General Description Of The Study English Language Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Studies related to raising verbs have been paid attention to. Some linguists talk about raising verbs in order to prove the rationality of their theories; some compare raising verbs with passive verbs or control verbs respectively. To have a better understanding of this linguistic phenomenon need unit relevant aspects to make it clearer because it is helpful for learners. To begin with, it is necessary to have a general description of the study.

General description of the study

The aim of the thesis is to analyze the specific characteristics of raising verbs by means of applying Government and Binding theory (GB theory for short) and making two groups of comparisons: comparison between raising verbs and passive verbs; comparison between raising verbs and control verbs. Moreover, it also shows how learners acquire raising verbs and use raising verbs appropriately based on the former analysis.

First of all, it is necessary to know the development of GB theory. Chomsky is famous for his Transformational-Generative (TG for short) grammar and Chomskian theory is an explanatory scientific theory. Chomsky thinks that linguistics should not only be applied to describe language structures, but also explain the reason why language comes into being according to certain rules as well as ungrammatical structures can not exist. Explanatory grammar can inspire linguists to go further in order to find rules underlying in the deeper structures and to offer evidence about ungrammatical situations.

Universal Grammar (UG for short) with a history of more than 50 years is a main idea of Chomskian theory. Informally, UG is a system of all the principles that are common to all human languages; this means languages as different as English and Italian and Japanese. To quote Chomsky himself: Universal Grammar may be thought of as a system of principles, common to the species and available to each individual prior to experience (1981b, 7). Government and Binding theory, as the fourth phase of TG grammar, has achieved a lot (1981a). Government and Binding theory, also called Principles and Parameters theory (1986a), is primarily developed by Chomsky in 1980s based on the former theory and developed by Chomsky and his followers later. The thesis makes use of GB theory because the theory is rather mature and the linguistic phenomenon discussed in this thesis is close related to the framework of Government and Binding theory.

Significance of the study of raising verbs

Nowadays, more and more language researchers are interested in the study of raising structure. There is a typical example for raising (Davies and Dubinsky, 2004, p3), as follows:

Barnett seemed t to understand the formula.

It seemed that Barnett understood the formula.

A guiding intuition has been that such pairs of sentences are truth-conditionally equivalent (Davies and Dubinsky, 2004, p4). From a semantic view point, the two sentences have the same meaning. The verb seem is treated as a most typical raising verb because seem has no semantic meaning with the subject of the main clause. As for the former sentence, the subject is based-generated from the subject position of the infinitival clause so that it leaves a trace in the lower clause. For the later sentence, the object of the verb seem is a that-clause, which refers to a expletive-it insertion in order to meet the EPP principle. Therefore, the same raising verb has different syntactic explanation under different syntactic structures.

Here is another example (Misha Becker, 2005, p52):

a. Janine tends [t to eat sushi] (raising)

b. Janine likes [PRO to eat sushi] (control)

The two sentences above look similar, but the underlying syntactic structures are different. One is a raising structure; the other is a control structure. The first sentence involves an NP movement, the subject of infinitival clause is raised to the subject of the main clause and leaves a trace t; the second sentence has no NP movement and the subject of the infinitival clause is PRO.

There is an example about W-verbs (Perlmutter 1979, p107-119):

a. It began to rain. (raising)

b. John began to eat a sandwich. (control)

The verb begin has two different meanings because of two different subjects. One is animate; the other is inanimate.

The following is an example about differences between raising verbs and passive verbs.

a. The boy was bitten by that angry dog. (passive)

b. *The boy was seemed to be bitten by that angry dog.

c. The boy seemed to be bitten by that angry dog.

It's obvious that the second sentence is ungrammatical, the asterisk shows it ungrammatical. The first sentence is an example of passive structure; the third sentence is right because the boy is the subject of the infinitival clause, and he is hurt by the dog, so the infinitival clause must be in a passive voice. In addition, raising verbs have no passive forms, which are explained in the following chapters.

Although raising verbs is just made up of a small class, including verbs like seem, appear, and tend and so on, raising verb as a kind of verbs shares something with other verbs. For example, they take regular verbal inflection, they follow negation, they do not invert in questions; however, there are still some differences between them. Particularly, they do not select a subject argument, or any other thematic arguments. Due to raising structure close related to other language aspects' study, this strong correlation is quite beneficial for one another. Specifically speaking, to explain Case Theory, raising verb becomes a piece of useful evidence. Conversely, Case Theory can make the syntactic properties of raising verb clear: 1, raising verbs do not assign subject theta roles; 2, the raising verbs do not assign subject structural case; 3, there is no passive forms among raising verbs; 4, there is not any semantic relationship between raising verbs and the subject of the main clause; 5, expletives including it and there can be the subjects in raising structures.

In the meantime, there are two comparisons: comparison between raising verbs and control verbs, and comparison between raising verbs and passive verbs. Differences, as well as similarities between them are clearer, which contributes to identifying the syntactic properties of raising verbs. To great degree, it also makes the characteristics of the other two groups of verbs clear.

Some of the former studies on raising verbs treat the linguistic phenomenon as an available example to explain theories; some studies it through being paralleled with passive verbs. In contrast, the thesis is in an effort to make full use of the explained theories to analyze the properties of the group of special verbs. Moreover, by the two groups of comparisons, raising verbs become the dominating topic. Most importantly, the study of raising verbs not only helps English research but also triggers language acquisition.

1.3 Structure of the thesis

The thesis consists of five parts:

Part one is the introduction of the whole thesis; it focuses on the phenomenon of raising and the significance of the study of the raising verb and the structure of the thesis.

Part two is about the literature review of raising verb. Through learning the historical or current study of raising verbs, we can have a better understanding of raising verbs.

Part three introduces relevant theories, such as Theta-Theory, Case Theory, X-bar Theory and NP movement, and based on the theories, it classifies raising verbs into two groups; one is subject-to-subject raising verb; the other is subject-to-object raising verb.

Part four is made up of two comparisons: one is between raising verbs and control verbs; the other is between raising verbs and passive verbs. By comparing with the other two kinds of verbs, the properties of raising verbs are clearer. Besides, it deals with the inspiration: acquisition of English raising verbs shows how English learners acquire English raising verbs.

Part five is the conclusion of the thesis. In the part, it makes a summary about the thesis. And it shows its shortcomings as well.

Chapter Two Literature Review

Before having a view of former studies, the definition of raising verbs should be introduced.

2.1 The definition of raising verbs

Look at examples given below:

a. He seems to be hurt.

b. It seems that he is hurt.

2) a. Lily turned out to be the manager.

b. It turned out that Lily was the manager.

3) a. They appear to be upset.

b. It appears that they are upset.

4) a. We happened to meet our professor.

b. It happened that we met our professor.

There are two levels of syntactic representations: deep structure (D-structure) and surface structure (S-structure). The two levels are related to each other by movement: elements at D-structure are moved to somewhere else at S-structure. Leaving all of sentence bs alone for the moment, those examples are revised as follows:

a. e seems he to be hurt.

He seems t to be hurt.

2) a. e turned out Lily to be the manager.

Lily turned out t to be the manager.

3) a. e appear they to be upset.

They appear t to be upset.

4) a. e happened we to meet our professor.

We happened t to meet our professor.

He, Lily, they, and we are NPs each of which is the subject of the infinitival clause. All of them are moved from the D-structure to the subject position of the S-structure. At D-structure, the subject of the matrix clause is empty, and the subject of the infinitival clause can be assigned Agent θ-role in that infinitival clause does not possess the ability of assigning any Case. And at the same time, the empty subject position filled by e receives a θ-role form the inflection; so, the subject of infinitival clause has to be moved to the only empty position of the sentence with θ-role in order not to violate θ-criterion. At S-structure, the trace t stands for the base position where the subject of the infinitival clause is moved. IP, the infinitival clause, can not assign Case to its subject, and e receives nominative Case from the inflection. The movement also meets the requirement of Case Filter.

It is time to talk about sentence bs.

First of all, consider following sentences:

b1. *seems that he is hurt.

b2. * turned out that Lily was the manager.

b3. *appears that they are upset.

b4. *happened that we met our professor.

These sentences are ungrammatical because they violate the EPP. It-insertion is used here. Expletive it receives no θ-role; therefore, the verbs in sentences above are one-place predicates which take a clausal complement. Take It seems that he is hurt as an example, as for the clausal complement, he is the external argument of hurt. Moreover, the thematic relations between sentence a and sentence b are the same. In other words, under the two situation, the subject of seem which is empty receives no θ-role at D-structure. The D-structure is:

[IP e seems [cP that [IP he is hurt ]]].

As illustrated above, the main verbs of sentence as are followed by infinitival clauses, and there exists NP movement at D-structure. The subject of infinitival clause is moved to the position of the main clause. The subject of the lower clause is raised out of the clause and moved into a higher clause, this movement is sometimes referred to as NP-raising or raising. Verbs such as seem which induce raising are called raising verbs (Hageman, 2004, p309).

In recent years, the theory related to raising verb has been paid more and more attention by several linguists. Like other linguistic theories, the theory of raising verbs also has its history. Western linguists have studied the phenomenon.

2.2 Foreign studies of raising verbs

Chomsky is famous for his Transformational-Generative grammar. Generative grammar means to define rules that can generate infinite grammatical sentences. Transformational Grammar seeks to identify rules (transformations) that govern relations between parts of a sentence. In 1950s, Chomsky almost turned over the whole previous linguistic ideas. The TG grammar went through five periods: Syntactic Structure, Standard Theory, Extended Standard Theory, Government and Binding Theory and the Minimalist Programme. In Chomsky's Syntactic Structures (1957), isolating grammatical analysis from semantics is the feature of Syntactic Structure. The so-called Standard Theory corresponds to the original model of Generative Grammar laid out in Chomsky (1965). A core aspect of Standard Theory is a distinction between two different representations of a sentence, called Deep structure and Surface structure. The two representations are linked to each other by transformational grammar. The so-called Extended Standard Theory was formulated in the late 1960s to early 1970s. Syntactic constraints and generalized phrase structures (X-bar Theory) are its features. The so-called Revised Extended Standard Theory was formulated between 1973 and 1976. It contains restrictions upon X-bar Theory (Jackendoff, 1977), assumption of the complement position and move α. The GB theory was developed principally by Chomsky (1995) in the 1980s, which is a radical revision of his earlier theories, and was later revised in The Minimalist Program.

Besides Chomsky, Haegeman and Ouhalla account for GB Theory in details (1994, 2001). All of them refer to raising structures while they discuss X-bar Theory, Case Theory, NP movement as well as Theta-Theory. Raising structures can illustrate those theories. Generally speaking, they prefer raising structures to be paralleled with passive structures.

The earliest version of the true raising analysis belongs to Rosenbaum (1967) and further developed in Postal (1974). In later work, it is defended by Authier (1991), Johnson (1991), Koizumi (1995), Runner (1998, 2006). Postal has made a great contribution to raising verb. In 1974, he explains raising verbs in details, and he classifies raising verbs into three parts; they are A-verbs (the subject-to-subject raising verbs), B-verbs (the subject-to-object raising verbs) as well as W-verbs. Moreover, he points out raising does work among some intransitive verbs such as begin and turn out with gerundive complements and transitive verbs such as prevent and keep. While part of his conclusion is arguable, the rest is still available up to now (Postal, P., 1974). Besides, Rosenbaum (1970) also identifies the verb begin as a raising verb because it shares similarities with the subject-to-subject raising verb. However, from a different point view, Chomsky (1973) argues that A-verbs include a subject-to-subject raising operation. What's more, he thinks B-verbs have nothing to do with raising operation. Viewing from the discussion of the history raising and control in English, Higgins (1989) shows that raising was the result of two changes for the relevant predicates. In addition, Jacobson (1990, p423-475) makes use of the notion of function composition as an approach to raising verbs. However, the approach does not offer an actual stage at which the subject of the raising predicate is actually in the specifier position of the infinitive. According to Barron (1997), raising predicates have arisen historically from predicates with full argument structure, and there is a semantic bleaching that may correspond to the historical dissociation of the function and the theta-role. This will in turn affect the syntactic representation of these predicates. She (1997) holds that a perception verb may become a raising verb if three conditions apply: i) the presence of secondary predication; ii) suppression of the perceiver argument through detransitivisation; iii) cognitive shift from a physical to a mental process. Haegeman (1994), Haegeman & Gueron (1999) and Barron (1997) explain the English raising verbs. Haegeman also argues that the word seem cannot assign structural case. Be and perfect have are also raising verbs in English (Haegeman & Gueron, 1999).

As for discussion of whether or not modals belong to raising verbs, Thomas Ernst (1994) gives two pieces of evidence to show that modals are quite different from control or raising verbs. First, while modals allow VP-preposing, control and raising verbs do not. Second, he argues that there is an empty inflection in Chinese. However, his study doesn't say anything about the differences between modals and raising predicates. His classification is not quite clear.

Although raising structures and control structures are quite similar in form, they are quite different in several ways, especially their influence on language acquisition. As noticed by Huang (1991), Radford (1997), Chomsky (1995), Haegeman (1994), Zhang (2001), raising verbs allow no referential subject like expletive there, but control verbs do not. The subject of control verbs must be a referential phrase. A control verb must assign a theta-role to its subject. Control verbs have an accusative case feature, while raising verbs do not. One of the primary differences between raising and control verbs is that control verbs stand in a thematic or selectional relationship with the matrix subject, but raising verbs do not. Raising is somewhat limited by lexical idiosyncrasies of a given language; in English it is possible with predicates like seem, appear, happen, be likely, be apt, turn out, and begin but not similar predicates like be possible (Davies and Dubinsky, 2004). Similar lexical items can be used as raising and control predicates across languages; for instance, modal and aspectual verbs are often raising predicates, and verbs of intention and desire are typical control predicates (Stiebels 2007, 2010). The most well established approach to control is developed within the framework of Government and Binding Theory (Chomsky, 1982). This paper employs empirical methods to examine verbs such as seem, for which the traditional raising to subject analysis relates pairs of sentences which differ by taking an infinitival or sentential complement (Scott Grim, 2009).

The acquisition of the distinction between raising verbs and control verbs has been widely studied, too. The paper addresses the problem of learning the class of raising verbs. These verbs are potentially problematic for learners in that unlike typical main verbs, these verbs do not stand in a semantic relation with any noun phrase arguments (Misha Becker, 2005). This paper addresses the question of how language learners come to distinguish the class of 'raising' predicates from other kinds of predicates. What distinguishes these predicates from other main verbs is that while they share the morphosyntactic properties of main verbs, raising verbs are auxiliary-like in their argument structure. In particular, they do not select a subject argument, or any other thematic arguments. There is a significant effect of subject animacy or expletive subjects on the choice of a raising vs. a control verb (Misha Becker, 2005). The learner may be biased to hypothesize one structure rather than the other. As Borer and Wexler (1987) argue, A-movement may take time to mature, and so the learner may be constrained to first hypothesize only the control structure. Frank's (1998) claims that a raising structure is computationally more complex than a control structure, and a learner should, therefore, be inclined to suppose a control structure first. Misha Becker (2005) argues that expletive-driven strategy is insufficient on both logical and empirical grounds: a class of verbs that are ambiguous between being raising or control provide a problem for the logical soundness of the expletive-driven learning approach. As noted by Perlmutter (1979), there are verbs that are ambiguous between being raising and control verbs. Verbs like begin, start, fail and continue can be raising verbs, but they can also function as control verbs. Misha Becker (2005) proposes two families of cues that could drive the learning of these verbs. Recent work by Yang (2002) supports this trend of moving away from a single trigger type of approach, and towards a strategy that makes use of multiple sources of evidence in a probabilistic way. The multi-cue type of approach admonishes that a learner will have to consider multiple sentence frames in which a verb appears in order to figure out the lexical meaning of the verb (Gleitman, 1990).

The phenomenon of raising verbs also attracts Chinese linguists. Based on the study of western linguists, they start to find raising verbs among Chinese syntactic structures.

2.3 Domestic studies of raising verbs

Wu (1998) discusses the potential principles of subject-to-subject raising from the viewpoint of semantics. Li (1985) assumes that all verbs in Chinese assign case from left to right. In the case of raising structures, Li argues for a one-to-one relation in structural case assignment and claims that clauses require case. Li (1990, P123-128) argues that verbs like keneng, nan, rongyi, and kaishi are raising verbs. Huang (1988, p35) has made a very thorough study on shi (be) and you (have) in Chinese. He has made a distinction between raising verbs and control verbs. Tsao (1996, P172) uses a functionalist approach and holds that the raised parts are topics. Based on his analysis, Tsao (1996, p176-180) classifies Chinese raising verbs into four types: i) aspectual verbs like kaishi, tingzhi, meiyou; ii) modal verbs such as yinggai, keneng, kanqilai, and shi; iii) tough verbs like rongyi, nan, zhide; iv) frequency verbs like changchang, nande. However, Tsao doesn't explain the syntactic motivation of raising. Shi (2000) points out the weak point of Tsao's theory of topics. Lin and Tang (1991) have made a through research of Chinese modals. They (1995, p46) believe that Chinese modals are not the constituents of INFL, but are main predicates.

As for discussion of modals, Chinese linguists also put forward their opinions. Lin and Tang (1991) argue that modals take clausal complement, CPs in particular. They have done a very extensive research on Chinese modals under Government and Binding theory. They divide modals into two types: one being raising modals and the other being control modals.

As mentioned above, linguists have discussed raising verbs from several aspects. Some explain raising structures by comparing with passive structures or control structures. Others further classify them into different groups. Still others explore how children acquire raising verbs. Both western and Chinese linguists work on raising verbs. It is not difficult to find that Chinese linguists still need to work hard so as to keep up with others. Moreover, only paralleling with passive structures is not sufficient for studying the syntactic features of raising verbs. Through comparing with passives and control verbs, features of raising verbs are more specific and clear. To sum up, having a better understanding of English raising verbs is beneficial for language acquisition.

Chapter three Characteristics and Classification of Raising Verbs

The verb seem is a most typical raising verbs. Now, take it as an example, it can be followed by DP, PP, AP, IP and CP.

a. That strong man seems an athlete.

b. The poor guy seems out of money.

c. Tom seems unhappy.

d. Lucy seems to go swimming.

e. It seems that Lucy goes swimming.

From a viewpoint of grammatical function, that strong man, the poor guy, Tom, Lucy and it are called subjects. However, they are called external arguments according to Theta theory. In d and e, it is obvious that the two sentences express the same meaning, but the meaning is described by two different sentence structures. Analyzing their deep structures can make the difference clear. For d, Lucy, the subject of the infinitival clause, is moved to the subject position of the matrix clause and leaves a trace in the based-generated position; and for e, Lucy, the subject of the subordinate clause, is moved to the object position of the matrix clause and also leaves a trace. A verb is called a raising verb, if it is raised from the subject or object of infinitival and subordinate clauses and leaving a trace in the based-generated position.

It is not difficult to see that raising verbs, unlike other verbs, have theirs own characteristics. The question is how raising verbs come into being and what factors lead to their emergence. This question is solved later in the chapter. First of all, relevant theories need to be introduced because they illustrate the factors resulting in raising.

3.1 Theoretical framework

The thesis talks about raising verbs under the theoretical framework of GB theory. GB theory, as the most mature theory of TG grammar up till the present moment, has achieved quite sub-theories. There is no doubt that these sub-theories are helpful to explicate the linguistic phenomenon. First of all, sub-theories are introduced one by one, and then, raising verbs are classified based on the combination of sub-theories.

3.1.1 X-bar theory

Before talking about X-bar theory, it is necessary to clarify two concepts: one is the Projection Principle; the other is the immediate constituent analysis.

The definition of the Projection Principle is that the properties of lexical items project onto the syntax of the sentence, lexical structure must be represented categorically at every syntactic level (Chomsky, 1986a, p84).

The Projection Principle shows the union of syntax and the lexicon. There is no denying that syntax is based on lexicon, and the characteristics of lexical items project onto the syntax. For example, the properties of the verb love [love, - NP] decide that it must be followed by a noun phrase in the sentence. If there is a sentence like "I love." The sentence is ungrammatical. If the projection is extended to the whole sentence, it is called the Extended Projection Principle. According to Chomsky (1982a, p10), each sentence must have a subject. For instance, it seems that today is a sunny day. Syntactically, the subject it has no semantic meaning with the predicate seem, it is an expletive. According to the Extended Projection Principle, it can not be omitted, otherwise, it is ungrammatical.

Immediate constituent analysis or IC analysis is a method of sentence analysis, which was first introduced by Bloomfield (1933) and developed by Wells (1947). In Chomsky's early works, it is a helpful strategy for analyzing sentence structure. The method produces most tree structures which can make sentences easier to be analyzed. IC analysis divides a sentence into immediate constituents, and these immediate constituents are divided into further immediate constituents. The process is not ended until each constituent consists of only a word or a meaningful part of a word. The result of the process is usually presented by diagrams which reveal the hierarchical relationships between constituents. These diagrams are often trees. For example,



Det N′ I VP


The boy V′ NP


love A′ N

A coats


From the example, it is clear that IP immediately dominate the constituents: NP and I single bar. And the two constituents are divided further. At last, each of them is ended by a single word or a meaningful part of word. The tree diagram demonstrates relationships between lexical items or phrasal structures. Their hierarchical or precedent relations are visible.

Through detailed explanation of the Projection Principle and the immediate constituent analysis, X-bar theory can be easier to understand.

The part of the grammar regulating the structure of phrases is called X-bar theory, which brings about what is common in the structure of phrases: all phrases are headed by a lexical item. The four lexical phrases used in X-bar syntax are: Verb Phrase (VP), Noun Phrase (NP), Adjective Phrase (AP) and Prepositional Phrase (PP). Each of the four phrases has a head. They are V, N, A, P. in return, these heads reflects their phrasal properties. Specifically, X can be replaced by V, N, A, P. What's more, lexical categories are related to phrasal categories, they belong to the same category because phrases are endocentric. In other words, if the head is a noun, the phrase must be a noun phrase. The other heads are used as well.

X-bar theory was first proposed by Noam Chomsky (1970) and further developed by Ray Jackendoff (1977). An X-bar theoretic understanding of sentence structure is possible in a constituency-based grammar or phrase structure grammar only. X-bar theory claims that all types of phrase need two internal levels of structure: one (X″, X double bar) consists of the head and possible specifiers; the other (X′, X single bar) consists of the head (X, X zero bar) and possible complements, and a single-bar category can also contain a further single-bar category and an adjunct (Vivian Cook, & Mark Newson, 2000, p144-147). It is presented by binary tree diagram as follows:


spec X'

X' adjunct

X complement


Take an NP as an example:


Det N'



P Det N'


the boy with an umbrella

Paralleling presented, it can be expressed as follows:

NP→ specifier (the) N'

N'→ N' adjunct (PP: with an umbrella)

N'→ N (no complement)

The X-bar framework was then extended to the functional categories I (inflection) and C (complementizer). The VP is taken to be the smallest representation of the clause, in that at deep structure (D-structure) it contains all the lexical elements which constitute a clause. The IP (Inflection Phrase) is built on top of this VP and contains the inflectional head (e.g. tense and agreement). CP (complementizer phrase) in turn is built on top of IP and contains a head complementizer in embedded clauses and an empty head position in matrix clauses. Both IP and CP conform to the usual X-bar pattern (Vivian Cook, & Mark Newson, 2000, p157).

X-bar theory offers such a general formula that complex language becomes easier. As mentioned above, the Projection Principle, the immediate constituent analysis and X-bar theory connect lexicon with syntax, which means grammar not only studies syntactic phenomenon from a separate view. The way just concerning spark words is meaningless. Words should be studied in the sentence.

3.1.2 Theta theory

Theta theory, as another sub-theory of GB theory, is also important for analyzing raising verbs. The reason is that Theta theory is involved in dealing with the semantic relationship between predicates and arguments. Besides, the theory also puts forward linguistic rules to restrict language points. Lexical entries talked about here are not about their pronunciation and meaning, but about syntactical connection with other words and phrases.

First of all, what predicate and argument are is a question demanding to be answered. Predicate refers to a lexical item that says something about an entity or the relationships between entities, and arguments are connected with one another by a predicate (Vivian Cook, & Mark Newson, 2000, p164). Therefore, a predicate expresses a semantic relationship between related arguments. For example, John plays football. For this sentence, the word play describes a semantic relation between two entities John and football. So, the word play is called a predicate, and the two entities are called arguments. Secondly, the semantic relationship needs to be talked about specifically because it is not arbitrary. Taking advantage of the example illustrated above, it is not difficult to find that the verb play requires one argument involving the entity doing the playing (John) and the other argument involving what is played (football). The doer is called Agent, and the accepter is called Patient. The relationship can be represented by θ-grid: play <Agent, Patient>. These semantic roles of arguments related to a predicate are called thematic roles or θ-role, such as Agent, Patient, Goal and Theme (Vivian Cook, & Mark Newson, 2000, p164, 172). Only certain arguments bear certain θ-roles. For example, TV watches Lily. TV can not bear the Agent role. A predicate, therefore, semantically selects a certain number of arguments bearing certain θ-roles. Besides s-selection properties, it is often assumed that lexical items also have c-selection (category selection) properties. C-selection is represented in the lexicon in terms of a subcategorization frame (Vivian Cook, & Mark Newson, 2000, p162). For example,

a. Mary found a book. find [ __ NP]

b. Mary found there was a book. find [ __ CP]

Combing θ-grid with subcategorization frame, a more detailed lexical entry for the verb find in the sentence a is:

find <Agent, Patient> [ __ NP]

There is a major difference between c-selection and s-selection. C-selection deals only with complements, while s-selection deals with all arguments including complements and the subject (Vivian Cook, & Mark Newson, 2000, p163).

The assignment of θ-roles to arguments is not at random. There is a one-to-one relationship between arguments and θ-roles, which is called θ-Criterion (Chomsky, 1981a, p36):

Each argument bears one and only one θ-role, and each θ-role is assigned to one and only one argument.

Θ-theory is concerned with the assignment of thematic roles such as agent-of-action, etc (Chomsky, 1981a, p5-6). Θ-roles are transferred from a predicate to its arguments by a process known as θ-marking (Vivian Cook, & Mark Newson, 2000, p169). Complements, as internal arguments, are θ-marked by the head under a sisterhood condition; subjects, as external arguments, are also θ-marked under the sisterhood condition, but it is the head and its complement that θ-marked the subject (Vivian Cook, & Mark Newson, 2000, p171-172).

The last question is concerned with A-positions (argument positions), non-A-positions (A-bar positions) and θ-positions. A-positions are the usual location for the arguments of a predicate. A-bar positions refer to positions that can not take arguments. A θ-position is one to which a θ-role is assigned (Vivian Cook, & Mark Newson, 2000, p178, 179). The following example illustrates locations to which the positions are respectively distributed.





Anne I VP



t V′ PP


play tennis P NP

at school

All the NPs including the trace t of the moved subject from VP occupy A-positions. Anne is an NP in subject position. Its root position is located in the specifier of VP to which the θ-role Agent can be assigned; tennis is an NP in object position. It is located in the complement of VP to which the θ-role Patient can be assigned; home is also located in the complement position, but it is in the object position of PP. For the example, two positions are A-bar positions: one is the specifier of CP (C, empty in the example); the other is the complement of I (VP). Because it is impossible for any lexical entry can project arguments onto these positions. Θ-positions are restricted by the sisterhood condition placed on the process of θ-marking (Vivian Cook, & Mark Newson, 2000, p179). Therefore, the trace t is sister of V′, TV is sister of V, and NP is sister of P. All of them are in θ-positions.

As mentioned above, it is obvious that θ-positions are A-positions, but not all A-positions are θ-positions, such as expletives it and there. For example, it seems that he wants to go home. In this sentence, it has no semantic relationship with the verb seem, so it can not receive θ-roles. It can not be omitted because of the EPP.

The θ-theory is concerned with the semantic relationship between predicates and their arguments. The assignment of θ-roles happens at deep structures. Once θ-positions are moved, they move with their θ-roles at the same time, which meets the demand of the Projection Principle.

3.1.3 Case theory

Case theory is a most important part sub-theory of GB theory as well. Case is a traditional topic. Case theory, however, deals not just with traditional morphological case, but with abstract Case. For example,

He fell from the wall yesterday.

She gave me an apple.

He lent his books to them.

They love their parents.

According to traditional grammar, he, she, they are the subjects of the sentences in the nominative case; me is the object of the verb gave in the accusative case; them is the object of preposition to in the Accusative case; and his, there are in a possessive relationship with books and parents respectively in the genitive case. These examples show that traditional case forms are visible in the surface structure while abstract Case handles case even when it does not appear in the surface. For example, a single NP John can not be judged from the surface whether it has a nominative or accusative case, but it does not mean that it does not have abstract Case. In some languages, like Chinese, case is not morphologically realized; in other languages, like French, case can be clearly shown by morphological change. In view of these linguistic facts, Chomsky (1986a, p74) assumes that it is assigned in a uniform way whether morphological realized or not.

Case theory deals with assignment of abstract Case and its morphological realization (Chomsky, 1981a, p6)

Seeing the examples given above, the structural position where NPs are located determined their Case: subjects usually have Nominative Case and objects have Accusative Case. Similarly to Theta theory, Case is also assigned by certain elements. It is known that verbs and prepositions are followed by objects, so verbs and prepositions should assign Accusative Case; and subjects of finite clauses should be assigned Nominative Case. There is a difference between subjects of finite clauses and those of infinitival clauses. For instance,

a. She has an apple a day.

b. Jack seems to be a superstar.

c. e seems Jack to be a superstar.

According to the Projection Principle, the finite clause is an IP and the infinitival clause "to do" is also an IP. The head I of the finite clause is inflection with [+ tense, + agreement], so it has ability to assign Nominative Case to its subject; however, the head I of infinitival clause to with [- tense, -agreement], so it has no ability to assign Nominative Case to its subject. The sentence a is a finite clause, the subject she has an agreement with its inflection [-s]; sentence c is the deep structure of sentence b. Jack, the subject of infinitival clause, can not be assigned Nominative Case at deep structure. It is moved to the subject position of finite clause which has Nominative Case assigned by the inflection [-s]. The movement is the result of joint action formed by Theta theory and Case theory. The reason why Jack must be moved to the empty position is explained in the following.

Cases are classified into structural Case and inherent Case. Nominative Case and Accusative Case are structural Case, while Genitive Case is inherent Case. The principle that forces case to be assigned is called Case Filter (Chomsky, 1986a, p74): Every phonetically realized NP must be assigned (abstract) Case. In other words, all overt NPs have Case. Case is a property of NPs in general. In English, Case is overtly reflected in phonetically NPs only. Therefore, the movement about the example discussed above is forced by Case Filter from a caseless position to a case-marked position.

Chomsky (1981a) proposes that Case is assigned under government. It accounts for the fact that certain cases assigned to certain positions. But there are still exceptions. For example:



Lily I VP

(-s) V′


believe NP I′

him I VP

to V′


be a good teacher

For the head of infinitival clause does not have ability to assign Case to its specifier of IP, and infinitival IP is not a barrier for outside government, the verb believe needs to be assigned Accusative Case to an NP position. Therefore, the verb believe can govern into the IP and assign Case to its subject. The Case is often referred to as exceptional Case-marking, abbreviated as ECM.

There is a connection between Case theory and Theta theory, Case assigned to NPs makes them visible so that they can be θ-marked. They can deal with some linguistic phenomena together, like NP movement. Case theory is concerned with overt NPs, which is useful in explicating NP movement.

3.1.4 NP movement

In the previous examples, some of them are involved in movement from deep structure to surface structure. It is not difficult to see that the relationship between deep structure and surface structure is movement. Deep structure tries to show the original location of elements in the sentence moved in surface structure. The principle of movement is Move α where α stands for any category. It proposes that any part of the sentence could move anywhere. The movement operation (henceforth Move α) is an invariant principle of computation, stating that a category can be moved to a target position (Chomsky and Lasnik, 1993, p522). There is another way expressing the relationship between deep structure and surface structure: a chain. A chain is the surface structure reflection of a history of movement, consisting of the positions through which an element has moved from the A-position it occupied at deep structure (Chomsky, 1986a, p 95). For example,

Teachers' Day is when?

When is Teachers' Day?

c. When1 is2 Teachers' Day t2 t1?

Sentence a is the deep structure, sentence b is the surface structure, and sentence c is the surface structure preserving their original locations through traces t2 t1. The example has two chains to link their relationship: one is (when, t1), the other is (is, t2). Movement of an element α always leaves a trace and, in the simplest case, forms a chain (α, t) where α, the head of the chain, is moved element and t is its trace (Chomsky and Lasnik, 1993, p522).

In the principle of movement, α stands for any category. Therefore, the movement is called NP movement when α stands for NPs. In English, the passive construction is a typical example of NP movement. For instance,

The angry dog bit the naughty boy. (active sentence)

The naughty boy was bitten by the angry dog. (passive sentence)

The naughty boy was bitten t by the angry dog. (surface structure)

e was bitten the naughty boy by the angry dog. (deep structure)

From the point of view of the traditional grammar, passive construction is that the object of the active sentence is moved to the subject location of the passive sentence by exchanging the angry dog and the naughty boy and adding was, -ten, and by. Within the GB theory, however, it explains passive construction through deep analysis. The passive sentence is considered as having a deep structure. In previous discussion, it is known that the verb bite has following lexical entry according to the Projection Principle: bite V, [ __ NP] <Agent, Patient>. It c-selects an NP and s-selects two θ-roles; the angry dog is assigned Agent role, the naughty boy Patient role. The verb bite has to be followed by an NP; otherwise, it violates the Projection Principle. Sentence b seems wrong because it is not followed by an NP. To meet the requirement of the principle, t, the trace, appears after the passive verb. So, at the surface structure, sentence b is written like: The naughty boy was bitten t by the angry dog (sentence c). As for its deep structure, sentence b is: was bitten the naughty boy by the angry dog. The subject seems to be missing. According to the EPP, it violates the principle. Under this situation, the deep structure of sentence b is written like: e was bitten the naughty boy by the angry dog (sentence d).

Movement is also limited by θ-theory. The angry dog and the naughty boy are subject and object respectively, so they are A-positions and receive θ-roles. Θ-roles are assigned at deep structure; movement does not influence its original θ-roles. As for the object the naughty boy in the sentence d, it has already been assigned a θ-role by the verb; the original θ-role is still with it when NP movement happens. However, once it is moved to the subject position, it receives another θ-role. One argument has two θ-roles, which goes against θ-criterion. With the GB theory, in English, the passive morphology is said to not only trigger movement, but also absorb the subject's θ-role leaving the subject position free to be moved into by the object (Vivian Cook, & Mark Newson, 2000, p194, 196).

Beside passive construction, raising construction is also a typical example of NP movement. For example,

a. Dan seems to be hard-working.

b. e seems Dan to be hard-working. (deep structure)

c. Dan seems t to be hard-working. (surface structure)

Like the passive, raising verb seem also triggers NP movement and absorbs agent θ-role; but at the same time, there exists differences between them, which is discussed in details in chapter four. No matter passive verbs or raising verbs which cause the movement, it is not difficult to find that movement is never determined by a specific rule, but rather results from the interaction of other factors (Chomsky, 1986b, p5).

To sum up, one linguistic phenomenon appears, the solution to explain it is not one and only; under most circumstances, it is often involved in several factors which should be explicated by relevant theories. In return, the characteristics of this linguistic phenomenon are more detailed and clearer with the help of those theories. The study of raising verbs is not exceptional. The characteristics of raising verbs are raised by making good use of the theories mentioned above.

3.2 The classification of raising verbs

In the former part of the chapter, relevant sub-theories of GB theory are introduced. Based on them, the second part is going to discuss the characteristics of raising verbs and classify them into two groups.

The definition of raising verbs is introduced in chapter two. It is certain that raising verbs are not identical to one another. They should be divided into groups depending on their specific characteristics. Postal (1974) has made a great contribution to raising verbs. In 1974, he explains raising verbs in details, and he classifies raising verbs into three parts; they are A-verbs (the subject-to-subject raising verbs), B-verbs (the subject-to-object raising verbs) as well as W-verbs. Generally speaking, there are two main groups of raising verbs, one is subject-to-subject raising verbs; the other is subject-to-object raising verbs.

3.2.1 Subject-to-subject raising verbs

Subject-to-subject raising verb is also called subject-raising verb or raising to subject verb. Subject-to-subject raising verb is just what the name implies: the subject of infinitival clause is raised up to the subject position of the matrix clause. Verb triggers the movement is called subject-to-subject raising verb. The verb seem is a typical subject-to-subject raising verb.

Jack seemed to live in a warm place.

The queen seems to be in charge of the whole country.

The poor seemed to revolt against exploitation.

Peter seems to work hard.

In many cases, different sentence patterns expressing the same meaning can help to solve some problems. Therefore, it makes sense to transform these sentences into the sentence pattern of the verb seem followed by a subordinate clause.

a1. It seemed that Jack lived in a warm place.

b1. It seems that the queen is in charge of the whole country.

c1. It seemed that the poor revolted against exploitation.

d1. It seems that Peter works hard.

These changed sentence patterns share the same meaning with the previous sentences. By making good use of them, it is easier to find the syntactic properties of seem. As for that-clause, it is a finite clause, each NP has been assigned one θ-role, and all overt NPs have Case. For example, Jack receives a θ-role from the predicate live, and place is assigned a θ-role by the preposition in. At the same time, Jack is assigned a Nominative Case by the inflection -ed; and place an Accusative Case by the preposition in. So, they do not need any θ-role or any Case from outside that-clause. As for the main clause, it is known to us that the expletive it receives no θ-role and it is meaningless. Seem, therefore, fails to assign an external θ-role, which is one of prominent characteristics of raising verbs.

Next, it is time to talk about the other important features of raising verbs. As for previous sentences a, b, c and d, all the subjects of the matrix clauses are empty at D-structure. In answer to Case Filter and θ-Criterion, the subjects of infinitival clauses have to be raised from the lower clause to the higher clause. Take sentence a as an example, the D-structure is:

[IP e seems [IP Jack to live in a warm place]].

Jack is the subject of live, and the subject position of seem, which receives no θ-role, is empty. The infinitival clause, IP, can not assign Case to Jack, and the inflection -s of the matrix clause assigned a Nominative Case to the empty subject. If seem also assigns an Accusative Case to Jack, it violates θ-Criterion. Although the subject position of the main clause is empty, it has Case. There must be a θ-role with it. So does Jack. Jack has a θ-role. There must be a Case with it. NP movement offers a solution to this situation, which meets the demand of θ-Criterion as well as Case Filter. Seeing this situation from a different angle, seem can not assign Accusative Case to the subject of infinitival clause, which is the other characteristic of raising verbs.

Appear, tend to, happen and used to are all subject-to-subject raising verbs because they share the same syntactic characteristics with seem. It is explained as follows:

Coco appears to leave.

Heavy drinkers tend to cause traffic accidents.

We happen to meet our old friends at his birthday party.

The teacher used to wear blue blouse.

Their D-structures are parallel to the D-structure of seem. All of them are triggers of the raising process. Then, it is necessary to have a better understanding of subject-to-subject raising verbs. Like seem, the D-structure of appear is:

[IP e appears [IP Coco to leave]].

The other raising words are similar to them.

There are two characteristics for raising verbs: one is that raising verbs can not assign external θ-role; the other is that raising verbs can not assign Accusative Case. Burzio (1986) relates these two properties by the descriptive generalization.

Burzio's generalization:

(i) A verb which lacks an external argument fails to assign ACCUSATIVE case. (Burzio, 1986, p178-179)

(ii) A verb which fails to assign ACCUSATIVE case fails to theta-mark an external argument. (Burzio, 1986, p184)

It is obvious that subject-to-subject raising verbs take an infinitival clause. There are two different subjects: one is empty, the other which needs to be raised locates in the lower infinitival clause at D-structure. As for raising verbs, the other group is subject-to-object raising verbs.

3.2.2 Subject-to-object raising verbs

Besides subject-to-subject raising verbs, English has subject-to-object raising verbs, also called object raising verbs or raising to object verbs. As the name implies, verbs triggers the process that the subject of infinitival clause is raised up to the object position of the matrix clause are called subject-to-object raising verbs, such as expect and believe. As for subject-to-subject raising verbs, there are two characteristics for raising verbs. It is necessary to find out characteristics among subject-to-object raising verbs.

Consider the following pair of sentences:

The whole family expects that Johnson will win the basketball game.

The whole family expects Johnson to win the basketball game.

Apparently, the two different sentence patterns express the same meaning. For the first sentence, expect is followed by a that-clause. Similarly to seem clausal pattern, every NP in the clause has its θ-role as well as Case. They also do not need any θ-role or Case from outside the clause. Expect assigns two θ-roles: one is to the subject of the main clause; the other is to that-clause.

It is clear that Johnson is the subject of the embedded clause in sentence a; but it can also appear as the object of the matrix verb. In both cases, however, its thematic interpretation comes from the predicate of the embedded, not matrix clause. Its D-structure can show its semantic meaning with the predicate of the embedded clause. The D-structures are:

[IP The whole family expects [CP that [IP Johnson will win the basketball game ]]].

[IP The whole family expects e [IP Johnson to win the basketball game]].

Johnson has to move out of the infinitival clause IP because it can not assign Case to its subject. Moreover, expect need assign Case to its empty object. So, Johnson moves out from the subject position of infinitival clause to the object position of the main clause to receive Case.

In the two cases as mentioned so far, the process raising the subject of the lower infinitival clause to the subject or object position of the matrix clause is treated as raising. Verbs triggering the process are called raising verbs.

Believe, want, like, prefer, and hate belong to subject-to-object raising verbs. They occur with the nominal plus the infinitival clause.

I believe Lily to be a good student.

People want the whole nation to stand up together.

They would like Kitty to sing this song.

She prefers Coco to stay at home.

We hate John to live with us.

The immediate nominal cannot be passivized like this:

The whole nation is wanted to stand up together.

Kitty would be liked to sing this song.

Coco is preferred to stay at home.

John is hated to live with us.

As for believe, it can be passivized as follows:

Lily is believed to be a good student.

So, there are subsets for subject-to-object raising verbs.

Subject-to-subject raising verbs and subject-to-object raising verbs represent two main groups of raising verbs. It is certain that raising verbs still include other verbs which are not absolutely raising verbs. There are some ambiguous raising verbs. The following discussion further displays the specific characteristics of raising verbs. Through comparing with passive verbs as well as control verbs, the specific characteristics of raising verbs can be clearly illustrated.