| English middle construction, which involves a complex interplay between syntax and semantics, is one of the most heatedly discussed topics in linguistic study. The formation of English middles is particularly appealing to most scholars in the literature. Previous studies are divided on this issue, generating two approaches, namely, the lexical approach and the NP-movement approach. This thesis takes a VP-shell approach, aiming to provide a unified account on English middle formation, regardless of typical middles, which are heatedly discussed by both the lexical approach and the NP-movement approach, or the atypical middles which escape notice by both approaches. Unlike traditional analyses, we start with middle semantics but take a syntactic approach, thus capable of incorporating merits of both sides while discarding their defects.The exploration I made in this thesis is manifested in the following 4 aspects.First, we have established a working definition of and assigned a new classification to English middle constructions. English middles have long been studied, but no consentaneous definition has ever been put forward, whether accurately or roughly. In this thesis, we establish a working definition as follows: The English Middle Construction refers to that type of construction [NP V ADV] in which (properties of) NP cause the V-ing (of NP) to be Adj, middle verb is active in simple morphology and there is usually an adverb or other adverbial phrase working as complement of the verb. This is a rough definition but it captures the apparent properties of English middle constructions in terms of semantics and syntax. It will serve as our guideline in discussing English middles. Following this definition, we assign a new classification to English middles into three types, namely, Class 1 middles, Class 2 middles and Class 3 middles. Traditional approaches set a dichotomy between typical middles and atypical middles, which is too ambiguous in terms of criterion, ours is based on a unified criterion - the subcategorization features of middle verbs. The working definition and the new classification will free us from controversies and categorize as many middles as possible under our unified account.Second, we have defined Responsibility property of grammatical subject (Responsibility condition) as the key restrictional factor for English middle formation. We take a syntactic approach, but we take into account middle semantics. Among the semantic properties, the responsibility property of the grammatical subject is singled out as the most fundamental one. We have combed out various restrictions on middle formation. Responsibility condition is a core condition working as a necessity to license middle formation. Aspectual condition, selecting for possible middable verb, Anti-effected condition, a corollary to Responsibility condition and Agentivity condition, a supplemental condition to Aspectual condition, they are all peripheral conditions in order for middles to be possible. Because they are all peripheral conditions, they are more diversified. Just like computer configuration, CPU is the core, while peripherals may vary greatly from a scanner, a webcam to a projector or whatever. The core CPU is the most fundamental, but the peripherals may be dispensed with. Even all the configurations, whether the core or peripherals are all available, the computer still may not work, for it may break down for various reasons. This may account for why some middle sentences still crash even when all the conditions are met. Their relationship can be diagramed as in (3):Third, we have parameterized the agentθ-role in English middles as an overt for-PP phrase and a covert IMP. Despite great disputes among scholars on the middle formation, there is one thing that most scholars agree on - the presence of an implicit agent. The lexical approach holds that it only exists at the level of lexical-conceptual semantic structure. It is not projected into the syntactic structure by combining with any lexical word to become an argument, because the agent in most cases bears an arbitrary interpretation. This approach overlooks the presence of overt agent argument and fails to account for the binding of subject-contained anaphors. The NP-movement approach is divided on the status of the implicit agent - pro, PRO or whatever. Stroik (1992) has proposed that it's PRO adjoining to VP. But his proposal is theoretically flawed, because in English PRO invariably occupies [Spec, T (-finite)] position and thus incompatible with adjunction. We stand with him that the implicit agent is adjunction but disagree with him in that the implicit agent, by our analysis, is IMP adjoining to vp rather than PRO. Under our analysis, the syntactic puzzles such as subject-contained anaphors and adverbial effect can be accounted for in a package fashion. IMP adjoines to vp and higher than DP contained in the subject (Under VISH ), so it can c-command and bind the anaphor. As for the adverbial effect in English middles, adverbs need to be agent-neutral and are therefore at VP-adjoined position under our analysis. We believe that they are two sides of the same coin. IMP adjoining to vp rather than VP is thus too far to exert (or some element blocks) its agentive quantification over the adverbs adjoining to VP, thus the adverbs agent-neutral; on the other hand, agent-neutral adverbs force the implicit agent to adjoin to vp, leaving only VP for themselves to adjoin to. That is to say, agent-neutral property of adverbs can be inferred from the tree diagram (as in 1). This reflects that our analysis is along the right line from one perspective.Fourth, we have formulated a unified account on middle formation, inclusive of Class 1 middles, Class 2 middles and Class 3 middles as well. The line along which we proceed with our analysis is as follows. Following Responsibility condition, we conclude that middle subject is base generated; the semantically perceptible, but syntactically covert causative verb is a light verb (v), laying the foundation of our VP-shell analysis; the implicit argument is not PRO, but IMP, a vp-adjunction position at the syntactic level, thus defining its satatus at the syntactic level; the internal argument of middle verb can be either suppressed (if any) or structurally discharged, providing a general picture of the subcatgorization feautures of English middle verbs. Therefore, we have got the general picture for the derivation of English middles, as in (4):Our analysis is far from perfect. Our analysis is mainly limited to English and thus lacks cross-linguistic evidences.Therefore, we hope that, in the future, we can have more time and chance to pursue this subject further for a better account.
|
PDF Full Text Request