作者：祁小雯
出版社：上海交通大学出版社
格式: AZW3, DOCX, EPUB, MOBI, PDF, TXT
最简方案框架下英语小句中名词合法性研究(英文版)试读：
前言
本书在最简方案框架下研究英语小句中名词性谓语的合法性问题，解释小句中位于谓词位置的非论元名词项合法性的检验机制。旨在进一步加强生成句法理论的一致性，并且验证最简方案解释的充分性。
现有针对小句名词性谓词的格检验研究主要观点分为4类：无格假设、默认格假设、一致格假设和结构格假设，但都无法充分解释英语小句中名词性谓语的格位获得机制。最简方案中所有词项都带有由语项决定的特征进入句法运算，名词项则带有一致特征和格特征，其中一致特征在名词项上属于可解释特征，而格特征属于不可解释特征。不可解释特征在推导过程中必须通过一致操作被删除，否则到达语义接口的表达式得不到充分解释，会导致推导失败。因此，处于小句谓词位置的名词项的合法性应该得到应有的重视并在现有句法理论框架下得到解释。
本书认为，由于谓语名词的格地位在最简方案理论框架中得到证实，也应该可以在同一理论框架中得到解决。并且最简方案对语言推导过程的新认识为我们解决这一问题提供了理论基础和技术手段：首先，由于名词在进入推导时就带有格特征，因此不需要在结构中存在格来源，只是需要在带有相应格特征的功能节点处进行特征核查；其次，根据特征核查理论中的多重一致操作机制（Multiple Agree）和特征传递机制（Feature Inheritance），语段中心语所携带的核查特征可以在语段中传递，这样就增加了可以供谓语名词进行格特征核查的功能节点数量；第三，多层指示语结构（Multiple Specs）允许在中心语的核查域内嫁接外层节点，扩大了单个中心语的核查域范围。这些理论基础和技术手段为在最简方案框架下进行谓语名词的格检验提供了保障。
本书还从句法和语义两个方面讨论了小句结构特点，认为小句是一个功能中心语F的最大投射FP，小句中的功能中心语F 为Pr。小句中的主语位于[Spec， PrP]，谓语位于功能中心语的补足语位置。PrP短语被主句动词所选择，与主句动词结构vP组成一个语段。
本书目标读者为广大高校语言学专业教师及句法学方向研究生。Abbreviation List
ACC Accusative
A-CT Agr-based Case Theory
AGR Agreement
A-less CT Agrless Case Theory
the A-P system the articulatory-perceptual system
C the central computational systemHL
the C-I system the conceptual-intentional system
DAT Dative
DS the deep structure
ECM the Exceptional Case Marking
FP Functional Phrase
GB Government and Binding
G-CT Government-based Case Theory
IL-predication individual-level predication
LF logic form
MP the Minimalist Program
NOM Nominative
NP Noun Phrase
NSC nominal small clause
PF phonological form
PPC the Percolation Principle of Case
PrP Predication Phrase
SC small clause
SL-predication stage-level predication
SS the surface structure
TP Tensed PhraseChapter OneIntroduction1.1The Anomaly of the Nominal Predicate Licensing in Small Clauses
The main concern of this book is the nominal predicates in the following so-called “small clauses” selected by consider-type verbs, see (1).
(1) a. I consider John (as) a teacher.
b. The general considered the attack (as) a failure.
c. The mother considered her son (as) a genius.
d. John considered Bill (as) a fool.
The term “Small Clause” (SC, henceforward) “refers to a string of XP YP constituents which enter into a predication relation, but the predicate YP, rather than containing a fully Inflected verb, contains an adjective phrase, noun phrase, prepositional phrase or an uninflected verb phrase” (Basilico, 2003). Examples in (1) contain a small clause whose subject is predicated by a nominal item. Hence, these kinds of small clause is named nominal small clause (NSC, henceforward).
Within the standard Case Theory as presented in Chomsky (1981, 1986b), examples like those in (1) pose a problem in terms of the θ-criterion and the Case-filter, which function together as the basic principles determining the distribution of NPs. For the purposes of this research, we shall first assume the following formulation of these principles, see (2):
(2) a.θ-criterion: Each argument bears one and only one θ-role, and each θ-role is assigned to one and only one argument.(Chomsky, 1981:36)
b. Case-filter:*NP if NP has phonetic content and has no Case.(Chomsky, 1981:49)
Taken together, these two principles impose a requirement that every NP, which is phonetically realized and has semantic content, must be assigned both a θ-role and Case, either directly or via a trace which it binds.
There are three NPs in each sentence in(1), the main subject, the subject in the embedded clause and the nominative predicate of the SC. Then main subjects are licensed perfectly as the external argument of the main verb and marked nominative case by the Infl (For convenience sake, the Split Infl will not be discussed in detail in this section, and will be considered in detail in later parts). The embedded subjects are licensed by being marked accusative case as the main verb consider is so-called ECM verb (Exceptional Case Marking verb), which selects an infinitival IP clause as its complement and case-marks the overt NP at the subject position in the embedded clause with a deficient T head. Semantically, the embedded subjects are assigned a θ-role by the predicates in the small clauses.
The story goes well until now, but when we examine the licensing of the third NP in the construction, the nominal predicate, problems rise. Semantically, there is no constituent that can assign a θ-role to it. It may be argued that as a predicate, this NP does not need a θ-role but rather assigns a θ-role. But there are two reasons to argue against this idea. First, at certain stage of derivation, two NPs, and only two NPs bearing the same categorial properties of [+ N,-V], interpretable φ feature set and uninterpretable Case feature, are selected from the Numeration and enter the derivation. Which one should be selected to assign θ-role during the computational process and which should be the one to be in the argument position? It’s implausible to assume the computation system select randomly because the sentences in (3) are not acceptable (hence the asterisk).
(3) a. *I consider a teacher is John.
b. *I consider a teacher is him.
It is too hush to say that the ungrammaticality in (3) results from that John or him is pronominal and a teacher can denote one’s job and show certain property of a person, and so the pronominals should be selected as subjects, and the rest predicate. The evidence against this argument lies in (1)-b, repeated as (4) here:
(4) a. The general considered the attack a failure.
b. *The general considered a failure the attack.
c. *The general considered the failure a attack.
It’s obvious that both attack and failure are not pronominals but attack cannot be the predicate no matter it’s modified by definite or indefinite determiner. So it’s worth probing into the semantic and syntactic rules that license an NP as a predicate.
Second, given that the predicate NP is licensed by the ability to assign θ-role to the subject NP and that the subject NP is licensed by being assigned this θ-role by the predicate, it seems that a licensing circle is formed as in (5) and a problem of circularity emerges.
With the circularity licensing pattern in (5), we cannot tell which item licenses the other. If we regard the ability to assign θ-role to the subject NP as the licensing of predicate NP, then the prerequisite is that the subject NP should have been licensed by a licensed predicate. But the subject NP is waiting the would-be-licensed predicate NP to license it. So the ability to assign θ-role should not be regarded as the licensing of the predicate NP but the ability of predicate following the licensing. Therefore it is rational to assume that there is a mechanism in universal grammar to put a teacher at predicate position and John the subject position. We assume the mechanism is closely related to the nature of predication relations and the properties of different kinds of nominals. All the syntactic items like “teacher”, “a teacher”, “the teacher”, “many teachers”, and “all teachers” bear the same category “NP” (or “DP” along with the DP Hypothesis), but they have various internal structure, semantic references and syntactic behavior as follows. For example (6):
(6) a. I consider him a teacher.
b. *I consider him the teacher.
c. I consider them teachers.
d. *I consider them many teachers.
e. I consider them all teachers.
Sentences in (6) show that the change of modifiers or determiners in NP phrases leads to the difference in grammaticality. We assume here that different determiners change the properties of the nominals and only certain kinds of nominals can be the predicate and trigger the predication relation in SC. This problem will be examined closely in Chapter 4.2.5.
Now, we turn to a more syntactic problem of the nominal predicate, the Case-marking problem. For convenience sake, we repeat (1) as (7):
(7) a. I consider John as a teacher.
b. I consider John a teacher.
In (7), the most possible case marker of predicative NP is the particle as preceding the predicate NP. Some linguists like Haegemman (1994) believe that the presence of as is the evidence of the existence of the functional head Agr in small clause, so small clause is a maximal projection of this head. As the head of the project, Agr selects the predicate NP as its complement and case-marks it naturally. But this assumption is not tenable for three reasons. First, the notion that a nominal item enters computation without Case feature and the case is assigned by a case-assigner has been discarded under the framework of Minimalist approach. Case-assignment mechanism within GB framework heavily depends on the notion of government. But the notion has been doubted in MP. Under the framework of MP, case-marking is conducted in a totally different mechanism named feature checking. Second, there are lots of disputes on the internal structure of small clauses. Linguists have yet come to an agreement about whether small clause is a maximal projection or not, which means it’s still a question whether there is a head in small clause, let alone the category label of it. So it seems that we should not rush into this case assignment pattern. Third, if we assume that “as” assigns case to the predicate NP, the situation in (7)-b is worse because the only potential case-assigner is missing. But we don’t take it as the end of the story. On the contrary, it’s the beginning of the story, because it means that the predicate NP is not licensed by “as” in case-assignment patterns but by some other syntactic mechanisms.
But before we probe into the syntactic mechanism licensing predicate NP, we need to tackle with a more fundamentally conceptual question: whether it is necessary for predicate NP being case-licensed since it is not an argument and bears no θ-role.
In the framework of Government and Binding, two views of Case have developed: the original Case Filter proposed by Rouveret and Vergnaud (1980) and Chomsky (1981) as listed in (1)-b rules out caseless noun phrases with phonetic content, regardless of their thematic status. However, this original view was quickly replaced by Chomsky’s (1981:336) alternative version which derives the Case Filter from the Theta Criterion and only applies to θ-marked noun phrases. This revision of the Case Filter became known as the Visibility Condition since case marking is treated as a condition making a noun phrase visible for Theta marking. Since most attention has been paid to argument nominals, it is not clear whether the theory extends easily to predicative phrases. However predicative NPs exist substantively not only in SC but also in copular sentences coming after copula “be”. In copular sentence, the overt copula “be” is taken as the accusative case assigner of the predicate NP (Maling and Sprouse, 1995). But it cannot account for the linguistic fact that predicate NP in SC is licensed without any overt case-markers.
Pereltsvaig (2008) pointed out the problems of the revised version of Case theory when studying the intra-structure of Russian copular structures. The first problem with the Visibility Condition is that, being an LF principle, it does not easily relate to the observable morphological case. The second problem is that it is not relevant to case marking on post-copular phrases since presumably post-copular nominals are not marked and therefore need not be marked with Abstract Case either, according to the Visibility Condition. Yet, at least in some languages with morphological case systems–including not only Russian but also Icelandic, Arabic, Korean, Finnish and others–post-copular phrases clearly bear morphological case marking, which a grammar system including the Visibility Condition remains unaccounted for. The solution proposed within the framework of the Visibility Condition is to treat such case marking as “a morphological default in the absence of syntactic case resulting from an independent morphological well-formedness requirement on nouns” (Pereltsvaig, 2007:105). However, this approach fails to account for two uncontroversial facts: first, some case-marking on post-copular phrases is not easily analyzed as morphological default, such as instrumental case in Russian and accusative case in Arabic; second, at least some case-marking on post-copular phrases has been argued to indicate certain syntactic configurations (Lee, 1989; Maling and Sprouse, 1995; Comrie 1997). This leads some researchers to challenge Visibility Condition as incorrect or insufficient, proposing to go back to the original Rouveret-Vergnaud formulation that Case serves as licensing mechanism for all and only overt noun phrases (Lee, 1989).
Besides the empirical data of the morphological case found on predicate NPs, we also encounter a theoretical barrier in assuming the caselessness of predicate NPs. Within MP framework, all lexical entities enter computation with features, including categorial features such as [+ N,-V] , or [+ V,-N] and the features determined by categorial features. In this system, nominal items bear φ features and Case features along with categorial feature N. And for a nominal, categorial and φ features are [+interpretable] , because as a noun it inherently expresses properties like number, gender and person. But Case feature is [-interpretable] because we cannot tell the case of a noun by itself. Case feature must be checked by an interpretable matching Case feature in a functional head for the nominal item to be interpreted at LF and hence the derivation converges. In this system, all nominals need to be case-licensed by checking the [-interpretable] Case feature, including NPs on predicate position. So, theoretically speaking, to exclude predicate NPs from Case Filter which is assumed to explain the distribution of nominals is a great loss in terms of the uniformity and the explanatory adequacy of generative grammar. And the licensing mechanism of the predicate NPs requires a theoretical account.
Therefore, we stick to the previous version of Case Filter and assume that any overt NP, including the non-argument predicate NPs, must be ruled out if it is not properly case licensed. So the gap between linguistic empirical data and linguistic theory we found in NSCs triggered this study and we try to present a solution within the framework of MP.1.2Four Puzzles in NSCs
First, nominal small clauses show lower acceptablity than finite clauses and infinitives. This is found in the Questionnaire on Acceptability of English Small Clauses (See Appendix) among English native speakers. The latter two structures are ranked between “perfectly acceptable” and “quite acceptable” with the average score of 4.53 and 4.33 respectively. But nominal small clauses are “just acceptable” with the average score of 2.9. Then can we attribute this degradation in acceptability to the anomaly in small clauses?
Second, there is a cross-linguistic distribution gap. In other Romance languages, like Italian, French and German, nominative small clauses are also found. Italian examples in (8), French examples in (9) and German examples in (10):
(8) a. Considero Jim (come) un insegnante.
I consider Jim (as) a teacher.
b. Considero Mary (come) una insegnante.
I consider Mary (as) a teacher.
c. Li considero (come) insegnanti.
I consider them (as) teachers.
(9) a. Je considère Jim comme un enseignant.
I consider Jim as a teacher.
b. Je considère Marie comme une enseignante
I consider Mary as a teacher.
c. Je les considère comme des enseignants.
I consider them as teachers.
(10) a. Ich betrachte Jim als Lehrer.
I consider Jim as a teacher.
b. Ich betrachte Mary als Lehrerin.
I consider Mary as a teacher.
c. Ich betrachte sie als Lehrer.
I consider them as teachers.
In French and German, there is no such bare nominative small clauses without particle comme/come. With this distributional advantage, we unavoidably come to the question that whether these as/come/comme constituents indeed help case-mark the following predicate. And what syntactic role does it play in small clauses or does it function more semantically? The answer to these questions may shed some light on the research and help us look into the syntactic and semantic functions of this adverb and nominative small clauses.
Third, with the previous distribution advantage, we may assume nominative small clauses with as should be more general and acceptable. But it is not the case. In the acceptability questionnaire, nominative small clauses with as in it shows no higher acceptability than bare nominative small clauses as expected．In fact, the average score of nominative small clause with as is 2.1, lower than that of bare nominative small clauses 2.9. This result leads to a confusing contradiction with the cross-linguistic distribution gap of bare nominative small clause and the one with as in it.
Last but not the least, the selection of main verbs influences the distribution of nominative small clauses. The constituent called “small clause” was originally identified as the complement of consider-type verbs (Moro, 1997:262). Other consider-type verbs like believe and think can also select small clause as the complement but present some distributional difference. Look at sentences in (11) and (12):
(11) a. I believe John innocent.
b. I believe John in the park.
c. ?I believe John a teacher.
d. *I believe John as a teacher.
(12) a. I think John innocent.
b. I think John in the park.
c. *I think John a teacher.
d. *I think John as a teacher.
(11) and (12) show that believe and think can select small clauses with AP or PP predicates but not small clauses with NP predicates. This is also proved in the acceptability questionnaire. While nominative small clauses selected by consider are ranked as “just acceptable”, those selected by believe and think are ranked approaching “not acceptable at all” with an average score of 1.19. All the three words, consider, believe, and think are categorized as psych cognitive verbs, expressing static cognition activities. And they bear the same theta-grid as in (13):
Semantically, they can assign a theta role of theme to the small clause, and syntactically, they can select a small clause as the complement. It goes so well when they select small clauses with AP or PP predicate but becomes problematic when it comes to NP predicate. Compared with AP and PP counterparts, the most prominent feature of NP predicate is that overt NPs need to be case-licensed. So what semantic or syntactic features of the main verb lead to this difference?1.3Our Proposal
We propose to account for the licensing mechanism of predicate NPs within the framework of Minimalist Program applying the feature checking operation. First, we assume in line with the conceptual idea of MP, that all NPs enter derivation bearing categorial feature [N] and the following canonical features such as [+interpretable] φ features and [-interpretable] Case feature. The derivation will converge at LF interface only when all the [-interpretable] features in the structure checked and deleted, including the Case feature of the predicate NP. Second, the prerequisite of this mechanism is that small clauses are analyzed as a constituent, a maximal projection of a functional head Pr (Predication). Predicate NP merges as the complement of the head and subject NP at the Spec position. Then the PrP (Predication Phrase) merges with a higher phrase VP and is contained in the v P phase. In this phrasal structure of small clauses and the main clauses, the co-operation of Multiple Agree, Case Inheritance and Multiple Specs in feature checking mechanism provides a way for predicate NPs getting Case feature checked, and hence licensed.1.4Significance of the Study
As we have mentioned previously, this research about the licensing mechanism of nominal predicate is theoretically important because excluding predicate NPs from case-licensing does undermine the uniformity and explanatory adequacy of the generative grammar.
First, the uninterpretable Case feature of every NP, whether argument or not, must be admitted, because all the syntactic operations in the framework of Minimalist Approach are motivated by features. It means, if the Case features of predicated NP are denied and NPs in this position are assumed caseless, those NPs are inactive in minimalist analysis and cannot participate in derivation. It’s not plausible to assume that an inactive item enters into a derivational system where most of the derivational operations involve Move. If we assume a predicate NP enters derivation without Case feature, a different derivational mechanism is required to account for the process
试读结束[说明：试读内容隐藏了图片]