的目标（比如实现最终分类的目标，这尤其适用于那些基于大数据的统计方法），那么整合可能就比较困难。尽管如此，当我们在区分讲话者内在语法和外在条件各自的影响时，如果我们可以合理的解决统计方面的问题，这将会变得非常有趣，并可能产生多样的结果。至于那些基于应用的方法，整合似乎很困难，因为在我看来这种方法不能解释人类语言能力最显著的特性，即语言的创造性。语言的创造性的最令人惊讶的表现是婴幼儿在语言习得过程中能创造出无数合乎语法的句子。语言习得的研究者都知道婴幼儿表现出的语言行为不同于目标语言的语言行为，也不同于他们接触到的语言输入。但是我们发现婴幼儿对非目标语言的表达有可能是正确的。对我来说，这一直是个令人震惊的发现。这就好像婴幼儿对语法的可能性有探寻空间，即原则与参数方法中的参数，并且在习得过程中，他们会做一些合理的尝试，尤其是在压力之下（可能是语法或计算方面，因为记忆还未发育完全）。实验方法的整合，尤其是心理学传统中的同时运用线上线下技术的实验方法的整合，早已成为事实，并且是最受欢迎的。因此，我认为处理语法的创造力，尤其是处理儿童语法的创造力是一个至关重要的挑战。语法创造力应该是，而且肯定是我多年来的理论语言学研究的核心，虽然这在其他方法中可能并不是那么重要。但对我来说，这就是理论语言学最有价值的地方，而且价值很大。 

如果我们从更具体的角度看这个问题，讨论局部整合，比如最简方案和句法制图，我认为没有必要讨论这些整合，因为它们解决的是同一问题，只不过侧重点不同。句法制图侧重丰富普遍理论的实证基础，而最简方案侧重假设简约性和形式简约性。这两者都很重要，能够完美融合。

English Version

01.

Prof. Simone Guesser (S.G.) and Núbia Ferreira Rech (N.F.R.): What is the importance of Cartographic Syntax for the future of Generative Linguistics?

Prof. Adriana Belletti: With the Principles & Parameters approach within the so-called GB framework in the eighties and early nineties, the field of formal generative linguistics has experienced a real empirical explosion, so to speak, with innumerable discoveries of similarities and differences manifested by very diverse languages and language families (from Romance to Germanic to East-Asian, Amerindian, Austronesian, Bantu, Ugro-Finnic…). My feeling is that something similar can happen again through the studies in syntactic cartography: the offering of fine-grained analytical tools by the cartographic approach, which design precise and rich maps of different areas of the clause, all built through the simple recursive Merge operation, as made explicit in the minimalist program. Thus, the available maps of the left periphery of the clause (from Rizzi’s 1997 seminal paper on, probably the area of the clause analyzed in most details), of the internal morphosyntactic functional spine of the clause (inspired mostly by Cinque’s 1999 highly innovative analysis of adverbial classes), of the clause internal discourse related area at the periphery of the verb phrase (see below)–all these available maps open several new questions as to the status and the behavior of different languages with respect to given positions in the maps. We can already see this empirical explosion in the amount of work that is being developed in the frame of syntactic cartography in different language families (RIZZI; CINQUE, 2016; BOCCI; RIZZI, 2017 for overviews).

The new discoveries enrich the empirical domain of the available database in different areas of syntax, morphology, morphosyntax (also with the contribution of work in nano-syntax (STARKE; 2010; PAVEL, 2009…) and in domains at the interface of syntax with discourse-pragmatics and prosody (BIANCHI; BOCCI; CRUSCHINA, 2016). Through their empirical discoveries, studies in cartographic syntax have generated new research questions that challenge theoretical assumptions and call for constant refinements, in particular due to the innovative work on the interfaces. It is a good feeling to see that not everything has already been discovered and that a lot is still in need of suitable explanations. Syntactic cartography also inspires work and new research trends in language acquisition, suggesting new perspectives (see also below) and a constant dialog between (i) the aim to reach large empirical coverage thus integrating in the database also acquisition data - and (ii) the search for theoretically sensible and constrained explanations (the latter is indeed the central issue of a large research project run in Europe over the last five years, funded by the European.

02.

Prof. S.G. and N.F.R.: Among your contributions to linguistic studies is the identification of the vP periphery, based on data from post-verbal subjects interpreted as new information focus in Italian. In your opinion, after 14 years of publication of the article Aspects of the Low IP Area, what are the main advances that have occurred in studies about the vP periphery?

Prof. Adriana Belletti: It seems that the low part of the clause is indeed implicated in various discourse related processes across languages. Somehow, the identification of its presence and position is less straightforward than in the case of the left periphery of the clause as explicit morphemes are rarely available in this area (recent work on some Arabic dialects may turn out to be relevant in this respect). Nevertheless, a variety of phenomena have found an interesting and principled account by assuming the presence of this discourse-related area of the clause containing positions of focus and topic across languages as diverse as, among others, Malayalam (JAYASEELAN, 2001), Chinese (Tsai, 2015), Italian (also in different constructions such as most notably cleft structures and doubling type phenomena, Belletti 2009, 2015), Brazilian Portuguese (in clefts, GUESSER, 2007, and also in the related wh-in situ phenomenology (Kato 2003)), Old Italian (POLETTO ,2006), Sicilian and other dialects (CRUSCHINA ,2012, and ongoing work on the analysis of the wh-in situ phenomenon in some Italian dialects exploiting the low vP periphery, as in BONAN, 2018, for example). I think that there is still much to be discovered in this area, also in the perspective of disentangling the aspects that it shares with the clause external one and those that it does not. Principled reasons are most likely to be involved in both possibilities (such as elements that require sentential scope should characteristically only concern the left peripheral area and not low clause internal positions). Among the new questions to ask, one can list the following: does the low area of the clause also contain positions where types of adverbial modifiers can be found, much as it happens in the left periphery? That is, are there also Mod-like positions in this area? Should we expect the same types of elements in these positions? How are the nature and the information content of the focus positions–the clause internal and the clause external one–parametrized across languages? To what extent does the interpretation associated with the focus positions in the clause correlate with a dedicated prosody? To what extent does the prosodic interpretation correlate with presence/absence of a dedicated marker of the relevant interpretation? These are among the numerous questions in the research agenda, some of which are guiding ongoing work, and there are others to come. These new lines of research will, in turn, undoubtedly raise new questions. I consider the fact that new questions have been and are being raised as a positive feature of the proposal and a sign that it continues to be a lively one.

03.

Prof. S.G. and N.F.R.: What is your opinion towards the introduction of experimental studies in research in theoretical linguistics?

Prof. Adriana Belletti: When I teach an introductory class, one of the first things I tell my students is that (formal, generative) linguists typically make use of ‘grammaticality judgments’to arrive at their descriptions of properties of the language(s) they work on; from such descriptions, they may start the attempt to look for explicit theoretical explanations. Since without a good and precise description we can hardly even attempt to start looking for explanatory accounts, the stage of getting grammaticality judgments is a rather crucial one. The ultimate goal of the research in theoretical generative linguistics is to provide an appropriate characterization of our language capacity as human beings (CHOMSKY, 2016 for recent non-technical (re)discussion) in its various manifestations on different levels (syntactic, semantic, phonological, morphological, at the discourse level…); grammaticality judgments are privileged lenses through which an otherwise hardly accessible cognitive capacity can become more accessible.

Given this general background, in order to achieve reliable conclusions, the (formal, generative) linguist must be very careful in asking the speakers the right questions. The questions must not be collected at random but always in a controlled way. In order to make relevant properties of the speaker’s internal grammar come out, one should always first have a precise research question in mind, (i.e. a theoretical guiding hypothesis to test), and then one should be extremely careful in formulating the grammaticality question(s) in such a way that the judgments be preferably comparative (i.e. giving pairs of sentences, not sentences in isolation). In this way, the sentences put to test should minimally differ from each other. Ideally, the only variable property should be the one under testing. This, to me, looks a lot like the experimental approach, as we have learned from the scientific method, whose origin dates back to Galileo. Indeed, I always also tell my students that the task of obtaining grammaticality judgments from speakers is in fact like performing an experiment: no need of a lab with this type of experiment, but clear hypotheses to put to test and controlled experimental material, as in any experiment. Research in generative grammar is thus experimental in nature from its origin, from Chomsky’s Syntactic Structures up to now. To quote one example, the so-called ‘new comparative syntax’ (HAEGEMAN, 1997) - through the micro-comparisons that it makes possible - is also a classical implementation of the experimental approach (Kayne’s seminal work is a leading example in this respect, KAYNE, 2000, 2005 and much related work).

From all perspectives, the more controlled, the material to test is, the better. Thus, I could not be more positive and welcoming about the use of explicit experimental methods and the studies they generate in theoretical linguistics. To the extent that these studies ask precise and explicit research questions, as in the proper application of the scientific method, and are precise as to the controlled variables they are putting into test as well as in the analysis of their results, to be performed also through use of precise statistical quantitative measures, to me, this is all more than welcome. It is in fact an enrichment of the classical experimental method, which is at the core of theoretical linguistics as I have tried to characterize above. Within the tradition of generative grammar, linguistics is an empirical science, so the methods of empirical sciences naturally suit it.

04.

Prof. S.G. and N.F.R.: Do you believe that in the future there may be some kind of integration between different theoretical perspectives of current Linguistics?

Prof. Adriana Belletti: If the shared aim is that of providing a characterization which is close enough to our language capacity as human beings, integrations are always possible and welcome. However, this aim has to be clear at the outset. If one has different aims (e.g. ultimately taxonomic ones, possibly extremely performing ones as in some statistical methods using big data), integrations may be hard. Nevertheless, it would be extremely interesting and potentially rich of consequences if some statistical factors were spelled out properly in the aim of factoring out and disentangling the respective role of the speaker’s internal grammar and external conditions. As for the so-called usage-based approaches, integrations seem hard, as it seems to me that they cannot face the most striking property of our human language capacity, which is our linguistic creativity. The probably most amazing manifestation of it is provided by the innumerable examples of the grammatical creativity that young children manifest in the course of acquisition. As is in the experience of anybody working on language acquisition, children often manifest a linguistic behavior that differs from that found in the target language, ultimately from the input that they have most likely been exposed to. Looking at other languages, however, we typically find that children’s non-target expressions are in fact possible there. This, to me, is always such an astounding discovery. It is as if children had a search space of grammatical possibilities–the parameters of the P&P approach–and during the course of acquisition they try out some of the available options, most likely under some pressure (possibly grammatical, computational, due to immature memory resources, etc…). The integration of experimental methods, in particular those coming from the psycholinguistic tradition that make use of both online and offline techniques, is already a fact, and a most welcome one, as it is also clear from the answer to the preceding question. Therefore, I think that dealing with grammatical creativity–in general and with children’s grammatical creativity in particular–is a crucial challenge. Grammatical creativity should be and it certainly is at the core of the theoretical linguistic research in the tradition I have assumed over the years; it may not be equally central in other current approaches. To me, though, this is what makes theoretical linguistic worthwhile, and a lot so.

If the question had a narrower perspective, and referred to more local integrations, such as aspects of minimalism and aspects of syntactic cartography, my impression is that there is no need of talking about integrations in these cases, as these are just partly different ways to address the same fundamental questions. It is just a matter of different emphasis. On the one hand, there is more emphasis on the heuristic ability to enrich the empirical basis of the general endeavor in aspects of syntactic cartography. On the other hand, there is more emphasis on the reduction of language-specific general assumptions and on their formal simplicity in aspects of minimalism. Both are crucial issues that can perfectly integrate each other and should do so.

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：闫玉萌 赵欣宇 雷晨 

排版：闫玉萌 赵欣宇 雷晨 

审校：陈旭 李芳芳 田英慧