管辖约束理论(语言学)

Government Binding TheoryRobert N. St. ClairUniversity of LouisvilleIntroductionThere were a series of major developments leading to the rise of Government and Binding Theory. In the atavistic version of syntactic theory which was published as Syntactic Structures (1957), Chomsky argued for syntax as the basis for correlating the linguistic meanings (the semantic component) with linguistic forms (the phonological component). At that time, syntax consisted of three kinds of rules: (1) The phrase structure rules ordered the parts of a sentence into linguistic categories and provided the lexical forms for nouns, verbs, prepositions, and adjectives; (2) the transformational rules operated on deep structures and reordered the phrase structural forms; and (3) morphophonemic rules merely changed lexical forms where necessary ( e.g. go + past = went). The phrase structure Rules created the deep structures of sentences. The transformational rules operated on these deep structures to produce the surface structures of the language. These transformational rules were ordered.By 1965, the model was greatly expanded and revised. Aspects of the Theory of Syntax by Chomsky included a detailed account of the semantic component (projections rules, global insertion rules, selectional restrictions, rules for the interpretation of subcategorization, and semantic distinguishers). The most innovative work on the semantic component was done prior to this time as evidenced in the publication of An Integrated Theory of Linguistic Descriptions (1964) by Jerrold Katz and Paul Postal. This was enhanced further by the co-authored work on The Structure of Language (1964) by Jerrold Katz and Jerrold Fodor. The phonological component was also greatly enhanced the inclusion of underlying phonemic forms, ordered rules, and phonetic outputs. The most definitive work on the phonological component can be found in The Sound Patterns of English (Chomsky and Halle, 1968). This publication provided the theoretical framework for a universal theory of phonology (distinctive feature theory, the principle of stress rules, the phonological cycle, phonological constraints, etc.).The aforementioned model became known as The Standard Theory. By 1972 more revisions in the model took place and this led to a renaming of the revised standard theory. It became known as The Extended Standard Theory. There were several reasons for these revisions. What was referred to as the Semantic Representation of a sentence was no longer seen as a single, uniform structure. The syntactic component interacted with the semantic component of the language many times during the processing of syntactic structures. Before the application of transformational rules, for example, the deep structure went to the semantic component where it was interpreted in terms of its functional structures. At this time, the semantic component provided information on the interpretation of the various semantic roles used in the sentence such as agent, patient, goal, experiencer, etc. This deep structure continues to be processed syntactically as it goes through the various cycles of transformational rules. This information is again turned over to the semantic component for further interpretation. This time the semantic component provides information on modal structures such as the scope of negation, and the interpretation of quantifiers in the language. Another kind of semantic information developed at this stage includes the establishment of a table of coreferences in the sentence being analyzed.Mary saw herself in the mirror. (co-referential)Mary saw her in the mirror. (not co-referential)Mary thinks that she is attractive. (no specification of referentiality)These rules, it should be noted, could not be processed at the deep structure level and the process had to be delayed until certain modifications and rule applications had taken place within the transformational component of the language. Finally, at the surface structure level, the output of the final transformational cycle is sent to the semantic component for further processing. This time the semantic representation is analyzed for focus, presuppositions, topicalization. The presupposition of a declarative sentence has to do with what the speaker assumes the hearer knows. Focus, on the other hand, has to do with what the speaker assumes that the hearer does not know. Focus is symbolized in the following sentences by means of bold-faced capitalized words. Such words have extra heavy stress.Mary drank the MILKPresuppositions: Mary exists, Mary did something, Mary drank something.Focus:She drank milk and not water.MARY drank the mild.Presuppositions: same as beforeFocus:Mary drank the milk and not John. John saw MaryOld Information: We know that John did something. New Information: We did not know that he saw Mary.Mary was seen by JohnOld Information: We know that Mary was seen.New Information: We did not know who saw her.Ray Jackendoff has been instrumental in the development of the Extended Standard Theory. The arguments for the current revisions are stated in Semantic Interpretation in Generative Grammar (1972). One of the concerns that he voiced during these revisions was the need for autonomous syntax. He proposed that transformations should be applied without having to mention semantic information such as referentiality within a table of coreference, the use of index markers, etc. This could be done, he argued, by changing the kind of information allowed in the deep structure in the Standard Theory. Consider, for example, the rules of reflexivization, pronominalization, and Equi-NP deletion. In each of these cases, full lexical forms are assumed to exist at the deep structure level. In reflexivization, John saw John is transformed into John saw himself. This requires semantic information on coreferentiality. In the case of pronominalization, coreferentiality is also needed to transform Mary wonders if Mary will be happy into Mary wonders if she will be happy. In the case of Equi-NP Deletion, coreferentiality is also need to transform Mary expects Mary to win into Mary expects to win. One could avoid these references to the semantic component, Jackendoff argued, if pronouns and dummy subjects (gaps) already exist at the deep structure level. This introduction of abstract elements and empty categories into the deep structure of sentences marked an important turning point in linguistic theory. It led to the emergence of GB Theory. Nevertheless, before turning to this major revision based on the premise of autonomous syntax, it is important to review the representation of the model of the Extended Standard Theory. This is the model of grammar without autonomous syntax.The development of Government and Binding Theory as a modular model really began in 1977 when Chomsky and Lasnik proposed some major revisions in the Extended Standard Theory. In this article on Filters and Control (Linguistic Inquiry 8.3), they questioned the necessity of phrase structure rules in the context of lexical subcategorizational rules in the lexicon which provided similar information. They also addressed the question of transformational rules with regard to stylistic rules versus meaning changing rules. The issues addressed in this early article were further addressed by Noam Chomsky in his writings on Lectures on Government and Binding (1981) and Some Concepts and Consequences of the Theory of Government and Binding (1982). He became less concerned with regard to the base generability in transformational grammar and more concerned with Structure-Preserving Constraints on language. The revisions led to a modular model of transformational grammar.Since the concepts of Deep Structure and Surface Structure were substantially revised by the inclusion of abstract elements, trace elements, and empty categories, they were given a new nomenclature, d-structure and s-structure respectively. These are related to each other by means of movement rules. S-structures were further developed into Phonetic Forms or Logical Forms. What is called phonetic form involves more than mere acoustic and articulatory information. It included semantic properties, and low-level transformational rules (such as stylistic rules), deletion rules, contraction rules and phonological rules. The Logical Form Component deals with the meaning of sentences.The distinction between PF and LF components was necessary because they represented very different aspects and concerns within grammatical theory. The LF component deals with semantic information.PF Component: John likes MaryLF Component: For every x, x is a person, Mary likes x.Concerns about Binding Conditions are dealt with in the LF Component because they involve semantic issues (referential dependencies, coreferences, etc.). Similarly, quantifier raising dealt with semantic issues within the context of the LF Component. This component began by representing information in terms of standard logic, but was modified to incorporate constituent command, trace theory, and other linguistic issues.The differences between a d-structure and an s-structure captures what changes have taken place once a movement rule has applied. But GB Theory involves a lot more than merely revising the deep structure of a grammar. In includes many new features:D-Structure, Fully Recoverable D-Structures, S-Structure, X-Bar TheoryMovement Rules, Trace Theory, Empty CategoriesCase Filters, Theta Theory, Well-Formedness Conditions,Consider how the surface structure sentence Did John invite Mary? was dealt with in the extended theory of syntax.Deep Structure: John past invite MaryQuestion past John invite MaryAffix Hopping Do Support do+past John invite MarySurface Structure: Did John invite Mary? In the GB Theory, movement rules are severely restricted. One can only move elements to unfilled (empty) or to certain designated categories . The sentence is now represented as an inflectional phrase (IP).One of the movement rules takes the INFL form (past) and moves it into only open and empty category of Complement. In the event that the INFL is not moved, Do is deleted by means of a Do Drop Rule. Within this model, there are three kinds of movement rules. One that moves NPs (maximal projections), another that moves Heads of Phrases, and a third that moves Wh-constructions.Another major change that took place in the transition from the Extended Standard Theory to GB Theory has to do with the shift from rules to representations. In the earlier models of transformational grammar, one accounted for grammatical changes by means of syntactic rules.Deep Structure: John past see MaryPassive Mary past be+en see by JohnAffix Hop Mary past+be see+en by JohnSurface Structure: Mary was seen by John Such a rule had to be language specific. Only English, for example, insert be+en before the main verb and insert by before the agentive noun phrase. If transformational rules are to be universal, they cannot also be language specific. Hence, the system has to be changed. What is needed, Chomsky argued, is a theory that relies on conditions on these representations. These conditions operate at all levels and employ general rules such as:MOVE NPThis change became known as the Principles and Parameters Approach. Later, it came to be called Government and Binding Theory. This shift from rules to conditions on representations avoided language specific rules. It also strengthened the concept of the language acquisition device (LAD) in psycholinguistics. In the earlier model, a child was given a toolbox and was asked to construct a grammar for himself. He has some idea of what a grammar should look like, but he has no idea about universal principles govern all grammars. This could happen only if he were provided with a set of principles which defined the parameters and the conditions operating on syntactic representations. This is exactly what GB Theory attempts to do. This is why the shift from rules to conditions on representations had to take place. It provides a child with the principles and the conditions on natural languages and allows him to set the parameters in that language such as SOV or VSO word order, etc.CATEGORIES AND PHRASE STRUCTURESThe Standard Theory of Syntax contained a base component. In this base were two subcomponents, one that generated the context-free Phrase Structure Rules for the deep structures of the language and the other which globally inserted lexical items into the final string of the deep structure of the sentence. The problem with this model is that it also generated many PS rule combinations which do not occur in language. What was needed was a PS component that avoided this problem. This led to a re-examination of PS rules and several things were discovered. One of them is that the labels for these categories are based on the traditional notion of the parts of speech. Outside of this tradition, they lacked cogent reasons for their nomenclature and their existence. The revision of this information became known as Structure Dependent Constructions, and X-Bar Theory. Another discovery which came out of this re-examination is that the head of a phrase turns out to be an important linguistic concept. This led to a closer look at endocentric and exocentric constructions. The revisions in this case were known as the Head Parameter, Head and Maximal Projections, The Hierarchical Organization Phrases, and X-Bar Theory. In every phrase, there is always an obligatory element. This element is called the head of the category. The head of a Noun Phrase is the Noun. The head of a Verb Phrase is the Verb. The head of an Adjectival Phrase is the Adjective. The head of a Prepositional Phrase is the Preposition. From this awareness, they noted a similarity among all phrases, viz. X (N, V, A, P) is the head of XP (NP, VP, AP,PP).N is head of NPV is head of VPA is head of APP is head of PP_ Therefore, X is head of XP This concept become known as X-Bar Theory where X stands for the head of XP (phrase). The more interesting questions to emerge have to do with the head of a sentence and the head of a complement phrase. The rationale for the head of a sentential phrase is the inflectional marker, INFL. The inflectional marker includes much more than mere tense markers. It also had to do with agreement features, etc. Given the analogy of X is the head of XP, one must reinterpret INFL as the head of the sentence or IP. As for the head of the complement phrase, it is obviously, the complement. Hence, C is the head of CP.INFL is the head of IP, or I is the head of IPCOMP is the head of CP, or C is the head of CPThis focus on the head and its projections became known as the core grammar. This means that any other elements added to these phrases are adjuncts (add-ons).The destruction of the city in April of 1945. Head of NP Adjunct John saw Mary this afternoon.Head of IP Adjunct whether John will see Mary tomorrow afternoon. Head of CP AdjunctIt should be obvious from the aforementioned examples that what has been normally called temporal adverbs are now treated as adjuncts to phrases. They are no longer considered to be peripheral part of the sentence and does not belong to the core grammar.Some of the terminology and insights used in X-Bar Theory are related to similar research done in the area of morphology. Linguists have always studied how words are put together to form morphological units or compounds. These compounds function as though they were single words: blackmail, skyscraper, bathroom, bookshelf, football, in-crowd, etc. Two combinations of words within compounds have drawn the most attention and commentary in linguistic analysis. One of these has to do with constructions in which one of the elements is a headword (endocentric) and the other element acts a modifier of that headword. Such a combination is known as an endocentric construction. Bookshelf, for example, is an endocentric construction because books modify the headword shelf. Such compounds are easy to detect because the combined elements function to define a subclass of one of the nouns in the construction. A bookshelf is a shelf for books. Similarly, a nosebleed is a kind of bleeding from the nose. The other kind of construction which morphologists are found of discussing is constructions with contrast with endocentric compounds. These are called exocentric constructions and they are defined negatively because the combined elements do not share a headword and modifier relationship. Scapegoat, for example, is not a kind of a goat. It is a kind of a person. Hence, this is an exocentric construction. What is interesting to GB Theory about this investigation of morphological constructions is the fact that endocentric constructions have many interesting parallels with X-Bar phrases. The headword functions as the head of the compound noun just as the noun functions as the head of a noun phrase. The conclusion that one draws from this information is that all XP constructions are endocentric. This discovery has been stated in even strong terms: there are no exocentric constructions among XPs.It is not enough to note that a phrase consists of a head and a modifying element. It is important to also consider how these relationships are structured. When all of the categories within a phrase are strung out in a linear pattern, this is called a flat structure.In a flat structure, all of the elements are connected by means of sister adjunction. They all belong to the same category (NP). There is abundant evidence that phrases are structured hierarchically. This evidence comes from the study of pro-forms. This is when a category is replaced by another form (pro-noun, pro-verb, pro-sentence, etc.). What is actually taking place with pro-form substitution is that some hierarchical structure is replaced by another form, a pro-form. These substitutions operate on structures. They are structure-based. These structure occur in layers. They are layered structures.NP: I saw John talking to Mary. I saw him talking to her. PP: John is going to London next week. John is going there next week.VP: John saw Mary and Bill saw Mary. John saw Mary and so did Bill. AP: John saw the pretty child and Mary saw the pretty child.John saw the pretty child and Mary saw her too. XPs consist of a head element within a phrasal construction. The head is a lexical category and the phrasal component is called a projection. Since these projects may occur in layers, one speaks of the top most layer as the maximal projection of a phrase. NP is the maximal projection of N. VP is the maximal projection of V. There are intermediate projections, however, that must be accounted for in describing the structure of phrases. These are called X-Bar projections.HEADMAXIMAL PROJECTIONX-BAR PROJECTIONVVPV-BARNNPN-BARAAPA-BARPPPP-BAROne would naturally ask how many levels or layers of structure can occur from the maximal projection of a phrase to its head. In GB Theory, there are only two such projections allowed - the maximal projection (XP) and the intermediate projection (X-Bar). It should be noted that in addition to a head element and its projection, there is another element within a phrase. The element is known as the specifier (SPEC). A similar element can be found at the CP level and it is known as the complement C. Both specifiers and complements are not syntactic categories. In this regard they differ from X-Double Bar (XP), X-Bar, and X categories. These elements are empty categories. They function as locations for parts of the structure of a phrase that may be filled in by actual syntactic categories in the process of the application o