人物专栏 | Yoshio Endo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19期，总第222期）“人物专栏”与大家分享本站采编人员赵欣宇对Yoshio Endo教授进行采访的访谈录。Yoshio Endo，日本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，曾在美国麻省理工学院担任访问学者。
本期访谈中，Yoshio Endo教授首先讨论了语言逻辑与数学逻辑的关系，就数学逻辑中的莱布尼兹定理及布尔代数交换律在语言中的作用进行了对比，然后举例论述了外在句法和内在句法的关系，表达了自己对wh-疑问词相关研究的看法。
本次访谈内容共分为上下两期，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。
人物简介
Yoshio Endo教授
Yoshio Endo（遠藤喜雄），日本神田外语大学教授，曾在美国麻省理工学院担任访问学者。他师从Luigi Rizzi教授，在瑞士日内瓦大学获得博士学位。他还曾任日本岛根大学助理教授、日本横滨国立大学副教授。其主要研究领域为理论语言学和认知科学。
Brief Introduction
Yoshio Endo is a professor at Kanda University of International Studies and visiting scholar of MIT. He received his Ph.D. at the University of Geneva under the supervision of Prof. Luigi Rizzi. He was assistant professor at Shimane University and associate professor at Yokohama National University. He mainly focuses on theoretical linguistics and cognitive science.
访谈内容
01.
赵欣宇：您认为语言逻辑与数学逻辑有何相同之处？
Yoshio Endo教授：正如我在“Boolean algebra in bi-clausal sentences”（2021）中所讨论的，我认为自然语言的语法中有数学逻辑结构的影子，许多看似与数学不相关的语法项其实都包含着某些功能范畴，有些功能范畴是显性的，有些则是隐性的。举例来说，布尔代数中的交换律如下：
a・b=b・a.
下句可以反映出交换律的本质：
Harry [plays the lute](a) and [sings madrigals](b).= Harry [sings madrigals](b) and [plays the lute] (a).
然而，描述序列事件的句子却不符合这一交换律：
I [went to the store] (a) and (then) [bought the whisky] (b). ≠ I [bought the whisky] (b) [went to the store] (a).
这种差异在Ross（1968）提出的并列结构孤岛效应（coordinate structure constraint , CSC）理论中表现得更加明显。CSC不允许如上述A和B这样的并列成分进行移位。只有当并列结构具有布尔代数属性时，CSC才允许其移位，例如：
*the lute which Harry plays and sings madrigals is warped. (Ross 1986: 100) vs. Here is the whisky which I went to the store and bought. (Ross 1986: 103)
下面三个例句都带有像“come”和“go”这样的运动动词，我们可以通过这些例句进一步研究上述问题。
Come go eat with us.
当“come”和“go”没有被“and”连接时，它们似乎同属于一个功能范畴。Cinque（1999，2006）认为当词项被当作功能范畴时，它们会按照层级结构进行排列，因此不能互换位置（也就是说，不具有布尔代数属性）。事实也是如此，我们可以直接观察到上述例句中“come”和“go”的序列具有层级结构，因此如下所示，“come”和“go”不能互换：
Come go eat with us! vs. ??Go come eat with us. (Shopen 1971) (cf. Zwicky 1969, Bjorkman 2016)
同预期一致，当功能范畴“come”和“go”没有被“and”连接时，就不会产生孤岛效应，如下所示：
??What meal did you ask him to come and go and eat with us? vs. What meal did you ask him to come go eat with us? (Shopen 1971)
Cinque和Rizzi认为功能范畴的层级具有普遍性。也就是说，尽管两种语言的基本语序不同，但二者的层级结构是相同的。例如，日语不属于像英语那样的中心语前置语言，而是一种中心语后置语言，那么英语和日语中的“come”和“go”的线性语序应互为镜像。但是，如下所示，“come”和“go”的层级结构在这两种语言中却是一致的，也就是说，“come”的层级高于“go”的层级。
[come [go [eat; tabeni ‘eat’] itte ‘go’] kuru ‘come’].
综上所述，我们可以看到CSC具有一种数学属性，即CSC只适用于具有布尔代数属性的句子。在研究自然语言所特有的一致性现象时，这一点变得更加清晰。例如Chomsky（“Genuine Explanations”，2021）提到的句子。
a. John arrived early, met Bill and got a good seat.
(三个独立事件)
b. To arrive early, meet Bill and get a good seat seems/*seem to be what John wants.
（单一复合事件）
c. arriving early, meeting Bill and getting a good seat seems/*seem to be what John. Wants
(单一复合事件)
d. To arrive early, to meet Bill and to get a good seat seem/seems to be what John wants.
(单一复合事件，三个独立事件)
这里的区别在于有三个动词短语的句子是代表一个（有序的）复合事件，还是三个独立的事件。为了能够理解这里的每个句子，我们先从直观的角度去讨论。首先，三个动词短语的时态似乎对判定该句子是单一复合事件还是三个独立事件来说很重要。Stowell（1982）研究了控制句中带有不定式“to”的动词短语的性质，发现带有“to”的动词有一个未实现的时态，如下所示：
John persuaded Mary to leave.
Martin（2001）完善了这一观点，在仔细观察了各种英语中带有不定式“to”的句子后，他得出结论：“to”具有某种时制特征。根据这一观点，再结合上述Chomsky提出的句子中三个动词短语包含不定式“to”时都解释为具有三个独立事件这一事实，我认为，有多少事件就有多少时制特征。考虑到一个时制特征代表一个独立事件，一个有三个不定式“to”的英语结构就会有三个时制特征，即代表三个独立事件。因此，这个句子有三个事件，动词用复数表示。(当这三个事件被视为一个整体事件时，则用单数表示，后文会做出解释）。
相反，如果每个动词短语都有“ing”，那么它就不具有时制特征，这与动词都用不定式“to”时的情形相对。因此，如果每个动词后面都有“ing”，那么该句子将只有一个时制特征。因此，当三个带“ing”的动词短语连接在一起时，动词用单数表示。
让我们用偏序格更精确地阐述这一点。例如，假设有<狗、猫、动物>的偏序集。当比较每个元素中蕴含的信息时，其中两个任意元素是狗和猫，它们共同的上位概念是动物，这里动物被称为“上确界”。带着这种思考，让我们回顾一下Chomsky的句子。在这里，我们可以得到事件集< arrive early, meet Bill, get a good seat >。如果这三个动词短语都含有“to”，那么这三种时态就代表着三个独立事件，所以主句动词要用复数形式。如果上确界包含的三个独立事件表达整体事件的含义，那么主句就只有一种时态，所以动词要用单数形式。
相反，如果这三个动词后面都有“ing”，那么它们就没有时态，因为动词短语中的“ing”不具有时制特征。上确界表达的整体事件中包含动词短语，整体事件需要借助主句的时制特征才能获得时态。因此一个时态代表一个事件，主句动词要用单数形式。
接下来，我们讨论一下Chomsky教授在“Minimalism: where we are now, and where we are going”（2020）中所遗漏的部分。需要注意的是，当三个动词短语以不定式“to”形式出现时，它们就具有了布尔代数属性，因此可以进行如下移位：
To arrive early, to meet Bill and to get a good seat seem/seems to be what John wants. = To meet Bill, to arrive early, and to get a good seat seem/seems to be what John wants.
相比之下，“ing”形式的三个动词短语则不具有布尔代数属性，因此，如下所示，它们不能移位：
Arriving early, meeting Bill and getting a good seat seem/seems to be what John wants. ≠ Meeting Bill, arriving early, and getting a good seat seem/seems to be what John wants.
因此，我们可以把自然语言分为两类：有数学属性的语言和无数学属性的语言。如果功能范畴使得人类语言不具有数学属性，那么我们可以说英文中的“ing”属于功能范畴，而“to”具有包括未实现时态在内的多种含义，因此不属于功能范畴。
顺便说一下，我在研究生课堂上讨论这个话题时，一位中国学生告诉我，汉语中可以通过“了”字的有无来反映英语中“ing”和“to”的这种差异，如下所示：
吃饭，唱歌和跳舞（separate event 独立事件）
to eat, to sing and to dance
吃了饭，唱了歌，然后跳了舞（single event 单一事件）
eating, singing and dancing
02.
赵欣宇：请问您如何看待莱布尼茨定理和布尔代数逻辑在语言学研究中的作用？可以详细说明一下吗？
Yoshio Endo教授：布尔代数交换律和莱布尼茨定理的相似之处在于它们都涉及交换元素的操作，但它们的适用范围不同。其一，莱布尼茨定理指出，当A=B时，若关于A的命题为真，则同样的命题代入B也为真，反之亦然。举个例子：
a. Remi(A)=Yoshio’s 13-year-old dog(B)
b. Remi(A) sleeps well.
c. Yoshio's 13-year-old dog(B) sleeps well.
（1a）为真，（1b）也为真，而在（1c）中，我们把“Remi”和“Yoshio’s 13-year-old dog”互换，命题也为真。这种情况下，元素的特征是通过非坐标结构中的元素互换来揭示的。
其二，布尔代数的交换律在坐标结构中所起的作用，上文已经讨论过。
03.
赵欣宇：在您看来，状语从句的外在句法和内在句法是什么关系？
Yoshio Endo教授：正如我上文所讨论的，内部语法处理内部结构问题，而外部语法处理成分及其内部结构与主句之间的关系。Haegeman和我在“Adverbial Clauses and Adverbial Concord”（2019）中提出了一种称为“状语协同合并”的操作来将两者联系起来。如上所述，该操作的要点是，在状语从句中当某个位置（例如，时体）发生移位时，移位成分决定整个状语从句的类型（例如，时体类型），然后将确定的成分（例如，时体类型状语从句）与主句中的相应成分进行连接（例如：时体成分）。在这种情况下，“协同”很重要，这意味着在两个成分之间有一个制约。我们认为这种制约使得这两个成分紧密相连。例如，在英式英语中，这种制约要求“since”与主句的完成体相关联，而与过去/现在时无关，如下所示。在这种情况下，“since”与主句的完成体一致。
I have been living in London since I was a child.
*I live in London since I was a child.
*I am living in London since I was a child.
而关于状语从句和相关主句之间的联系，Haegeman和我的观点将会引发一个有趣的预测。当一个状语从句与一个较高的外部成分关联时，它应该出现在另一个较低的内部成分之前。这一点通过下图更容易理解，其中，高层的外部状语从句（1）与高层的外层功能中心语（FH（1））相关联，而低层的内层状语从句（2）与低层的内层功能中心语（FH（2））相关联。因此，较高的外部状语从句（1）出现在较低的内部状语从句（2）之前。
[high Adverbial clasue(1) [low adverbial clause (2)... FH(2)] FH(1)].
在我们的研究生院中有很多中国学生，所以我向他们取证核对，想确认这个观点在中文中是否成立，下面让我们看看结果。首先，在日语中，表达言语行为情绪的功能范畴出现在表达说话者情绪的功能范畴的外部，因此相应的状语从句也出现在它之前。这与中文的情况似乎是一样的，如下所示。
要我说（A-mood）虽然可能性很低（S-mood）但可能会落选。vs.?虽然可能性很低（S-mood）要我说（A-mood）但可能会落选。
在许多其他情况下，日语和汉语似乎表现出相同的状语从句限制，如下所示：
事实上(A-mood) 上田先生(top) 秋天可能会搬家。vs. ？上田先生(top)事实上(A-mood)秋天可能会搬家。
大学毕业并继续攻读研究生课程时（T），我一边保留了学籍（Asp），一边担当非营利组织的董事长。vs. ？我一边保留了学籍（Asp），大学毕业并继续攻读研究生课程时（T），一边担当非营利组织的董事长。
虽然汉语是否具有与日语相同的功能范畴及层级排列这一问题还有待考察，但是如上所述，汉语中也发现了日语中出现的状语从句限制条件，那么汉语中可能存在与日语相同的功能范畴层级。如果日语与汉语的状语从句序列存在差异，则意味着两种语言的功能范畴层级略有不同。欧洲语言中也存在这种功能范畴层级互换的现象，Rizzi针对这一问题做过相关论述。
04.
赵欣宇：英语中，询问原因的wh-词并不唯一。请问其他语言中是否也存在类似的现象？针对询问原因的wh-词的研究对句法制图的发展有何益处？
Yoshio Endo教授:英语中有几种询问原因的疑问句，如“why”，“what for”，“how come”等，这些表达各有不同的特征。除了英语，俄语和日语中也有类似表述原因的方式。在句法制图框架中，我们假定这些原因成分出现在ReasonP位置。对这些原因成分在句中位置的研究，意在通过为其指派详细的制图结构，从而为句法制图学稍做贡献。此外，了解哪些原因成分在哪些语言中不会使用/没有语音形式也很重要。
05.
赵欣宇：有人认为，以“for”开头的原因表述总是在否定辖域之内（Endo，2015）。但在某些情况下，例如“John didn’t do it for love of his country.”，原因成分既可以在否定辖域内得到解释，也可以在否定辖域外得到解释。请问这一现象背后有什么样的深层次原因呢？
Yoshio Endo教授：这个问题很有意思。我尝试着将这一例句译成日语。然后我发现，在这句话中，可以很自然地将“for”改述为“because”。我的意见是，英语中可能有两种“for”，一种出现在比否定辖域高的位置，另一种出现在比否定辖域低的位置。根据这一假设，ReasonP中出现在高于否定辖域位置的“for”将会生成为“because”的含义。当同一个词有两种不同表现的时候，便假设存在两个不同的句法位置，Cinque（2006）对长被动句的分析中运用的也是这一方法。
English Version
01.
Xinyu Zhao: In your opinion, does linguistic logic have something in common with mathematical logic?
Prof. Yoshio Endo: As I discussed in Endo (2021b), my impression is that the logical structure of mathematics is reflected in the grammar of natural language and many grammatical items seemingly unrelated to mathematics seems to involve functional categories, where the functional categories may or may not be pronounced. For example, consider the commutative law of Boolean algebra:
a・b=b・a.
The nature of this law is reflected in the following sentence:
Harry [plays the lute](a) and [sings madrigals](b).= Harry [sings madrigals](b) and [plays the lute] (a) .
In contrast, in the case where the following sequence of events is represented, Boolean algebra property of a・b = b・a is absent:
I [went to the store] (a) and (then) [bought the whisky] (b). ≠ I [bought the whisky] (b) [went to the store] (a).
These differences become more apparent when looking at Ross's (1968) coordinate structure constraint (CSC), which forbids moving elements from one part of a coordinate structure like A and B. The CSC is only operative when a coordinate structure has the properties of Boolean algebra, as follows:
*the lute which Harry plays and sings madrigals is warped. (Ross 1986: 100) vs. Here is the whisky which I went to the store and bought. (Ross 1986: 103)
Let us explore this point further with a case study consisting of three sentences with motion verbs like come and go.
Come go eat with us.
When come and go are not coordinated by and, they seem to serve as a functional category. Cinque (1999, 2006) claims that when lexical items are used as a functional category, they are arranged in a hierarchical structure and thus cannot be swapped (=non-Boolean algebra property). In fact, the sequence of come and go we saw immediately above has a hierarchical structure and thus they cannot be swapped as follows:
Come go eat with us! vs. ??Go come eat with us. (Shopen 1971) (cf. Zwicky 1969, Bjorkman 2016)
As expected, the effect of coordinate structure constraint does not occur for the functional categories come and go that are not connected by and, as follows:
??What meal did you ask him to come and go and eat with us? vs. What meal did you ask him to come go eat with us? (Shopen 1971)
Cinque and Rizzi assume that the hierarchy of functional categories are universal. That is, even if two languages have different basic word orders, their hierarchical structure is the same. For example, unlike head-initial languages like English, Japanese is a head final language, so the linear word order of come and go of English and Japanese is mirror image. However, as seen below, the hierarchical structure of come and go of these two languages is the same, that is, come is more hierarchically higher than go.
[come [go [eat; tabeni ‘eat’] itte ‘go’] kuru ‘come’].
To summarize so far, we have seen that the CSC has a mathematical property that only applies to sentences with the Boolean algebra property. This point becomes even clearer when looking at the phenomenon of agreement, which is unique to natural language. For instance, consider the following sentences, recently discussed by Chomsky (2021).
a. John arrived early, met Bill and got a good seat. (three independent event)
b. To arrive early, meet Bill and get a good seat seems/*seem to be what John wants. (single compound event)
c. arriving early, meeting Bill and getting a good seat seems/*seem to be what John. wants. (single compound event)
d. To arrive early, to meet Bill and to get a good seat seem/seems to be what John wants. (single compound event, three independent event)
The difference here is whether the sentence with three verb phrases connected represents a single (sequential) compound event, or three independent events. To understand the nature of each sentence here, let’s first look intuitively at what is happening here. First, the nature of the tense of the three verb phrases seems to be important in determining whether the sentence is interpreted as a single compound event or three independent events. Stowell (1982) examined the nature of verb phrases with the infinitival to involved in the interpretation of control sentences to find that verbs with to have an unrealized tense, as follows:
John persuaded Mary to leave.
Refining this idea, Martin (2001), after careful observation of various sentences with the infinitival to in English, concludes that to has some tense property. Based on the fact that infinitival to has some kind of tense property and Chomsky’s sentences above have a three independent event interpretation when each of the three verbs is prefixed by the infinitival to, I suggest that there are as many events as there are tense features. Considering that a single tense feature represents a single event, an English construction with three infinitival to’s has three tense features and represents three independent events. For this reason, this sentence has three events and the verb agrees in the plural. (When these three events are taken as one event as a whole, as will be discussed later, agreement occurs in the singular.)
In contrast, if each verb phrase has an ing, it does not have a tense feature, as opposed to to. Therefore, if each of the three verbs is followed by ing, the sentence will have only one tense feature by borrowing one from the matrix clause. For this reason, when three verb phrases with ing are connected, the verbs agree in the singular.
Let's represent this somewhat more precisely with the nature of lattice. For example, suppose there is the partial order set of <dog, cat, animal>. Comparing the amount of information each element carries, two arbitrary elements are dog and cat and their common superior concept is animal, where animal is called “upper bound”. With this in mind, let us now recall Chomsky’s sentences. Here, we see set of events < arrive early, meet Bill, get a good seat >. If each of the three verb phrases has the element to, the three tenses will represent three independent events, and the verb in the main sentence will agree in the plural form. The upper bound expressing the whole event that subsumes the three events also has a tense, so the verb in the main sentence can also agree in singular.
In contrast, if each of the three verbs is followed by ing, then there is no tense, as there is no tense feature carried by the element ing in the three verb phrases. The upper bound expressing the whole event that subsumes verb phrases acquires a tense by borrowing one tense feature from the main clause. Therefore, one tense represents one event and the verb in the main sentence agrees in singular.
Next, let's look at what is missed in Chomsky (2020). Note that when the infinitival to occurs in three verb phrases, the verb phrases have the characteristics of Boolean algebra, and thus free to swap, as follows:
To arrive early, to meet Bill and to get a good seat seem/seems to be what John wants. = To meet Bill , to arrive early, and to get a good seat seem/seems to be what John wants.
In contrast, the case where ing occurs in three verbs has the properties of non-Boolean algebra, which is not interchangeable, as follows:
Arriving early, meeting Bill and getting a good seat seem/seems to be what John wants. ≠ Meeting Bill, arriving early, and getting a good seat seem/seems to be what John wants.
Thus, it can be seen that natural languages are divided into two types: those with mathematical characteristics and those without. If the functional category gives rise to human-specific sentences that do not have mathematical properties, then we can say that English ing is a functional category, while to, which has various meanings including unrealized tense, is not a functional category.
Incidentally, when I discussed this topic in my graduate class, a Chinese student told me that the same difference between ing and to in English is expressed in Chinese by the presence or absence of the element “了”, as shown below.
吃饭，唱歌和跳舞（separate event）
to eat, to sing and to dance
吃了饭，唱了歌，然后跳了舞（single event）
eating, singing and dancing
02.
Xinyu Zhao: How do you think of the role Leibniz law and Boolean logic play in linguistic studies? Could you elaborate on that?
Prof. Yoshio Endo: The commutative laws of Boolean algebra and Leibniz’s theorem are similar in that they both involve the operation of swapping elements, but they differ in their scope of application. For example, Leibniz's theorem states that if A = B, then whatever A is true is also true for B and vice versa. For instance, consider the following;
a. Remi(A)=Yoshio’s 13-year-old dog(B)
b. Remi(A) sleeps well.
c. Yoshio's 13-year-old dog(B) sleeps well.
(1a) is true. (1b) is also true and (1c), replacing Remi and Yoshio's 13-year-old dog, is also true. Here, the characteristics of the elements are revealed by the operation of swapping elements in the non-coordinate structure.
In contrast, the commutative law of Boolean algebra comes into effect in the coordinate structure, as we saw above.
03.
Xinyu Zhao: In your opinion, what’s the relationship between external syntax and internal syntax of adverbial clause?
Prof. Yoshio Endo: As I discussed above, internal syntax deals with the internal structure while external syntax deals with the connection between the element with its internal structure and the main clause. Endo and Haegeman’s paper (open access) propose an operation called “adverbial concord merge” that links these two. The gist of the operation, as described above, is that when a movement occurs from a certain position (e.g., aspect) in an adverbial clause, the moving element determines what the entire adverbial clause is (e.g., aspect type), and the determined element (e.g., aspect type adverbial clause) is then linked to the corresponding element in the main clause (e.g., aspect element). Here, the concept of “concord” is important, where concord means that a restriction can be seen between two elements. The intuition there is that because there is a restriction, the two elements are strongly connected. For example, in British English, the restriction is that the element since is associated with the perfect aspect of the main sentence rather than past/present tense, as seen below. In this case, the element since concords with the perfect aspect of the main sentence.
I have been living in London since I was a child.
*I live in London since I was a child.
*I am living in London since I was a child.
Now, as for the connection between an adverbial clause and the associated main sentence, Endo and Haegeman’s idea will make an interesting prediction. When an adverbial clause is associated with a higher outer element, it should occur before another lower inner element. This point is easier to understand by the following diagram, where the high outer adverbial clause (1) is associated with the high outer functional head (FH(1)), and the low inner adverbial clause (2) is associated with the low inner functional head (FH(2)), so that the higher outer adverbial clause (1) occurs before the lower inner adverbial clause (2).
[high Adverbial clasue(1) [low adverbial clause (2)... FH(2)] FH(1)].
There are a lot of Chinese students in my graduate school, so I checked with them to see if this point holds in Chinese, and let’s see the results. First, in Japanese, the functional category expressing the mood of the speech act occurs outside the functional category expressing the speaker’s mood, so the corresponding adverbial clause also occurs before it. And this seems to be the same in Chinese, as we see below.
要我说（A-mood）虽然可能性很低（S-mood）但可能会落选。vs.?虽然可能性很低（S-mood）要我说（A-mood）但可能会落选。
In many other cases, Japanese and Chinese seem to show the same adverbial clause restriction, as follows:
事实上（A-mood）上田先生（top）秋天可能会搬家。vs. ?上田先生（top）事实上（A-mood）秋天可能会搬家。
大学毕业并继续攻读研究生课程时（T），我一边保留了学籍（Asp），一边担当非营利组织的董事长。vs. ? 我一边保留了学籍（Asp），大学毕业并继续攻读研究生课程时（T），一边担当非营利组织的董事长。
Whether Chinese has the same functional category and the same pattern of arrangement as Japanese remains to be seen, but since the same restrictions on adverbial clauses found in Japanese are also found in Chinese above, it seems that the same functional category arrangement is present in Chinese as well. And if there are differences between the Japanese and Chinese adverbial clause sequences, it means that the functional category sequences are slightly different between the two languages. This interchange of functional categories is also present in European languages, as Rizzi describes.
04.
Xinyu Zhao: In English, there are different forms of wh-expressions asking for a reason. Are similar phenomena available in other languages? How can studies of wh-expressions asking for a reason contribute to cartography?
Prof. Yoshio Endo: In English, there are several question expressions that ask reason like why, what for, how come, etc., each with different characteristics; the language in which the expression corresponding to what expresses reason is Russian and Japanese, in addition to English.In cartography, these reason expressions are assumed to occur in ReasonP. Studying where these reason expressions occur in sentences is expected to contribute to cartography because we can create a detailed reason expression map here. It is also important to know which reason expressions are not used/pronounced in which languages.
05.
Xinyu Zhao: There is a belief that reason expression headed by for is always inside the scope of negation (Endo, 2015). But on some occasions, such as “John didn’t do it for love of his country.”, it can be interpreted both inside and outside the scope of negation. Are there any deep reasons behind this?
Prof. Yoshio Endo: This is an interesting question. I have tried to translate the sentence you pointed out to me into Japanese. Then I found that it is natural to paraphrase for with because. Then I suggest that there might be two kinds of for, one occurring higher than negation and the other lower than negation. In that case, the for occurring in ReasonP in a position higher than negation would be naturally interpreted as because. This idea of assuming two syntactic positions when the same word behaves in two different ways is also found in Cinque’s (2006) analysis of long passives.
往期推荐
知名语言学家陆俭明教授接受CCTV-13 “吾家吾国”栏目独家专访
Chris Collins | A Conversation with Noam Chomsky
Andrea Gajardo-Vidal博士、Diego Lorca-Puls博士访谈
Luigi Rizzi & Ángel Gallego | Phases & Minimality（语段与最简性）
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 聂简荻 丁子意 郭思源
排版：闫玉萌 赵欣宇 雷晨
审校：李芳芳 田英慧
英文编审责任人：雷晨